Goler’s fleet is comprised of the following vessels:

· FLNG Gimi: operating in Mauritania/Senegal (BP GTA project), consistently overproducing (Q2 2026, 15% overproduction);

· FLNG Hilli: currently in Singapore after 8-year Cameroon contract, to be deployed in Argentina (SESA) for a 20-year contract;

· FLNG Esperanza: under construction at CIMC Raffles (China), 75% complete, to be deployed in Argentina (SESA) in H1 2028;

· 4^th FLNG (order disclosed in Q2 2026): EPC agreement signed with CIMC Raffles, with 3.5 mtpa of capacity, approximately $2.45 billion in total budged and targeted delivery in 2029. Expected to add $400mm in adj. EBITDA;

Golar also co-owns 10% of Southern Energy (SESA) (alongside Pan American Energy 30%, YPF 25%, Pampa Energía 20%, Harbour Energy 15%), which is the contracting counterparty for the Argentina LNG exports. Beyond the FLNG charter fees, this equity stake gives Golar direct commodity exposure to Argentine LNG economics.

The company´s business model offers highly predictable cashflows as its tolling agreements have long durations, ranging between 10 and 20 years and significant switching costs for charterers, mainly because disconnecting, towing and redeploying a vessel could take between 18 and 36 months, and is extremely capital intensive. This structural stickiness is also reinforced by the fleet’s mobility, unlike fixed onshore liquefaction trains, Golar’s units can be redeployed across basins, giving gas-field owners a credible alternative for diversifying supply concentration. This is gaining relevance given recent geopolitical tensions in chokepoints such as the Strait of Hormuz and attacks on Qatar’s Ras Laffan facility, the largest LNG production complex in the world.

Apart from being benefited by macro tailwinds, the Hilli and Esperanza tolling agreements carry commodity upside through a tariff component embedded in each charter, entitling Golar to 25% of the FOB gas price above $8/MMBtu, upside not priced into today’s valuation (JKM currently at $22.94/MMBtu, TTF at €66.73/MWh, equivalent to $22.83/MMBtu).

The combination of highly predictable, long duration cashflows, along with fleet expansion assuming the company is able to add a 5^th vessel before 2029, excellent capital allocation towards buybacks and commodity tailwinds provides the company a very asymmetric risk/reward profile. The main reason for this asymmetric profile is that in the case of a extreme early termination scenario, we see a 15% downside from current levels assuming only 50% of cashflows from each vessel are received, and after termination vessels are sold for 20% of their initial FV at deployment plus a 25% liquidity discount. Decommissioning costs are minimal, as Golar is only responsible for mooring disconnection and towage in the event of early termination of its charter agreements. For example, the Hilli asset retirement obligation (ARO) was only $6.4mm as of the latest Q2 report.

4 vessels run-off sensitivity:

5 Vessel run-off sensitivity:

Valuation

The valuation method is based on adjusted EBITDA, as GAAP EBITDA significantly understates the company’s cash generation. When Gimi reached commercial operations date (COD) (June 12, 2025) under the BP GTA lease, the FLNG asset under development was derecognized and replaced by a sales-type lease receivable. The contracts are structured such that the counterparty covers operating and capital expenditures, as well as withholding taxes. Therefore, to get the discounted cash flows, we only need to deduct the associated asset financing costs.

The valuation therefore consists of the core discounted cash flows of the company, adjusted for each vessel scenario, the stake in SESA, and the potential commodity upside.

For simplicity over the SESA stake valuation, we assumed an 8x multiple on Goler’s share of share of SESA EBITDA, getting to a valuation of $1.7 billon. We do not have a high conviction view on the spot price. For that reason, the valuation is performed excluding the potential upside from commodity price increases, which are instead treated as free optionality as opposed to a core pillar of the valuation.

Assuming that COD for Hilli in SESA is in Q4 2027 and COD for Esperanza is H1 2028 and discounting the option value 9% we get $1.759mm (35% of current market cap) in commodity upside not fully priced at the current valuation. We are aware of the limitations of assuming constant volatility, however the added complexity of accurately estimating the IV will only moderately change the output.

Catalysts

Among the catalysts that could drive the price to its FV, we note the following:

· Conclusion of the strategic review process: Golar hired Godman Sachs as its advisor in 2025 to explore the potential sale of non-core assets, and we think a scenario is possible where we even see a take out of the entire company.

· Buybacks: Golar approved a $150mm buyback in Q4 2025, of which $41mm was used to repurchase 1.1mm shares at $37.76 per share. Management also flagged a potential upsizing of the Hilli sale-leaseback, which could unlock liquidity without additional equity capital. Given current leverage of 1.7x EBITDA, we believe Hilli could support 4-5x EBITDA leverage, potentially generating $650-950mm of excess liquidity for Golar.

Risks

Primary risks include delays in COD, particularly for Esperanza and the recently ordered fourth vessel, as well as significant swings in asset financing conditions. However, we do not view this as a significant risk given management’s extensive track record in fleet expansion and on-time deliveries. Regarding counterparty risk, Gimi is the least exposed asset, given the GTA project is primarily owned by BP (56% ownership). Esperanza and Hilli will be deployed in Argentina, where the current Milei government’s priority is to bring Vaca Muerta’s oil and gas reserves (4th and 2nd largest globally, respectively) to market. We believe YPF’s involvement, recent political reforms (RIGI) and alignment, along with fleet mobility significantly reduce this geographic risk.

Appendix

Early termination run-off model:

Disclaimer: As always, none of this is financial advice, do always your due diliguence before investing in any securities.

Financial Overview

The Q1 2026 earnings were key to show the Comerica integration. Reported EPS was $0.15 per share (Adj. EPS $0.83 accounting for $635mm a-related charges). Adjusted ROA came in at 1.12% and adjusted ROTCE ex-AOCI at 13.7%. That’s a step down from the pre-deal level (Q4 2025 standalone was adjusted ROA 1.41%/adjusted ROTCE ex-AOCI 19.6%). This represents a partial integration of the acquisition since it closed in mid Q1 therefore we should see a similar performance in terms of asset growth for Q2, and from there with the Comerica numbers fully baked in start observing significant cost cutting synergies from Q2 onwards.

Under the current loan-to-deposit ratio of 76% (Q1 2026) from 70% in the 2025, we believe the company has room to further increase its amount of loans which will add around $300 mm to NII (accounting for 7 bps of NIM expansion) making the conservative assumption that the loan book grows in 2026 only to 78% of total deposits. The main reason for this estimation is that most of its peers already have a loan-to-deposit ratio of between 80% and 85%, however, for the sake of conservativeness we prefer to assume no expansion of this ratio going forward and the company would still be at an attractive valuation.

Growth Drivers

The asset growth paired with the current balance sheet mix is the main driver of earnings growth over the next years. NIM expanded 17bps sequentially to 3.30% in Q1, FTE NII jumped 26% QoQ to $1.939B, and management is guiding another 3-5bps of NIM expansion in Q2. On the asset side, management has been direct, they’re leaning into commercial growth, targeting a 60/40 commercial/consumer loan mix over time, funded by a 60/40 consumer/commercial deposit mix.

From an asset liability perspective, we also see the company´s balance sheet aligned with our macro view of rates higher for longer in the long end of the yield curve, and therefore we forecast the following sensitivities:

• ^{Note: Using our own deposit beta estimates}

• ^{Note: Using average balances}
  

Valuation

We see the current valuation of the company as an attractive entry point both in terms of valuation and risk/reward asymmetry. From just looking at the numbers, we understand that the price may seem expensive, however we have to factor in the inclusion of Comerica in the latest Q1 results, and there is a lot of room for asset and earnings growth.

Expecting total assets to grow by more than 10% in 2026, we believe the company could end the year trading at approximately 2.0x P/TBV, slightly above the peer average. This highlights how little TBV dilution the Comerica acquisition creates despite its scale. Beyond 2026, we assume more modest asset growth to remain conservative over the longer term. Even under these assumptions, earnings growth remains attractive, driven by operating leverage, synergies, and cheap balance sheet mix, leading us to forecast a FY2027 ROTCE of 18.9%. Combined with a growing loan book funded by a low-cost deposit base and management's track record of consistently outperforming its own conservative guidance, we believe the market continues to underestimate the company's earnings upside potential. Our price target is approximately $72 per share in our base case.

The Comerica Deal

Fifth Third closed its all-stock acquisition of Comerica on February 1, 2026 (announced Oct 2025, deal value went up to $12.7B by close as FITB’s stock ran). Legacy FITB holders kept 73% of the combined company. The combined bank is now 9th largest in the US ($294B assets), sitting in 17 of the 20 fastest-growing large metros, largest retail deposits in Michigan, and a real Texas/Southeast footprint they didn’t have before. By 2029-2030 they’re expecting 150 new Texas branches and 1,750 branches total, more than half in the Southeast, Texas, Arizona, California.

We view the acquisition as a compelling long-term growth catalyst for Fifth Third. Beyond materially increasing the size of the balance sheet, the transaction provides a significant opportunity to improve the earnings power of an underperforming franchise. Comerica has historically suffered from weak execution, inefficient capital allocation, and persistent operational underperformance. Under Fifth Third’s stronger management team and disciplined capital allocation framework, we believe the combined company can unlock substantial value through balance sheet optimization, expense synergies, and improved profitability, ultimately supporting higher dividends and the resumption of share repurchases.

The transaction is also attractive from a valuation perspective. Tangible book value dilution was limited despite the acquisition premium because Comerica’s tangible book value was heavily depressed by unrealized AOCI losses. At approximately 1.1x tangible book value excluding AOCI, Fifth Third was able to pay a premium while preserving tangible book value for existing shareholders.

HoldCo Asset Management, an activist investor in Comerica, previously argued that the franchise represented an attractive turnaround opportunity under stronger management. Notably, Fifth Third acquired Comerica at a valuation below HoldCo’s lowest implied valuation range, reinforcing our view that the company acquired a high-quality deposit franchise and attractive banking assets at a meaningful discount to intrinsic value.

^{Source: HoldCo Asset Management Presentation: To The Board of Directors of Comerica Inc.: We Echo Mayo – If Not Now, Then When?}

Over $850 mm were targeted as synergy pool from the acquisition, with management flagging potential to over-deliver; brand/systems conversion targeted for Q3 2026. Once of the main drivers of cost-cutting synergies is the geographic overlap in operations mainly across Florida, Milwaukee, Cincinnati and Detroit.

Catalysts & Risks

The main catalysts for Fifth Third looking forward are:

• Systems/brand conversion (Labor Day weekend 2026). This is the trigger for the remaining part of the $850mm annualized cost-synergy target.

• Buyback resumption (H2 2026). Repurchases were paused during the acquisition and haven’t restarted, meaning the capital-return lever on top of the 3.5-3.6% dividend yield remains an attractive shareholder return policy, especially given the significant size of the program, which contains 100 mm shares, of which 93.1 mm remain undrawn.

• Q2 Earnings. NIM already moved 17bps to 3.30% in Q1, with guidance for another 3-5bps in Q2, driven by falling/stable deposit costs (interest-bearing deposit costs -27bps YoY). Management also has a history of being conservative with guidance, so it is possible that another raise in guidance comes for FY2026 in the following quarter as well.

Key Risks:

• Integration execution risk: The full systems and brand conversion will not be completed until Q4 2026, making any disruption during the integration process the single largest risk to our thesis. However, given the significant overlap between the two franchises and Fifth Third management's strong execution track record, we believe the upside at the current valuation is materially larger than the downside. Assuming synergies impact is over $200mm lower than expected in our base case scenario (and guidance), we see a downside of 15% from the current price of $55 per share. That would imply an EPS 20% lower than consensus estimates, so it is a significantly bearish assumption that should set a floor in terms of the maximum risk of the thesis.

Disclaimer: As always, none of this is financial advice, do always your due diliguence before investing in any securities.

Fundamentals

Building Automation Segment

Building Products (61% of revenue): Hardware-driven, fire detection and suppression systems (VESDA brand for aspirating smoke detection), security and access control (LenelS2 brand widely deployed in data centers), HVAC controls, building management systems (BMS), sensors, and energy management instruments. Building products grew 8%in Q1 2026 fueled by double-digit growth in the fire business in North America.

Building Solutions (39% of revenue): The services and projects arm. This includes installation, commissioning, long-term service contracts, and systems integration. It’s a stickier, recurring-revenue-oriented business that’s been the standout growth driver recently. Building solutions grew 8% organically in Q1 2026, led by strength in services.

The assumptions we made for the segment in particular were the following:

The main drivers for these assumptions are the strong data center build out that we are currently seeing given heavy AI investment has no sign of stopping and this continued trend accelerated the need of some key products that both of these segments provide. Among these products we can find:

• VESDA Aspirating smoke detectors: gold standard for smoke detectors in data centers given its speed advantage and its gas-based suppression mechanism that avoids water intervention.

• LenelS2: dominant in enterprise software for multilayer control, including biometric, badge readers, video integrated door management and visor management. The large size of some projects being built make the need for this system urgent.

• Li-ion Tamer: leading battery runaway thermal detector that detects off-gas signature lithium-ion battery thermal runaway before it becomes fire, critical for UPS battery banks. Total estimated market is over $11 billion as of 2025 and growing at over 17% yearly.

• Niagara Framework: An open software platform (via Tridium) that connects and unifies disparate building systems (HVAC, lighting, access control) across any protocol into a single web-accessible interface. Effectively the operating system for smart buildings, with recurring licensing revenue.

• Energy Management & Power Distribution: A BMS-driven suite that uses automation and machine learning to give building managers real-time control over energy efficiency and occupant experience. The power distribution layer manages load balancing and demand response, reducing peak energy charges across a facility.

Aerospace Segment

The aerospace segment is composed of the following subsegments:

· Defense and Space: covers products and services used by national governments, defense agencies, and prime contractors across military aviation, naval, ground, space, and unmanned systems. Honeywell equips over 150 defense platforms including the F-35, Blackhawk helicopter, Apache rotorcraft, and various unmanned aerial systems. Products include inertial navigation units, electronic warfare systems, cockpit displays for military aircraft, APUs for fighter jets, and space components, Honeywell’s components are on approximately 90% of satellites in orbit.

o Growth drivers: Defense and space grew 10% year over year in Q4 2025, its seventh consecutive quarter of double-digit growth, as global demand remains elevated. Q1 2026 has experienced a decline in organic revenue growth of 4%, driven mostly by mechanical supply chain disruptions that pressured output growth. However, we see the trend of demand drivers, NATO budget increases, Indo-Pacific rearmament, F-35 ramp, are all structural and multi-year in duration. We believe that in the following quarters, we will see growth in the segment go back to the trend of low double digit growth sale assuming 10% sales growth for the next 2 years.

· Commercial Aviation: Commercial aftermarket covers replacement parts, retrofits, modifications and upgrades (RMUs), and MRO services sold to airlines, aircraft operators, cargo carriers, and independent MRO shops to maintain and upgrade aircraft already in service. This is where Honeywell’s installed base moat is most visible, once its systems are on an aircraft, the operator comes back to Honeywell for spare parts, software updates, and regulatory-mandated upgrades for the entire life of that airframe. The aftermarket has structurally higher margins than OEM because pricing power is stronger as operators have limited ability to substitute non-OEM parts on certified aircraft, and regulatory mandates create non-discretionary demand that is price-inelastic. RMUs also particularly valuable given they are often regulatory-driven (ADS-B, ACAS X, connectivity mandates) and therefore non-discretionary from the airline’s perspective.

o Growth drivers: Commercial aftermarket sales increased 19% in Q3 2025, supported by ongoing supply chain unlock, with balanced growth across business jet and air transport end markets. The supply chain unlock is the critical near-term driver Honeywell had significant unmet demand for aftermarket parts in 2023–2024 because its own supply chain was constrained. As those constraints ease, backlogged demand converts to revenue with minimal additional selling effort. In Q1 2026, commercial aftermarket sales grew 3% with ongoing demand strength across the installed base, a deceleration from the 19% peak, but note this is on top of a very high prior-year base, and a 1.1x book-to-bill indicates continued net order intake ahead of current shipment levels.

· Commercial OEM: Commercial OEM covers new hardware sold to aircraft manufacturers (Airbus, Boeing, Embraer, Gulfstream, Bombardier) for installation in new aircraft being built. Revenue here is directly tied to production rates on the major platforms Honeywell is specified into, primarily the Airbus A320 family and Boeing 737 MAX family for narrowbodies, and various business jet programs for the business aviation segment.

o Growth drivers: Volume is directly tied to OEM production rate decisions made by Airbus and Boeing, making this the most cyclical and least controllable channel. Boeing’s ongoing quality and certification difficulties have been a persistent headwind to OEM revenue growth, as 737 MAX production has not ramped at the originally expected pace. The long-term OEM growth case is underpinned by the combined Airbus/Boeing backlog, which represents over 10 years of production at current rates. Every aircraft in that backlog has Honeywell content specified into its design, it will generate OEM revenue when built and aftermarket revenue for the next 25 years after delivery.

Industrial Automation Segment

Industrial Automation is a global provider of industrial automation solutions that deliver intelligent, sustainable, and secure operations for customers in oil and gas, petrochemicals, life sciences, metals and mining, and warehouse and logistics segments. With millions of installed assets, Industrial Automation deploys outcome-based solutions to increase asset utilization; improve operational efficiency and labor productivity; reduce carbon emissions with less energy consumption; and enhance cybersecurity for critical infrastructure and operational assets.

From this segment, two of its sub-businesses have been exited: Warehouse & Workflow Solutions (WSS) and Productivity Solutions & Services (PSS). WWS was sold for approximate $1.5 billion while the PSS sale figure was not fully disclosed, with both transactions closing in Q2 2026. The remaining businesses are the following:

Process Solutions (historically the largest sub-unit): Advanced process control, instrumentation, distributed control systems (DCS), and lifecycle services for heavy industry, refineries, chemical plants, LNG facilities. Also includes smart energy metering and measurement. This unit is being transferred to the new “Process Automation & Technology” segment in 2026.

Sensing & Safety Technologies: Sensors, gas detection equipment, and personal protective equipment (PPE). Honeywell makes industrial sensors across pressure, temperature, position, and flow. The PPE business (respirators, harnesses, fall protection) has been a drag and is being sold. The sensing business is recovering sensing delivered double-digit orders growth in Q4 2025, with strong tailwinds in industrial and aerospace and defense end markets

The assumptions made for the growth trajectory of both of these segments are the following:

One of the main questions we had is whether the company will actually keep the remaining parts of the IA segment or they will further divest. However, we actually see a path in which a turnaround of both segments is possible, specially driven by reindustrialization and exploiting a dominant position in a structurally higher margin industry.

The main assumption for keeping both of these businesses is the reindustrialization of the US, which will significantly boost both the process control systems and lifecycle services for heavy industries (Process Solutions), but will also increase the demand for sensors among other factors.

Given the recent political positioning of the Trump administration is favoring local manufacturing, we believe it makes a lot of sense to actually keep these businesses. Key figures that support this is the estimated 3.8 million new jobs in manufacturing estimated by 2033, many of which will be driven by capacity expansion and industrial investment.

On the Process Solutions front, the current political stand of the US also benefits the company significantly, as the US is already the largest LNG exporter, and the expansion trend in this area is set to continue with the US LNG export capacity expected to more than double by 2029, with planned additions of over 13 Bcf/d:

These figures are key as LNG facilities require distributed control systems (DCS), turbine and process monitoring and gas analytics, among many other services which the Process Solutions segment offers.

We therefore the following points as attractive to maintain Process Solutions as a subsegment:

· Asymmetric switching costs: A refinery running Honeywell Experion cannot migrate to Emerson DeltaV without a multi-year project, plant downtime, requalification of thousands of I/O points, and retraining of every control engineer on site.

· Lifecycle compounding: Approximately 30% of Honeywell’s revenue is generated from recurring aftermarket services, creating stable income with strong customer loyalty and high barriers to entry.

· DCS Market Growth: Honeywell’s Experion leverages low-code blocks that citizen developers can reconfigure without writing script, creating an accessibility moat in the migration of plant engineers toward software-defined control environments.

In terms of sensors, The US industrial sensors market was valued at roughly $17.7 billion in 2025 and is estimated to reach $26.5 billion by 2031, growing at a 7% CAGR. Globally, the market is projected to expand from $30 billion in 2026 to $47 billion by 2031 at a 9.5% CAGR, with the fastest growth in edge-AI sensor nodes that cut latency below 10 milliseconds, and with life sciences and pharma leading end markets at a 10% CAGR.

Given May 2025 divestiture from the PPE subsegment inside of Sensing and Safety Technology, we have seen a severe drop in the revenue for that segment in 2025, however this does not represent the full picture behind the real progress and potential of the sensing business.

Among the sensing portfolio of the company, we can find high-margin systems (DCS field instruments, safety controllers, fiscal metering) which are specification-driven and are deeply embedded in plant architecture, recurring calibration/service tails.

The turnaround in this division has already been underway quietly. Through the first three quarters of 2025 (Q1, Q2, and Q3), the business delivered high single-digit growth, with only some moderation in Q4 2025, when sales increased by 3%. However, this progress has been largely offset by declining sales in the low-margin PPE segment, now fully exited. We remain cautious about the outlook in the segment as the recent geopolitical tensions and cost driven inflation could affect sales, so all these risks are already embedded in our estimates for the segment going forward.

Financial Overview:

In terms of the valuation, we believe that the most efficient way to value this business is through a SOTP model:

Base Case:

Bull Case:

In our bear case we see an upside of around 5%, which means that the stock is actually fairly priced. Even assuming a 15% downside in our bear case should the thesis be completly wrong, the stable nature of the business with clear AI and industrial tailwinds, and the elimination of holding company discount through a well defined catalyst, provide a very attractive risk/reward profile.

Another key assumption in the valuation of the company is the expected IPO value of Quantinuum, the industry leading quantum computing company expected to go public in June 4, 2026. It is the lead company both in terms of performance and advancement in the field, and Honeywell currently has a 49% stake in the company. We assume that is valued at around $25 billion given that they raised over $800 million with pre-money valuation of $10 billion in 2025, and the current US government stake in quantum computing companies sustains this valuation.

In addition, we have been seeing a quantum computing rush in both spending and development as the hype around this technology often overcomes the technical difficulties as well as the poor performance of all model available including IBM and Google quantum computers which often are limited given ion-traps that make it impossible for qubits to process even 10-digit series, let alone have any real implication in daily operations.

In a trapped-ion system, each qubit is a single, perfectly identical atomic ion. Atomic ions provide perfectly identical, high-fidelity qubits. In contrast, superconducting qubits are macroscopic circuits, their properties (resonant frequencies, coupling strengths) vary from qubit to qubit due to fabrication imperfections, requiring individual calibration of every gate.

What they do differently from all competitors is how they trap and manipulate ions. Their approach is the Quantum Charge-Coupled Device (QCCD) architecture, analogous to how a classical CCD image sensor shuttles charge packets between zones.

In order to contrast our valuation and target upside for each scenario, we have attached also the letter of Elliott to the Board of Directors of Honeywell, explaining the potential upside in case of separation of non core businesses.

Elliott´s Letter to the Board of Directors of Honeywell

Risks & Catalysts

In terms of the catalysts that could significantly affect our thesis, we see the following:

· Q2 2026 Earnings

· Honeywell Aerospace (HONA) spin-off in Q2 2026

· Honeywell Industrial Automation Investor Day June 11 2026

With regards to the risks regarding the company, we see three main risks could affect performance going forward:

· Spin-off execution risk

· Lower than consensus guidance in Industrial Automation segment in IA Investor Day

· Further supply chain disruptions in the Aerospace segment that cap revenue growth

Disclaimer: As always, none of this is financial advice, do always your due diliguence before investing in any securities.

Read more

Subscribe now

Fundamentals

In the following image we can see an overview of the company:

Alibaba is listed both in China in the Hong Kong Stock Exchange and also in the US. As of the time of this report the stock currently trades both US and Chinese listed shares at around 129 per share. The stock has fallen from around $155 to $129, an approximately 17% drawdown. In March 2026, the company reported earnings, and results were somewhat disappointing as to what the market expected in terms of both margins and its AI spending. However, we first will dissect all the main businesses of the company and then show our forecast for each of the segments outlining the reasons for each forecast.

The main business of the company involve:

· Ecommerce: accounting for approximately 58% of 2025 revenues, ecommerce can be divided into both China retail and wholesale and International Commerce Retail and Wholesale. These involve the retail sales of the international and Chinese platforms of the company. The Chinese division is currently starting to shows signs of improvement as national consumer trends improve with expectations of China CPI expected between 0.6% and 1% by Nomura, indicating that the strong growth in terms of GDP that the country has been experiencing over the last years will finally convert into an increase in spending capacity. We expect moderate growth of around 3 to 6% for FY2026 in the Chinese segment, driven primarily by the reversion of deflationary trends. The International Retail segment has shown much stronger results over the last year, yet we assumed conservative growth between 10% and 15% across our bear, base and bull scenarios given the current macro volatility and tensions.

· Cloud: The cloud division has grown rapidly, now contributing over 13% of total revenue, driven by enterprise digitalization and AI-related demand. Management targets $500bn in cloud revenue (implying 40% CAGR); we apply a more conservative 35% five-year growth assumption, which still reflects a compelling expansion runway. Structurally, the US cloud market remains 6x larger than China’s, but accelerating migration from private to public infrastructure, combined with rising AI workloads, creates a meaningful catch-up dynamic. Alibaba is best positioned to capture this shift, commanding over 35% of China’s cloud infrastructure market. AI models like Qwen serve as a demand multiplier across compute and PaaS, reinforcing top-line cloud monetization.

On the model side, Qwen has scaled from 3 to 30 trillion training tokens, expanded to 235B parameters across 119 languages, and accumulated over 300 million downloads with over 200 derivative models, early indicators of platform-level adoption dynamics. Within Asia, primary competition comes from Baidu’s ERNIE and Tencent’s Hunyuan, though Qwen leads both on ecosystem scale and developer adoption, consolidating its position as the region’s de facto open AI infrastructure layer.

The latest release, Qwen 3.6-27B (April 22, 2026), provides the clearest evidence yet of the model family’s technical trajectory. The model is a fully dense 27B-parameter architecture, notable because it outperforms Alibaba’s own previous flagship, Qwen3.5-397B-A17B, on nearly every coding benchmark despite the predecessor carrying 397B total parameters. The caveat worth flagging: the 397B model is a Mixture-of-Experts (MoE) architecture that activates only 17B parameters per inference pass, making the comparison one of architectural efficiency rather than raw scale. Even so, a 27B dense model surpassing a 397B MoE on agentic coding tasks is a meaningful signal of training and architectural progress.

On benchmark results, Qwen 3.6-27B scores 77.2% on SWE-bench Verified vs. 76.2% for the 397B predecessor and 80.8% for Claude Opus 4.6, 59.3% on Terminal-Bench 2.0, matching Claude 4.5 Opus exactly, and 87.8% on GPQA Diamond, marginally ahead of Claude Opus 4.5 at 87.0%. Against Google’s Gemma 4, Qwen 3.6 leads across all four primary coding benchmarks: HumanEval (94.8% vs. 92.1%), SWE-bench Verified (68.2% vs. 61.4%), MBPP (93.1% vs. 90.3%), and LiveCodeBench (71.4% vs. 64.7%). The year-on-year trajectory is clear: no open-weight model cleared 40% on SWE-bench Verified in April 2025; Qwen 3.6 clears 68% a year later. On cost efficiency The Qwen 3.5 9B model scores 81.7% on GPQA Diamond, ahead of GPT-OSS-120B at 71.5%, a model 13x its size, and the broader family is available via API from $0.10 per million input tokens (Qwen3.5-Flash tier), approximately 13x cheaper than Claude Sonnet 4.6 for comparable workloads. The chart below benchmarks Qwen 3.6 against key peers across SWE-bench Verified, Terminal-Bench, and GPQA Diamond, the three evaluations most relevant to enterprise agentic and reasoning workloads underpinning cloud monetization.

Another important tailwind on the path to monetizing these AI models is their integration into the automotive industry, a move that could significantly reshape the competitive position of Alicloud across Asia. As of late 2025, management has indicated ongoing discussions with several major automotive manufacturers, including BYD, BMW, XPeng, and Changan, aimed at embedding Cloud solutions into the electric vehicle ecosystem.

This integration would go beyond simple cloud services. It implies the direct deployment of AI models developed through Cloud PaaS, enabling capabilities such as real-time vehicle telemetry, continuous model training, and advanced software-defined vehicle features. In effect, Alicloud would become a core infrastructure layer within next-generation EV architectures.

Given the company’s dominant position in the cloud market, holding roughly twice the share of its nearest competitor, and its strong margins, the case for continued investment in capex and R&D becomes more possible. These investments are not only sustainable but also strategically aligned with capturing growth in an underpenetrated market.

China’s EV market alone was valued at approximately $450 billion in 2025 and is projected to grow at a CAGR of 18.5% through 2030. If Alicloud successfully establishes itself within this ecosystem, the upside could exceed even optimistic projections. In such a scenario, our current bull case may prove conservative, particularly if the company secures a first-mover advantage in integrating AI-native cloud infrastructure into the automotive sector.

Regarding the training of the Qwen model, which has recently been running on NVIDIA H800 GPUs in 32-node blocks holding approximately 30 GB of optimizer and activation states per GPU, it is worth noting that Alibaba has been developing its own silicon through T-Head, its in-house semiconductor arm: the Zhenwu PPU.

T-Head didn’t start out making training chips. They began with RISC-V CPUs and a simpler inference chip called the Hanguang 800, but as Qwen and other large models kept growing, inference-only chips stopped being enough. So, they shifted gears and went after the full GPU-class accelerator market. The result is the Zhenwu 810E, a chip they describe as fully built in-house, both the hardware and the software stack, designed to handle training and inference together.

In terms of positioning, the 810E sits meaningfully above the chips it was designed to displace. It outperforms NVIDIA’s A800 and A100, both Ampere-generation chips built on a 7nm process, and matches the H20 across key indicators, completing practical scale verification in Alibaba Cloud’s thousand-card cluster. To understand what that means technically, it helps to map the full stack of relevant chips.

The A100 and A800, built on NVIDIA’s Ampere architecture (7nm, TSMC), deliver 312 FP16 Tensor TFLOPs with 1.6-2.0 TB/s of HBM2e memory bandwidth, but critically lack a Transformer Engine, forcing reliance on less efficient software-level precision scheduling for FP16/BF16 workloads; the A800 further suffers from reduced NVLink bandwidth (400 GB/s vs. 600 GB/s) due to export restrictions, constraining multi-GPU scaling. In this context, Zhenwu, with 96 GB of HBM2e and 700 GB/s inter-chip bandwidth via T-Head’s ICN, surpasses both in memory capacity and cluster-level communication, supporting its relative performance advantage. Compared to the H20, a Hopper-based (4nm) chip optimized for inference, Zhenwu is competitive in memory capacity (96 GB) and interconnect scaling, but trails significantly in raw memory bandwidth (700 GB/s vs. 4.0 TB/s with HBM3) and likely lacks a hardware-level Transformer Engine with FP8 support, limiting its efficiency in precision-sensitive pre-training while remaining viable for inference and fine-tuning. The gap becomes materially larger versus the H800, which shares the full Hopper architecture with the H100, offering 1,979 FP16 Tensor TFLOPs, 3.35 TB/s memory bandwidth, and Transformer Engine-enabled FP8 mixed-precision training; while its NVLink bandwidth is reduced (300 GB/s) for export compliance, this primarily impacts large-scale distributed workloads. In those regimes, such as Qwen’s frontier training runs requiring tensor and pipeline parallelism, the H800’s hardware-level precision optimization significantly reduces memory and compute overhead, whereas Zhenwu would require larger cluster footprints and incur higher per-token training costs to achieve comparable throughput.

Looking forward, the trajectory of these chips is asymmetric. NVIDIA’s export-controlled tier, the H20 and its successors, will remain capped by US policy regardless of engineering progress, which structurally limits the ceiling for China-accessible Nvidia hardware. Zhenwu’s roadmap faces the inverse constraint: engineering progress is possible, but manufacturing is bottlenecked by foundry access. T-Head almost certainly relies on SMIC at the 7nm node or its N+2 equivalent for domestic production, given TSMC’s tightening restrictions on advanced Chinese customers. SMIC at N+2 imposes real penalties in transistor density, power efficiency, and yield compared to TSMC’s 4nm, meaning a hypothetical Zhenwu 2 could close the architectural gap on paper while remaining behind on physical implementation quality. If T-Head retains TSMC access through a third-party arrangement, a next-generation chip incorporating a proper Transformer Engine analog and HBM3 could plausibly reach H800-class performance within two to three product cycles. Without it, Zhenwu remains a strong inference and fine-tuning platform and a credible H20 replacement, but the frontier training gap to H800 persists.

This has direct implications for Qwen’s development. Inference and fine-tuning workloads, which generate the majority of Alibaba Cloud’s AI revenue today, are already well-served by the 810E, enabling full vertical integration from silicon to model to cloud. Frontier pre-training, however, still benefits materially from the H800 clusters, and that dependency is the residual vulnerability in Alibaba’s self-sufficiency narrative. It is precisely this gap, and the strong investor appetite for any credible Nvidia alternative in the Chinese market, that has driven speculation around a T-Head IPO, which Alibaba was actively considering earlier in 2026, though no timeline has been confirmed.

In terms of the valuation, these segments were modelled individually to estimate the actual revenue of the company following the previously outlined assumptions:

Now having exposed our assumptions, lets do a reverse engineering of what the market is currently pricing in terms of both multiples and also EBITDA per segment:

Even though the SOTP valuation may appear overly simplistic, it highlights a key point: the market is effectively pricing in either a severe decline in TTG Group’s EBITDA margins, beyond the already estimated 100bps contraction we assumed as a margin of safety in our estimation given macro volatility and cost driven inflation could affect our consumer spending expectations in the short term. In addition, we are assuming cloud revenue growth below the CAGR that leads to 5-year guidance of management and assuming margin expansions in line with the past years average increase. In our view, both scenarios suggest a mispricing, particularly in the case of the cloud segment, considering its strong performance and clear path for future growth.

In terms of the valuation, it can be analyzed from multiple perspectives, so we used both DCF and a SOTP approach alongside sensitivity tables for the discount rate and terminal growth rate. It is worth noting that a DCF may not actually be the best way to value the business specially given the very diverse valuation multiples and performance that each segment has. Consequentially, our target in our bull case given the previously exposed assumptions is around $300 (over 133% upside) and our bear case is $160 (18% upside).

Macro

After addressing the fundamentals of the company, we have a clear perspective of the key aspects of the business. However, both the ecommerce valuation and the subsequent valuation of the whole business are discounted give macro factors and uncertainty given recent geopolitical tensions.

In regards to the “macro discount” that the stock currently has implied, we believe that even if we apply a 20% discount across each of our fundamental targets the current discount is unjustified for the following reasons:

· Alignment of interest of the government with hyperscaler’s to become more competitive relative to their US peers, given that the government invested over 57% of the RMB 700 billion in AI investment in China in 2025 through diverse mechanisms (subsidies, special-purpose bonds, etc.). This represents a shared goal for competitiveness that should ease the worries regarding constraining regulation as it has occurred in the past with Alibaba.

· US & China conflict probability remains low in the short and mid term as none of both countries are in economic positions survive any potential conflict.

· Deflationary trends in Chinese consumer are set to revert given the raise of nominal wealth transfer mechanism from the growth in equity indexes and GDP growth into actual consumers.

Deflationary Trends

China GDP grew by 5% in 2025, and per capita disposable income grew by 5% in nominal terms, with consumer prices broadly flat. With very little price increase, this translated into real increase in purchasing power. For two consecutive years over 5% of real income growth with near zero inflation makes inflation only a matter of time. Wages, which are approximately 56% of household disposable income, grew 5.3%, business income grew by 5% and transfer income by 5.7%. The issue was not the income generation but the actual transfer into consumption given wide conservativism in spending due to the wealth destruction caused by the real estate crisis.

Total Chinese household savings currently stand at 160 trillion yuan, with only 5% of this figure invested into equity markets. Meanwhile, saving rates has run above 43% of GDP for years. This figure will not experience very large declines, but rather, we expect that the propensity to consumer per unit of income to increase as a product of continued GDP growth paired with expansionary fiscal policies implemented in 2025 and the also aggressive increase in monetary supply. To this we have to add another layer, which is two consecutive years of double-digit positive return in domestic equity markets, which is a rare event in the case of Chinese markets.

Among the fiscal policies implemented, the local government issued significant debt (approximately 10% of the GDP) to refinance the local government debt and recapitalize banks to provide a floor to the financial system. The government also created a “Trade-In” Program which provided subsidies to consumer goods, which provided strong growth in Q1 and Q2 2025 with 4.6% and 5.4% of growth respectively.

In terms of monetary policy, they implemented a Reserve Requirement Ratio cut twice, cutting over 100bps, injecting roughly 2 trillion yuan in long term liquidity to help banks absorb local government debt swaps. In addition, the 1-year Loan Prime Rate was lowered to 3.10% by late 2025 and the gap between the M2 and M1 showed a clear stabilization of the banks. It is worth noting that as of mid-2025, the M1 began including personal demand deposits and non-bank payment institution reserves.

Looking into 2026, the consensus sees potentially more aggressive easing by the PBOC should consumption data continue be weak, and the deposit rate environment compounds to all of these measures. With one-year deposit yields at around 1.5% and CPI now running at 0.5% (despite it being cost driven rather than demand driven). As real deposit yields erode, households face increasing pressure to either spend, or reallocate into risk assets both of which are inflationary.

Catalysts

Near-term catalysts are centered on Q1 2026 earnings, where the key watch items are consumer trend momentum in the e-commerce segment and the rate of cloud revenue acceleration. Additional potential catalysts include any formalized EV EOM partnerships, further disclosure on the T-Head IPO timeline, and incremental updates on Zhenwu chip deployment and roadmap.

In the medium term, the trajectory of China’s deflationary cycle will be a critical determinant of multiple expansion in the e-commerce business, particularly as consensus begins to price in the 2027-2028 demand recovery.

Conclusion

Alibaba presents a highly attractive asymmetric risk/reward profile at current levels, underpinned by the convergence of solid fundamentals, emerging macro tailwinds, and a valuation that we believe does not yet reflect the full optionality embedded in the business. At its core, the investment case rests on a structurally sound e-commerce business, the largest contributor to group revenue, paired with a cloud business that carries both the highest growth trajectory and the highest SOTP multiple of any segment in the group.

Both segments are, in our view, candidates for meaningful re-rating: e-commerce on a macro recovery and sentiment normalization, cloud on continued AI-driven monetization and vertical integration through Qwen and Zhenwu. We maintain a conservative stance ahead of upcoming earnings, which will be critical in validating whether our $230/ADS price target is achievable, or if fundamentals are strong enough to support our bull case.

Disclaimer: As always, none of this is financial advice, do always your due diliguence before investing in any securities.

Business overview

Marex Group plc is a financial services platform acting as a middleman for commodities, energy and derivatives markets. It operates in four core segments, clearing, agency & execution, hedging & investment solutions, and market making.

The firm is occupying a mixed position where bigger banks cant access due to regulatory and liquidity requirements, but a level which the infrastructure required is enough to prevent smaller companies to enter the space.

Clearing

In the clearing segment the company guarantees and settles trades acting as the clearing broker when a fund buys derivatives (on CME for example). The segment accounts for the 26% of the revenues.

Trading volume, interest rates and the number of active clients are the main drivers for the segment.

Clearing is one of the hardest services to replace in financial markets because it’s built into everything a client does day to day, from managing margin and collateral to risk control . Once a client is set up with a clearing broker, switching is a hassle and can be risky, so most just don’t do it. That means relationships last a long time. By sitting in the middle of all this, Marex becomes a core part of how clients manage their money and risk, making it very difficult for another firm to step in.

Important to say that Marex isn’t a clearing house but a FCM (Futures Commission Merchant), which although being in the same paradigm, its different from CCPs.

- Clearing houses (CCPs) are market utilities. They sit in the center of the system, standardizing risk across participants. Their business is highly regulated, capital-light, and driven by volumes. Once liquidity is concentrated in a CCP, it’s almost impossible to move, which creates a very strong moat.

- Clearing brokers (FCMs), on the other hand, are client-facing. They connect clients to those clearing houses and handle everything around it, margin, collateral, funding, execution, and risk management. Unlike CCPs, they take on balance sheet risk and operational complexity, but in return they earn multiple revenue streams tied to their client relationships.

So, knowing the difference, it is sufficient to say CPPs can’t develop Marex business as they do not get involved with credit related and must be neutral risk.

This creates a structurally advantaged position. Access to CCPs is limited, banks are pulling back due to regulatory capital constraints, and clearing is now mandatory across most derivatives. As a result, clients are effectively forced to go through a small group of intermediaries like Marex, turning the business into a regulated bottleneck within global markets.

Regulation has also reshaped who can profitably operate in clearing. Under Basel III and IV, the combination of capital, leverage, liquidity, and operational constraints has pushed bank returns down to unattractive levels. Many have responded by pulling back or exiting entirely. Importantly, the underlying demand for clearing hasn’t changedit has simply moved to non-bank clearing brokers (IB have strong capital requirements and small peers lack infrastructure), creating a long-lasting and regulation-driven opportunity.

The economics are particularly attractive. Revenues are driven by client clearing fees and, more importantly, by interest earned on client margin balances. As volumes, volatility, and rates rise, both of these revenue streams scale naturally. At the same time, relationships are highly sticky, as clearing is deeply embedded in clients’ daily operations and difficult to switch, creating high margins derived the monopolistic position they find themselves in.

Despite this, Marex is often viewed as a cyclical broker, but far from reality. It sits between exchanges and clearing houses, with scarce access, recurring revenues, and a structural tailwind from regulation and market growth.

Agency and Execution

When a client wants to execute a trade (place the trade), Marex trades on their behalf sending the order to the exchange, so they make money through commissions per trade and execution spread. Offering also prime brokerage services

Main drivers are client trading activity, elevated volatility and number of clients.

Clearing and execution work together to improve returns. Clearing provides the infrastructure but is capital-intensive, while execution is capital-light and higher margin. By layering execution on top of clearing clients, Marex boosts profitability with little added cost or risk, as Clearing enforces strict controls through margining and real-time oversight, which extends to execution activities.

This is most evident in Prime Services, where clearing enables the service and execution drives revenues. The same setup also strengthens risk control, as clearing discipline carries over into execution.

Overall, this integrated model differentiates Marex from pure brokers and has allowed it to scale efficiently. Execution is now the largest segment, with lower margins but much higher returns on capital.

Hedging services

Marex provides tailored hedging strategies across commodities and derivatives, generating revenues through structuring spreads, pricing margins, and financing.

This segment benefits from rising volatility and growing demand for risk management. It sits on top of Marex’s client base, deepening relationships and capturing higher-margin, more bespoke flows, particularly as activity expands in structured products and Asia.

Market Making

Marex provides liquidity by acting as principal, earning bid-ask spreads. Revenues depend on trading volumes, volatility, and spread conditions, but the business requires balance sheet and introduces earnings variability (the integration of Winterflood Securities has boosted the segment a bit).

The segment accounts now for the 11% of revenues, compared to the 24% it accounted for in 2022 or almost half the revenue stream in 2021. The segment will probably continue to lose importance in the company’s sources, but still, it plays a complementary role, supporting liquidity provision and client service, while contributing opportunistically to earnings.

Now, for the valuation logic, risks and catalysts.

Read more

Pillars of the Thesis: Why is the business cheap?

Read more

Shawbrook is a recently re-listed small-cap UK specialist bank that, in our view, trades at a structural discount to its sustainable return profile. The Group is a UK specialist lender targeting underserved, complex or event-driven borrowers in SME, real estate and consumer niches, funded primarily by retail deposits, operating with a technology-enabled but human-underwritten model.

Subscribe now

Read more

VNET Group, Inc., an investment holding company, provides hosting and related services to Internet companies, government entities, blue-chip enterprises, and small-to mid-sized enterprises in China. The company is characterized by its large and high quality portfolio of carrier-neutral and cloud-neutral datacenters across the whole country. We strongly believe that this is one of the most asymmetric ways to play the expansion in capital expenditures led by Chinese hyperscalers. Descalation of macro tensions, shipment of NVDA chips to China, the precise execution of management in securing location critical land and delivering above average delivery rates in its datacenters will also be crucial factors that will help the company outperform.

We will base the analysis in the following segments: investment thesis, macro landscape, payoff asymmetry and risks.

Financial Overview (Base Case):

Investment Thesis

We believe VNET Group stands at the center of China’s AI infrastructure supercycle, where hyperscaler capital expenditure, state-directed policy, and accelerating GPU deployment converge to create sustained wholesale data center demand. China’s AI-related investment is projected to reach approximately RMB 700 billion in 2025, up 48% year-over-year, with roughly 57% funded directly or indirectly by government mechanisms including subsidies, special-purpose bonds, and designated computing hub incentives. Unlike prior technology buildouts, this cycle is explicitly framed as a national strategic priority, reducing cyclicality risk and extending duration of capital deployment. Against this backdrop, China’s leading hyperscalers (ByteDance, Alibaba Cloud, and Tencent) are committing several hundred billion RMB annually toward AI compute and infrastructure, materially increasing power density requirements and accelerating data center construction timelines. We believe VNET is structurally positioned to capture an outsized share of this incremental demand, and that the market has yet to fully price in the durability and scale of this investment cycle.

One of the biggest risks for this type of business model is actually the occupancy rate and the predictability (“stickiness”) of their customers, both in the retail and wholesale divisions. One of the main mitigants of this risk is VNET’s anchor partnership with ByteDance which provides multi-year revenue visibility and direct exposure to the most aggressive AI spender in China. ByteDance doubled capital expenditure from RMB 80 billion in 2024 to RMB 160 billion in 2025, with approximately RMB 90 billion allocated to GPUs and AI accelerators and RMB 70 billion directed toward data center infrastructure. Multi-phase deployments at VNET’s Ulanqab campus, including 28 MW and 55 MW contracted capacity with further phases under discussion, create expansion optionality. While customer concentration represents a risk, the scale and persistence of ByteDance’s AI investment materially de-risks near-term utilization and underpins contracted revenue growth.

However, this anchor partnership not only serves as a recurring revenue measure but also could lead to more long term contracts with several hyperscalers as the company´s delivery timeline of 9-12 months for each of their solely built up datacenters exceeds the industry average of 13-14 months, providing a structural advantage. Also, the strategy of the construction model implemented with the development of pre-acquired land bank as well as regulatory execution for building permits and licenses has set a signaficant difference that makes VNET more attractive for hyperscalers as they need to deploy capital and infrastructure as fast as possible.

The company’s 9–12-month delivery timeline, versus an industry average of 14–18 months, represents a structural competitive advantage in a compressed deployment cycle. Hyperscale’s procuring advanced GPUs face narrow windows to operationalize clusters, particularly given regulatory constraints around chip approvals. VNET’s modular construction model, pre-acquired land bank, and regulatory execution allow it to bring capacity online approximately 40% faster than peers, improving customer time-to-revenue and enhancing project-level IRRs through earlier cash flow realization.

Another key advantage of VNET over peers is location. VNET’s presence in three of China’s eight designated national computing hubs (Inner Mongolia, Hebei, and the Yangtze River Delta) provides meaningful cost advantages and policy alignment. These hubs offer land discounts of 30–50%, electricity costs of roughly RMB 0.28–0.32 per kWh compared to RMB 0.55–0.70 in non-hub regions, and expedited permitting processes. The resulting 25–30% structural total cost of ownership advantage enables either pricing flexibility to secure large hyperscale contracts or incremental margin expansion relative to non-hub competitors. the proximity to these large computing hubs and big cities significantly decreases latency. This translates into a significant cost advantage for the company, especially given their carrier neutrality by allowing for direct, high-speed peering and reducing expensive long-haul data transport.

Datacenter Locations across China:

Datacenters.com (2024) VNET Group, Inc.: China Data Centers - Providers Map in China. Available at: https://www.datacenters.com/providers/vnet-group-inc/locations/china?view=Map

As of 2023, the strategic partnership with Shandong Hi-Speed has created an integrated electricity-computing model that enhances cost stability and margin resilience. The $299 million investment established co-located renewable generation capacity supplying Ulanqab at an estimated discount to prevailing grid rates. With 4,799 MW of wind and solar capacity generating approximately 860 million kWh annually, the arrangement provides material energy cost savings at scale and mitigates volatility risk in one of the largest components of data center operating expense. Despite of this, in January 3 2026, it was announced a partnership of the CEO of the company, Josh Sheng Chen, with game developer Youzu Interactive to buy back the Shandong stake. This can potentially lead to both a strategic partnership with a private equity sponsor or a secondary offering, However, both the intention of the CEO to take this action as well as the institutional interest that this company has spiked, with Point 72 and Blackstone holding over $150M and $100m worth of shares respectively points out that the company may be severely undervalued.

On the operational front, VNET has demonstrated execution capability through rapid scaling while maintaining high utilization. Capacity expanded from 486 MW to 783 MW within nine months, representing 61% growth, while mature facilities sustained 95% utilization and under-construction capacity achieved a 98.7% pre-commitment rate. This combination of aggressive expansion and pre-leased visibility materially reduces speculative build risk and supports confidence in forward revenue realization.

The company’s capital-light “Hyperscale 2.0” strategy enhances return on invested capital and mitigates balance sheet strain. Through pre-REIT structures, joint ventures with provincial governments, and build-to-suit partnerships, VNET targets substantial capacity expansion toward a long-term 10 GW objective without proportional leverage growth. Retaining management and operating fees, the model preserves financial flexibility while sustaining growth.

Finally, VNET’s AI-optimized infrastructure, including rack densities exceeding 100 kW and power usage effectiveness (PUE) in the 1.15–1.25 range versus industry norms of 1.35–1.45, enables support for high-performance GPU clusters such as H200 deployments requiring 30–35 kW per rack. This technical capability differentiates the platform and supports pricing premiums relative to legacy air-cooled facilities that cannot accommodate equivalent load densities.

In order to value and analyze more granularly the company´s operational performance, we proceeded to create the following revenue model:

It is worth noting, that our base case (case 2), is the basis of the valuation of the company, however when we take a more bullish approach that given the progression of the land acquisition and in terms of Wholesale MW expansion we see feasible, there is where most of the asymmetry of the payoff from this opportunity comes from. In terms of the shares outstanding used in the calculation of the EPS, the company currently trades in ADS which are equivalent to around 6 class A shares.

Bull Case Valuation:

Macro Tailwinds

China’s AI infrastructure buildout is being executed as a state-directed strategic priority rather than a purely private sector drive expansion cycle. AI-related investment is projected at RMB 700bn in 2025, growing 48% YoY, with 57% of funding linked directly or indirectly to government mechanisms. That funding mix, via subsidies, special-purpose bonds, and computing hub incentives, reduces cyclicality relative to prior private-led technology cycles and extends the duration of capital deployment. We believe this causes an alignment of interests between public and private sector that suggest a rerating from the China macro discount that these types of companies currently face as the political uncertainty is partially reduced.

Semiconductor export control developments introduce complexity but also urgency. Conditional allowances for GPU exports and capped volumes create deployment windows that incentivize hyperscalers to operationalize compute clusters quickly. Regardless of external supply constraints, domestic AI capability remains a strategic objective, supporting continued infrastructure investment.

With the significant divergence between the investment of hyperscalers across US and Asia, another key advantage is that the Chinese valuations have not yet been punished as severily as US companies given that in nominal terms their investments is orders of magnitude larger. This serves as a relief effect for the Chinese companies as the market is not being so demanding in terms of both the timeline and the cashflows schedule as their US peers. Therefore, marginally we can expect larger increases in AI capex over extended periods of time, whereas we cannot be so sure that this is the case for many US companies.

Variant View

In order to see a bull case scenario valuation, we believe that this could be the main drivers that lead to exceeding street expectations:

1. Wholesale Revenue Mix Shift Driving 300+ Basis Points EBITDA Margin Expansion

Q3 2025 wholesale IDC revenues surged 82.7% YoY to RMB 955.5 million, while adjusted EBITDA margin reached 29.4%, up 140 basis points YoY from 28.0%. The upside case: wholesale now represents a rapidly growing revenue share (37% of Q3 IDC revenue vs. 35% in Q2), and wholesale contracts carry structurally higher margins than retail given lower customer acquisition costs, longer contract durations, and operational efficiencies at scale, and this in enhanced already by the big contract of the company with ByteDance. Management confirmed Q3 wholesale pricing remained stable, indicating no competitive pricing pressure. If wholesale reaches 45-50% of total IDC revenue by Q4 2025 (from current 40% trajectory), EBITDA margins could expand to 32-33% versus consensus of 28%.

2. ByteDance CapEx Acceleration Creating 2026 Revenue Pull-Forward

ByteDance’s confirmed trajectory RMB 80 billion in 2024 to RMB 160 billion in 2026 represents sustained YoY growth in data center infrastructure spending, with China approving over 400,000 H200 chips for ByteDance, Alibaba, and Tencent in early 2026. The catalyst: ByteDance’s Ulanqab deployments (28 MW + 55 MW contracted, with 100 MW framework agreements under discussion) require customers expecting delivery within six months of contract signing. If ByteDance accelerates H200 cluster buildouts to capture conditional chip import windows before potential policy reversals, VNET could deliver 50-75 MW earlier than planned, adding RMB 400-600 million in incremental 2026 revenue at minimal marginal cost given pre-committed infrastructure. In addition, we also see quite possible that this contract with BydeDance creates a “domino” effect with that leads to other potential equally attractive contracts in terms of economics as well as further land expansion beyond company guidance.

Another factor worth noting, is that VNET’s delivery plan targets 306 MW over the next twelve months, with 132 MW expected in 2025-2026 and 174 MW in 2026, while capacity under construction stood at 306 MW as of September 30, 2025, with 741 MW of total committed capacity. Management stated customers expect delivery within six months after signing, creating revenue recognition velocity that could pull forward 2026 revenues into late 2025. With mature wholesale capacity utilization at 94.7% and management confirming minimal speculative build risk, every MW delivered converts to immediate revenue at contracted rates.

Risks & Management

A key point about the company that we could not leave untouched is the company´s management, currently led by the CEO and founder of the company Josh (Sheng) Cheng. Josh takes the CEO role after the resignation of Jeff Dong in April 2024 and as a Tsinghua electrical engineering alumnus, pioneered China's first carrier-neutral data center in 1999 and maintains influential ties through the Internet Society of China and Tsinghua entrepreneur networks, embedding him deeply in the country's tech policy ecosystem. In February 17 2026, Blackstone-affiliated funds triggered a "Transfer Event" and "Minimum Shareholding Event" under VNET's amended $250 million 2% convertible notes due 2027 (originally from 2022), announcing secondary sales of the full principal amount convertible into 147.3 million Class A ordinary shares (at $11/ADS). This dropped their beneficial ownership below 5% post-settlement on February 19, terminating associated corporate governance rights from the 2022 notes and 2020 preferred shares agreements (board observer rights and vetoes), which had been amended on February 13 to extend maturity to October 2027 and recognize $20.1 million accumulated interest. VNET disclosed these events via 6-K on February 16-17, confirming the rights' end and noting Blackstone's remaining direct stake of about 62.4 million shares (~0.64%), with the stock tumbling on the news over 10%. We interpret this market reaction as a consequence of the loss of Blackstone capacity to intervene in management decisions, expressing concerns about management as they had a previous history of volatility in terms of both operational results and financials.

Given previous history of the company, it is clear that the value unlock of the assets that the company currently has are entirely contigent in the CEO and the relationships between China and US, and this is probably the main risk that this idea has. We are comfortable taking this risk given that we believe in terms of cashflows payoff is incredibly asymmetric and therefore it is worth the risk of our base case which is relatively in the shorter term. However, from an asset based valuation framework, the company we believe could even have more upside, yet this approach to this trade is so contigent on the risk factors previously mentioned that we did not use it for the purpose of having a more conservative view. Bearing this in mind, we took a moderately sized position in the VNET with a target of $17.3 per share.

Note: This material is provided for informational purposes only and does not constitute financial advice. Investors should perform their own due diligence and seek guidance from a licensed financial advisor prior to making any investment decisions.

Humm Group Limited (ASX: HUM), formerly FlexiGroup Limited, is an Australia-based diversified non-bank financial services institution founded in 1988 and headquartered in Sydney, New South Wales. The company operates across five markets, Australia, New Zealand, Ireland, the United Kingdom, and Canada, with approximately 484 employees. As of November 2025, the company maintains a market capitalization of A$361 million with a stock price of A$0.75 per share. Andrew Abercrombie serves as Non-Executive Chairman with a substantial 29.2% ownership stake, while Angelo R. Demasi assumed the CEO role.

The company experienced significant board instability, including mass resignations in June 2022 following the failed sale of Humm Consumer Finance to Latitude Group due to valuation disagreements. Frequent director changes raise questions about strategic consistency and governance stability. Insider selling of A$1.8 million occurred in May 2025 during a board overhaul, presenting mixed signals despite executives being net buyers overall. Concentrated ownership with the top five shareholders controlling 51% of voting power creates governance dynamics requiring monitoring. However, due to questionable actions taken by Andrew Abercrombie besides the clear conflict of interest related to his concentrated ownership position, we believe this company is materially undervalued due to a management discount that should likely rerate if Raper Capital´s proxy fight in February 2026 succeeds.

Read more

Read more

Read more

The managed-care sector has been navigating a high medical-cost cycle, with elevated MLRs pressuring margins across Medicaid and Medicare plans. Premiums have lagged behind rising utilization and reimbursement complexity, creating temporary valuation dislocations. At the same time, budget-constrained states are tightening RFP standards, rewarding the most cost-efficient operators. As 2026 premium resets approach, the sector is positioned for margin normalization and selective re-rating among disciplined players, and we believe this creates attractive opportunities for players like Molina.

Why is the company undervalued:

· Molina Healthcare presents an attractive valuation relative to its peers: the company has managed to survive while avoiding cash bleeding one of the hardest cycles of raising medical costs, reaching an MLR ratio of 92,6% in the last quarter. Given our model this shows significant stress in margins compared to previous years. However, given the company’s strong cash position, great management and cost efficiency, we think this is the best suited company to navigate these hard circumstances.

· Strong cash position, distressed valuation despite being of the top players in the Medicare and Medicaid sector, with the company repurchasing stock. The company is likely to be able to acquire unprofitable smaller insurers and managed care providers in highly favorable terms given its cash position, bootstrapping the clients of the acquired insurers and resulting in significant synergies while gaining market share.

Risks:

· Extreme shift in government policy towards Medicare and Medicaid, which seems unlikely given that it’s the most cost-efficient healthcare program in the US and there should be an alternative proposed which would imply a significant restructuring of the whole healthcare and insurance sector in the US. This risk is being characterized by structural, large size shifts in policy, but in our long thesis is embedded that we may see “policy volatility” as well as changes according with the political agenda, these are not the changes that would risk the thesis, but rather much larger and structural ones.

· Higher for longer medical costs while setting a structure that does not allow, or delays the catching up of premiums to the increasing raise of medical costs.

Deep Dive into the Fundamentals

Before getting into the financials of the company, let’s do a deeper dive into how does this company make money. Molina Healthcare is a managed healthcare provider that offers care programs in PMPM (fixed per patient per month) programs funded by government, 40% of these programs are funded by the state and the other 60% are funded by the federal government. Molina, Centene Corporation, and other MCOs (Managed Care Organizations), basically have to win an RFP to win this government contract and be able to get paid their fixed PMPM from these government subsidized programs. Once the bid for the RFP is presented, the state chooses, and the performance of the company is mostly determined by its cost efficiency, since the payments are fixed. Over time, premiums catch up to the medical costs, but when the lags occur as contracts are set for 3 to 5 years, it presents a great opportunity if you can identify which company is better posed to pass all necessary quality standards while operating cost efficiently and maintaining a healthy margins, since at these times all the MCOs will trade at a discount. Also, it’s worth noting, that for extreme cases, the State puts a floor and a cap in the margin of the company, such that costs can be reduced extremely to the point where quality standards are compromised. And the same thing occurs for the downside, the state offers floors, in which if claims were to raise to a point where it could endanger the profitability or even solvency of the company, the state would cover these expenses.

Now that we have defined the main idea behind how the company makes money, let’s dive into the fundamentals.

Currently Molina Healthcare trades at $142,38 per share. The current market cap is $7,32B, with an EV of $6,76B. The company has over $4,2B in cash and equivalents, and $3,36B of total debt. Our financial projections are the following:

Our main assumptions for the model are the following:

In our model, we assume a gradual normalization of profitability and efficiency metrics over the forecast period (2025–2031), reflecting the company’s adaptation to the high medical cost environment and its continued operational discipline.

• Operating Margin: We project the operating margin to decline from 5% in 2023 to 4% in 2024, stabilizing around 3% from 2025 onward as the company absorbs cost pressures but maintains efficiency gains.

• Interest Expense (% of Operating Income): We assume interest expenses will temporarily rise from 7% in 2023 to 14% in 2025, before moderating to around 10% from 2027 onward, consistent with deleveraging and lower financing needs.

• Net Profit Margin: We forecast a gradual decline from 3% in 2023 to 2% from 2025 onward, reflecting a more conservative profitability outlook amid persistent cost pressures.

• SG&A (% of Premium): We assume SG&A expenses will improve from 8% in 2023 to 6% from 2027 onward, driven by operating leverage and continued cost discipline.

• Medical Loss Ratio (MLR): We model a modest increase from 88% in 2023 to 91% from 2025 onward, reflecting the industry’s elevated medical cost cycle.

• EBITDA Margin: We forecast EBITDA margins to remain broadly stable, decreasing slightly from 5% in 2023–2024 to 4% from 2025 onward, consistent with our assumptions on cost normalization.

We began our valuation with the P/E multiples approach, as we like to compare different methods and assess how their results differ. In this case, we also include a DCF analysis, since revenue growth primarily depends on the company’s ability to secure new contracts. While a DCF model places significant weight on the terminal value and requires long-term projections, we believe it is particularly appropriate here. The company’s business model provides highly stable cash flows from existing contracts, making the DCF a solid framework for assessing its intrinsic value.

In addition, we are likely not the only ones who believe the company is undervalued, as it is already executing a $1 billion share repurchase program. The following table presents a scenario analysis showing the estimated EPS accretion under each case:

Building on our earlier assumption that, given the current high-cost cycle in the industry, the company will be able to acquire smaller unprofitable competitors, it is worth noting that this strategy has already been successfully implemented several times. Some notable examples include the following:

· ConnectiCare: acquisition closed in February 2025 for $350m with 140.000 members across Marketplace and Medicare.

· My Choice Wisconsin: acquired in July 2022 for $150m, for long-term care and Medicaid services.

· Affinity Health Plan: acquired in September 2020 for $380m, for Medicare and Medicaid services in New York.

Comparables

We strongly think that Molina is one of the most efficient MCOs with respect to the SG&A ratio, given their exclusive focus in the managed care operations in Medicaid and Medicare. In the following table, we provide a detail historical comparison of some key financial metrics of the company compared to two of its competitors:

Molina trades at deep discounts across P/E, P/S, and EV/EBITDA despite peer-level fundamentals, with its 5× EV/EBITDA far below UNH/ELV. Versus Centene, a modest premium is justified by stronger execution and Medicaid mix. The table highlights clear multiple upside for MOH relative to the group.

The company despite being currently cheap relative to its peers, which compared to some we can understand given its political risk, it had operating outperformance over the past 7 years as the company achieved lower SG&A ratios while maintaining in line MLR ratios. In the following table we can se a historical comparison of key financial metrics for Molina and two of its biggest competitors:

Catalyst and trade execution

Our catalyst for this stock is the premium readjustment expected at the beginning of 2026. We believe this adjustment will trigger the margin expansion the market is waiting for, allowing the company to re-rate toward our $200 fair value estimate, even under conservative assumptions.

We are starting with a small portfolio weight, as we prefer to see early confirmation of our thesis and continued fundamental strength over the next few quarters. Should we observe a deterioration or persistence of the risks previously explained, we would reassess and potentially reduce the position.

Read more

We currently have reason to believe the market is seriously undervaluing the stock given political measures took by president Donald Trump regarding immigrants (which are the strongest least elastic demand of the core products of the company, beer) and this has led to an overly pessimistic market valuation, having approximately a 10% FCF yield and with the following year being the Football World Cup, which takes place in US, the strongest commercialization point of the company and with lots of incoming tourists from many nationalities that do have strong drinking cultures.

Investment Thesis

The company is currently undervalued due to:

· Sale of low return assets to focus on the core business:

o In April 2025, Constellation divested a set of mainstream, lower-margin wine and spirits brands to The Wine Group. These “toxic” assets were dilutive to growth, carried weaker brand equity, and consumed disproportionate capital for limited returns.

o Post-sale, the retained portfolio is premium and growth-oriented: Corona, Modelo, Robert Mondavi, The Prisoner, Ruffino, Casa Noble, High West, and other high-end spirits/wines. This move mirrors a strategy seen across global beverage leaders, shedding commoditized brands to focus on premiumization, where margins and pricing power are structurally higher

o Management expects $200M+ in annual cost savings by FY 2028, improving free cash flow and strengthening balance sheet flexibility.

· Extreme pessimism over FY 2027 sales:

o Current political measures of president Donald Trump alongside with a younger generation that is not too inclined to drinking and with strong presence of body weight drugs, made the company cut sales outlooks for the second time for FY2026. However, the current stock price is discounting negative growth in the revenue of the company, assuming there is persistence in these trends, however in the last football World Cup years the company did over 8% of revenue growth in the last 2 World Cups and they were not even in the US, the key selling point of the company.

· Efficiency relative to peers is higher despite of underperformance in sales:

o Using peer comparison (TAP, BUD, ABEV), Constellation demonstrates superior efficiency and profitability, yet trades at distressed valuation levels give recent news. This shows that market is selling the news rather than actually pricing the highest probability outcome for the company in the mid-term.

o Capital Management:

Constellation Brands currently has a share repurchase program of approximately $4 billion, of which it used $660 million in the first half of 2025 to start repurchasing. In addition, the company has a dividend yield of 2.9% with a strong growth of 11.07% over the last 5 years. This capital management policy is much more attractive than the one of competitors such as Anheuser-Busch InBev (BUD), which has a lower dividend yield of approximately 1.8% and a negative 5-year dividend growth rate (-4.6% CAGR) despite having completed a $2 billion share repurchase program in June 2025. Similarly, it compares favorably to Ambev (ABEV), which offers a dividend yield of approximately 4.1% but provides no consistent, large-scale share repurchase program and has a highly variable dividend growth profile. Also, the policy is more balanced than Molson Coors (TAP), which has a higher dividend yield of approximately 4.0% and recent strong 3-year growth (around 7.72%), but does not have any share repurchase program.

In the following table, we present a detailed comparison of the company’s main competitors, analyzing profitability margins, year-to-date (YTD) valuation metrics, and stock performance. This comparison provides a clearer view of the company’s relative positioning within its peer group and helps contextualize its current valuation and operating efficiency:

Deep Dive into the Fundamentals

The company’s Q2 2026 results underscore the ongoing headwinds facing the alcoholic beverage sector, with revenue down 15% quarter-over-quarter. Nevertheless, management delivered an EPS beat through swift and disciplined cost rationalization, demonstrating strong operational agility amid weakening demand. For FY 2026, revenues are projected to fall approximately 11%, a decline that appears fully priced into the current valuation.

Beneath the surface, the company’s fundamentals remain notably resilient. Trading at a 9% free cash flow yield and maintaining robust pricing power, the stock offers an attractive opportunity to own an inflation-hedged, cash-generative franchise at a discounted valuation, particularly as market capital continues to chase high-multiple technology and AI names. Investors are effectively acquiring a defensive, inflation-protected asset at a level that already embeds a bearish macro narrative.

Looking ahead, we project nearly $2 billion in free cash flow by FY 2027, supported by both volume and pricing momentum. Management’s expansion plans target brewing capacity of 55,000 hL by FY 2028, a 14.6% increase from the current 48,000 hL base. Our model assumes a $2.6 increase in revenue per liter and a 2.38% annual price escalation, without incorporating any margin expansion, an intentionally conservative stance given ongoing efficiency initiatives. Even under these assumptions, the company’s earnings and cash flow trajectory remain intact.

From a balance sheet standpoint, the company remains financially sound, with Debt-to-EBITDA at 1.83x and the majority of maturities extending beyond 2028, ensuring ample liquidity and flexibility to sustain both debt service and shareholder distributions.

Relative valuation further strengthens the investment case. Sell-side analysts acknowledge that the disparity in FY 2027 revenue expectations between this company and peers lacks a clear fundamental basis. Despite comparable exposure to changing consumption habits, particularly among Gen Z, other brewers trade on stronger growth expectations (ABI +5.2%, Ambev +4.9%, Molson Coors +1%). Given this company’s superior governance, shareholder alignment, and forthcoming World Cup catalyst, the current discount appears unjustified.

Lastly, our buyback FY 2026 accretion analysis highlights the capital allocation strength underpinning intrinsic value creation. The company’s $4 billion repurchase program generates $3.3–$5.3 of value per share repurchased (execution range $135–$160), equating to 5–6% EPS accretion before any operating leverage effects. With an intrinsic value of $228 per share, management is effectively buying $1.00 of value for $0.629 (base case), an attractive arbitrage that compounds per-share value independent of revenue recovery. This reinforces our conviction that the company’s risk-reward remains highly asymmetric, with downside protection and compelling upside into the 2026 World Cup cycle.

Our estimates for the income statement of the company are the following:

Regarding the assumptions below the income statement, these represent our base case assumptions that drive the model’s outputs. We assume a base revenue growth of 5.5% for FY 2027, justified by the average sales growth observed during previous World Cup years. A moderate increase in COGS is projected, proportional to the rise in revenue, while SG&A expenses are expected to remain stable as a percentage of revenue. Interest expenses are assumed to decrease slightly over time, reflecting lower interest rates. We have assumed little to no margin expansion in our base case, which we consider conservative, though this could change in the foreseeable future given the higher margins of the brands retained in the company’s portfolio.

In terms of net profit margin, we believe our assumptions are prudent, yet this metric is not central to our investment thesis, as we place greater emphasis on cash flow generation rather than accounting profits. A key assumption in our model is the reduction of capital expenditures as a percentage of free cash flow, supported by our conviction that this ratio will decline significantly over the coming years due to the sale of low-return assets that historically required high capital spending.

We assessed multiple valuation methods to draw conclusions from the differences in results. Although we do not typically use a Discounted Cash Flow (DCF) model for other companies under coverage, we believe it is the most suitable approach in this case, given the company’s stable revenues and predictable cash flows, which make it well-suited for long-term projection and terminal value estimation. Nonetheless, valuing the company using P/E multiples would yield an implied valuation of approximately $200 per share under our FY 2027 assumptions, providing a useful point of comparison to our DCF-based analysis.

Given this discounted cashflow model we can observe how the company´s cashflows adjusted moderately for the cut in Capex produced by the divestiture in the wine business are posed to generate significant gains in the mid-term, yet the current stock price is far from what the fair value is.

And by this table we can observe in the column side the equity risk premium assumptions and in the rows, we can see the terminal growth rate. For the valuation in the Discounted cashflow model we believe that we have chosen base case assumptions, yet the discount rate could be even lower considering the stability of cashflows, the concentration of the company in high margin brands and the improvement of management over the recent years.

Given the calculation of the sensitivity table and that the results of the DCF valuation are strongly contingent in the free cashflows of the company and therefore in the sales growth, margins and other assumptions, we have identified three cases for the stock:

· Bear Case: The consensus is right and the company will severely struggle to increase sales and will not improve margins over the next three years despite the operational improvements and the reduction of capex caused by the sale of the low return business. We think this scenario is not probable given the significant change in the brands held by the company and the impact that these brands had in the cashflows of the company. We also think is contradicts directly the management of the company as the guidance previously mentioned by management contradicts strongly this case. We assign to this case a probability lower than 20%. With these assumptions the model would return a price ranging from 150 to 160 based in DCF method and multiples.

· Base case: Our base case is clearly stated in the assumptions of the model below the income statement. Revenue growth is assumed at a 5.5% base for FY 2027, with SG&A stability as a percentage of revenue and a slight decrease in interest expenses. The model adopts a conservative view with little to no margin expansion. The capex decreases over the period forecasted, which along with the sales growth expected boost cashflows. We assign this scenario a 60% probability. With these assumptions we see a valuation ranging $210-220 through the utilized valuation methods.

· Bull Case: In the bull case, the revenue growth exceeds the estimated 2% growth that we have assumed for the rest of the model, which both by multiples and by DCF would still return a price ranging between $230 and $240. This would essentially reflect an improvement in margins due to operational efficiency enhancement, continued policy of return to the shareholder by management and also the already assumed strong decrease in capital expenditures. We assume this scenario has a 20% probability.

Macro

In terms of macro dynamics, we have strong reasons to believe that the demographic headwinds affecting the company in FY2026 could reverse significantly in the coming year, driven by the Football World Cup, which will be hosted in the United States, the company’s largest and most profitable market. This event represents a major commercial catalyst, as it is expected to substantially boost demand for the company’s leading brands.

Importantly, the World Cup is expected to attract a large influx of Latino fans, who are the core consumer base for brands such as Modelo and Corona. In both the 2018 and 2022 tournaments, the largest groups of traveling fans came from Mexico, Brazil, Argentina, and Peru, underscoring the strong cultural and emotional connection between these consumers and the event.

Moreover, hosting the tournament in the U.S. makes travel far more accessible for Latin American countries due to geographic proximity and lower costs. Recent reports have already highlighted surging flight bookings from Mexico, Argentina, Colombia, and Brazil, suggesting robust travel demand and, by extension, a potentially strong tailwind for consumption of the company’s products throughout the event period.

Catalyst

The main catalyst for the company is the Football World Cup and the demographic and travel dynamics it will trigger. We expect these effects to translate into a meaningful boost in demand for the company’s products. However, we anticipate that the market will begin to price in this catalyst during the last quarter of FY2026, as management guidance is likely to reflect a more optimistic outlook. This should prompt the market to start discounting the stronger performance we forecast for FY2027, ahead of the event’s full impact on revenues and profitability.

Expected Value of the trade and risk reward asymmetry

In assessing the risk-reward profile of this trade, we find that the company offers a notably asymmetric setup, limited downside with substantial upside potential. The business generates exceptionally stable cash flows, is guided by a disciplined and experienced management team, and benefits from a strong shareholder base that includes Berkshire Hathaway. Its consistent focus on shareholder returns through both dividends and buybacks further reinforces the investment case.

While near-term headwinds such as U.S. deportation policies affecting Latino consumers and tariff-related pressures have weighed on sales, these challenges are already reflected in the share price. The key question for investors is whether the company will continue to underperform even as the largest cyclical tailwind in its industry (the upcoming World Cup) approaches. Although sector fundamentals remain soft and revenue acceleration may take time, we believe downside risk is limited.

Our valuation work supports this view. A multiples-based approach, assuming modest 1% sales growth in FY2026, yields a fair value of roughly $175 per share, while a DCF analysis implies an intrinsic value near $198 per share. Reverse-engineering the current market price of $139.50 suggests investors are pricing in negative sales growth for the coming year, a scenario that has never occurred in any prior World Cup cycle. This indicates that current weakness is largely driven by external macro factors rather than by company-specific fundamentals.

Even under a conservative scenario, we estimate fair value around $175 per share plus a stable 2.91% dividend yield, implying a potential total return of over 21% from current levels. Over the medium term, we expect valuation to normalize closer to $210 per share, supported by portfolio divestitures, balance sheet strength, and renewed volume growth around the World Cup catalyst.

Finally, the divergence between the company’s strengthening debt prices and its declining equity valuation underscores that credit investors retain confidence in its fundamentals. We attribute the equity weakness partly to position crowding, exacerbated by Berkshire’s ownership, and a broader rotation toward AI and technology names.

Based on our model assumptions, we estimate the expected value (EV) of the trade at $202 per share, reinforcing the view that risk is well-contained while upside potential remains significant.

Note: This is not financial advice and the publisher may hold positions in the companies mentioned.

Sources:

Aviation Week (2018) World Cup 2018: where are fans travelling from? Available at: https://aviationweek.com/air-transport/airports-networks/world-cup-2018-where-are-fans-travelling

TOI Sports Desk (2025) FIFA World Cup 2026: Ticket sales surpass 1 million, fans from 212 countries join the frenzy. The Times of India, 23 October. Available at: https://timesofindia.indiatimes.com/sports/football/top-stories/fifa-world-cup-2026-ticket-sales-surpass-1-million-fans-from-212-countries-join-the-frenzy/articleshow/124618449.cms

Constellation Brands, Inc. (2025) Form 10-K: Annual report for the fiscal year ended February 28, 2025. Available at: https://ir.cbrands.com/sec-filings/all-sec-filings/content/0000016918-25-000022/stz-20250228.htm

Constellation Brands, Inc. (2025) Form 10-Q: Quarterly report for the period ended August 31, 2025. Available at: https://ir.cbrands.com/sec-filings/all-sec-filings/content/0000016918-25-000141/stz-20250831.htm

Constellation Brands, Inc. (2025) Form 10-Q: Quarterly report for the period ended May 31, 2025. Available at: https://ir.cbrands.com/sec-filings/all-sec-filings/content/0000016918-25-000075/stz-20250531.htm

Brewers Association (2025) Brewers Association Reports 2024 U.S. Craft Brewing Industry Figures. Available at: https://www.brewersassociation.org/association-news/brewers-association-reports-2024-u-s-craft-brewing-industry-figures/?utm_source=chatgpt.com

Amidst the noise of credit worries in regional banks our focus has to be to analyse each security individually to determine if they really are at risk and whether we should have real concerns about their credit quality, or instead it should be acknowledged that they are now driven by outflows and fundamentals remain unchanged, thus providing an interesting opportunity.

In early September 2025, privately held Mechanics Bank, owned by Ford Financial Fund, executed an all-stock reverse merger with struggling HomeStreet Inc. (HMST), a publicly listed Seattle bank, to create Mechanics Bancorp (NASDAQ: MCHB) (deal was announced back in March 2025 and the ownership split is 91.7% Mechanics; 8.3% HomeStreet).

Subscribe now

Read more

Soleno Therapeutics is developing diazoxide choline controled-release (VYKAT XR), an oral therapy designed to address hyperphagia, the pathological, uncontrollable hunger that drives severe morbidity in Prader-Willi Syndrome (PWS).

PWS is a multisystem genetic disorder resulting from the loss of paternally expressed genes on chromosome 15q11–q13. This genetic defect disrupts hypothalamic development, giving rise to an evolving neuroendocrine phenotype characterized initially by neonatal hypotonia and feeding difficulties, followed later in life by insatiable appetite, reduced energy expenditure, altered body composition, and high risk of severe obesity.

Formulation and Pharmacokinetics

Diazoxide choline represents a new chemical entity that combines diazoxide with a choline salt to enhance bioavailability and control release kinetics. The compound is formulated as an extended-release oral tablet, available in 25 mg, 75 mg, and 150 mg strengths. Upon administration, it is hydrolyzed to diazoxide prior to systemic absorption, achieving peak plasma concentrations 12–16 hours post-dose and steady-state exposure within 7 days. This pharmacokinetic profile enables once-daily dosing with sustained therapeutic plasma levels and improved patient adherence—key advantages in managing a chronic neuroendocrine disorder such as PWS.

Molecular Mechanism of Action

The mechanism of diazoxide choline is rooted in modulation of β-cell membrane potential and insulin secretion. Under physiological conditions, pancreatic β-cells act as the body’s primary glucose sensors, linking circulating glucose concentrations to insulin release. Glucose uptake increases intracellular ATP through glycolysis, which in turn closes potassium channels on the β-cell membrane. The resulting depolarization triggers calcium influx, promoting exocytosis of insulin granules—a process known as glucose-stimulated insulin secretion (GSIS).

Diazoxide choline antagonizes this depolarization process, keeping the potassium channels open and maintaining the β-cell membrane in a hyperpolarized state. By binding to the regulatory subunit of the channel complex (SUR1), diazoxide prevents excessive insulin release, thereby stabilizing plasma glucose and insulin dynamics. This mechanism not only attenuates insulin secretion but also has downstream effects on central energy regulation.

Mechanistic Link to Hyperphagia and PWS Pathophysiology

In Prader-Willi Syndrome, the loss of paternal gene expression on chromosome 15 disrupts key pathways responsible for energy homeostasis and satiety control. Within the hypothalamus—particularly in the arcuate nucleus and paraventricular nucleus—neuronal circuits fail to integrate peripheral hormonal signals such as leptin, insulin, and ghrelin. The result is a chronic imbalance in the activity of orexigenic and anorexigenic neurons: neuropeptide Y (NPY) and agouti-related peptide (AgRP) pathways remain overactive, while satiety signaling is blunted.

Diazoxide choline appears to modulate hypothalamic neuron excitability by stabilizing neuronal membrane potential and enhancing sensitivity to peripheral metabolic cues. This mechanism is thought to restore partial responsiveness to satiety signals, leading to a measurable reduction in hyperphagia and associated food-seeking behavior in PWS patients.

The mechanistic profile therefore bridges two interlinked biological domains:

1. Peripheral metabolic regulation via modulation of insulin secretion.

2. Central appetite regulation through hypothalamic stabilization and improved hormonal signal integration.

This dual action differentiates diazoxide choline from purely behavioral or hormone-targeted therapies and provides a biologically coherent rationale for its observed clinical benefits.

Valuation

In the United States, the condition is estimated to affect approximately 12,000 individuals, with an additional 200 new cases diagnosed each year. On a global scale, the patient population is believed to range between 300,000 and 400,000, underscoring the considerable unmet demand for an effective treatment.

The company’s modeling assumes an average daily dose of 353.8 mg, corresponding to a total annual dosage of approximately 129,000 mg per patient. At a listed price of €5.92 per mg, and after accounting for customary gross-to-net adjustments, the effective realized price is projected at €3.55 per mg. This translates into annual net sales per treated patient of roughly €458,700.

Applying this pricing framework to the expected addressable patient pool yields an estimated Total Addressable Market (TAM) of approximately €6 billion annually at peak. This projection is consistent with an assumption of full penetration of the U.S. population or partial penetration of global cases, reflecting a disciplined and realistic commercial outlook.

Between 2026 and 2028, Soleno’s TAM is projected to rise sharply from approximately €1.1 billion in 2026 to nearly €3.9 billion in 2028, reflecting both increased diagnostic penetration and improved awareness of hyperphagia in Prader-Willi Syndrome (PWS). During this same period, the company’s revenue is expected to grow from €495 million to over €1.78 billion, representing more than a threefold increase in just two years. This acceleration highlights the strong initial uptake anticipated following market entry and the scalability of Soleno’s pricing model.

Operating expenses are forecast to expand in line with commercial execution needs. SG&A costs are expected to rise from €47 million in 2026 to €177 million in 2028, reflecting investments in market access, field force deployment, and patient support infrastructure. R&D expenditures also increase from €94 million to €353 million, as Soleno continues to broaden its development pipeline and explore additional therapeutic indications leveraging its core technology platform. These increases, while significant, are strategically aligned with the company’s growth phase and will underpin long-term sustainability in both product performance and market leadership.

Despite these front-loaded investments, Soleno is expected to achieve rapid profitability, with net income expanding from €247 million in 2026 to €876 million by 2028. Operating leverage remains strong, with gross margins consistently near 90%, demonstrating the high-value nature of the company’s rare disease business model.

Our model also conservatively accounts for share dilution as Soleno scales operations and potentially raises additional capital to fund expansion. The assumed increase in shares outstanding (SO) from 53 million in 2026 to 55 million in 2028 reflects prudent financial planning and does not materially impair per-share value creation.

Based on these near-term growth dynamics, our valuation model supports a fair value estimate over 35% above the current market price assuming a 10% discount rate over the forecasted cashflow. This upside potential is driven by rapid expansion of the diagnosed patient base, disciplined execution in commercialization, and early profitability, all of which position Soleno Therapeutics as one of the most attractive growth stories in the rare disease sector. We strongly believe that the market is currently underestimating the speed of market penetration the drug can have and that combined with the Scorpion Capital short report (which we strongly disagree with based on our experience investing in rare diseases) have caused the share price to be well below fair value.

To better understand this disease, we first have to understand the dense tumor microenvironment of the pancreatic cancer, which makes it very hard to treat. Components:

· Cancer cells themselves

· Immune cells (T cells, B cells, macrophages, NK cells, etc.)

· Fibroblasts and other stromal cells

· Blood vessels and endothelial cells

· Extracellular matrix

The KRAS, is a gene that encodes a small GTPase protein involved in cell signaling. It is a part of the RAS-RAF-MEK-ERK pathway, which controls cell proliferation and survival. Normally, KRAS acts like a switch, it cycles between on (GTP bound) and off (GTP bound) states. Mutations in KRAS lock in the “on” mode, causing continuous growth signals, which causes uncontrolled cell proliferation. KRAS mutations are very common in pancreatic, colorectal and lung cancers.

There are types of T lymphocytes that are critical in the immune response against cancer:

· CD4+ T cells (“Helper cells”)

o Coordinate immune responses

o Activate other immune cells like CD8+ T cells and B cells.

o Can sometimes be immunosuppressive

· CD8+ T cells (“Cytotoxic T lymphocytes (CTLs)”)

o Directly kill tumor cells by releasing perforin and granzymes

o Main t cells for recognizing cancer cells

Elicio Therapeutics Trial

Trial identifier / phase: AMPLIFY-7P — Phase 1/2 study of ELI-002 7P. The Phase 2 portion is randomized (NCT05726864).

Randomization and allocation: In Phase 2, patients are randomized 2:1 to ELI-002 7P versus observation. The trial is open-label, with crossover permitted from observation to vaccine upon confirmed progression.

Enrollment: Company communications report differing totals: approximately 135 patients enrolled as of a December 2024 announcement, and 144 patients in later materials (August 2025 investor/press updates). Both figures are cited by the company in different releases.

Geography / sites: Multicenter trial conducted in the U.S. Company press release notes 24 participating sites.

Indication / population: Patients with mKRAS-driven pancreatic ductal adenocarcinoma (PDAC) in the adjuvant, post-resection setting (after surgery with or without chemotherapy and/or radiation) who remain at high risk of recurrence.

Primary endpoint (Phase 2): Disease-free survival (DFS), assessed by investigators. Company materials describe this as investigator-assessed DFS, with modified RECIST confirmation for new lesions referenced in Phase 3 planning.

Secondary endpoints: Overall survival (OS), safety, and biomarker measures (e.g., ctDNA, tumor biomarker reduction/clearance). Immunogenicity is included as an exploratory endpoint.

Interim and final analysis timing: An event-driven interim DFS analysis was conducted. Following review, the Independent Data Monitoring Committee (IDMC) recommended continuation without modification (i.e., no stop for futility or safety). Final DFS results are expected in Q4 2025.

Immunogenicity (headline result): A company press release (September 17, 2025) reported approximately 99% mKRAS-specific T-cell responders among evaluable Phase 2 patients (89 of 90). The average fold-increase was ~145.3× over baseline. This represents a reported Phase 2 immunogenicity outcome.

The immunogenicity headline essentially indicates that mKRAS-specific T-cell responders—that is, the ability of the mutated KRAS protein or a therapeutic derived from it—to elicit a measurable and functional adaptive immune response mediated by T cells. Among these responders, CD8⁺ cytotoxic T lymphocytes (CTLs) play the key role as “killer cells,” directly recognizing and destroying cancer cells. In parallel, CD4⁺ helper T cells support and coordinate the activity of CD8⁺ T cells, thereby sustaining and amplifying the anti-tumor response.

If the investigators already have access to the data, it is unclear why they chose to delay publication of the p-values and confidence intervals, while simultaneously seeking financing before releasing the results. This raises the possibility that the data may be unfavorable for the primary endpoints, or at least not statistically significant. It is worth noting that immunogenicity is only an exploratory endpoint, so emphasizing it over the predefined primary outcomes could suggest a lack of meaningful efficacy signals.

The primary endpoint of this trial is disease-free survival (DFS), which is assessed using iRECIST (Immunotherapy Response Evaluation Criteria in Solid Tumors), a standardized guideline for evaluating tumor response to immunotherapy.

A key statistical challenge arises from the crossover design, which is common in oncology trials. In this design, patients in the observation arm who experience disease progression are allowed to switch to the experimental treatment. While ethically appropriate, this crossover can confound the analysis by influencing both DFS and overall survival (OS) outcomes, increasing variability and complicating the interpretation of survival curves. Similar issues were observed in their phase 1 trial.

The main catalyst for this program is disease-free survival (DFS), yet the endpoint appears highly compromised. The only trial with publicly available data reported that 13 of 25 patients were censored due to disease progression, a practice that significantly inflates recurrence-free survival (RFS). In effect, patients who failed the study on its primary endpoint were excluded from the analysis, leaving results with little statistical significance.

While the reported immunogenicity headline appears encouraging, it lacks the necessary statistical rigor and adequate sample size to support the conclusion that immune responses are functional or durable enough to eradicate tumors. Furthermore, considering the Correlate of Protection (CoP) threshold, even a strong immune response would be clinically meaningless if it fails to translate into tangible outcomes. Without clear clinical benefit, the trial is unlikely to succeed.

Elicio Therapeutics: Management Overview

The company’s leadership raises significant concerns. The Chief Science Officer lacks substantial experience in advancing late-stage oncology programs, while the board of directors has a poor track record across prior ventures. Companies such as Anfion Biomedica, Connect Biopharma, Excella Health Inc., Vaxart Inc., and Pulmatrix Inc. each suffered losses of more than 90% in value under their stewardship, underscoring a history of underperformance and further undermining confidence in the company’s ability to deliver.

Conclusion

Given the limited reliability of the company’s reported data, the noise and statistical biases observed in the Phase 1 trial, and the unresolved methodological flaws in the Phase 2 trial design, caution is warranted. The decision to highlight immunogenicity results while withholding comprehensive biostatistical data further raises concerns. On this basis, I would recommend shorting with moderate size Elicio Therapeutics stock. Moreover, should the treatment unexpectedly meet Phase 2 DFS endpoints, it could present a compelling case for a high-conviction short ahead of Phase 3, although such an outcome remains unlikely.

Stock data

- Sector: Banks – Regionals

- Price: 15.17

- Market capitalization: 290m

- P/E: 13.3x

- Forward P/E: 10.6x

- P/TBV: 0.8x

- Shares owned by insiders: 19.68%

Pitch

WaterStone Bank is a community bank that has served the banking needs of its customers since 1921. WaterStone Bank also has an active mortgage banking subsidiary, Waterstone Mortgage Corporation.

The story today is simple, the community bank is quietly delivering robust earnings growth, while the mortgage platform was posting losses that are cyclical, not structural. Losses have been reversing since the start of 2024 (with the exception of Q1 2025) sharply when rates decline. This sets up a situation where we have durable cash flow providing a floor and a powerful, rate-sensitive earnings lever providing upside optionality.

In recent years banking results have been decent, in Q1 2025, Community Banking net income nearly doubled year-over-year, reaching $4.6 million compared to $2.6 million in Q1 2024. Net interest income increased by $805,000 to $12.4 million, reflecting strong deposit franchises and stable local lending. The underlying stability comes from the nature of community banking: low-cost, relationship-based deposits; diversified loan books with shorter repricing cycles; and conservative underwriting that shields the bank from outsized credit volatility. These mechanics mean the community arm generates recurring, predictable cash flow across rate cycles, a base that underpins the WSBF equity story even in periods of mortgage weakness.

In contrast, the Mortgage Banking segment posted a net loss of $1.6 million in Q1 2025, down from a $298,000 profit a year earlier, primarily due to a $97.4 million, or 20%, decline in origination volume and tighter gain-on-sale margins. This is the direct result of the national mortgage market’s slump following the sharp rise in long-term rates since 2022. To understand the setup, recall that in 2020–2021 WSBF’s mortgage branch thrived during the refinancing boom. When rates collapsed to record lows, homeowners rushed to refinance, generating an extraordinary surge in origination. Mortgage lenders earn economics on both loan sales into the secondary market and the servicing rights attached. Volumes exploded, gain-on-sale margins widened, and fully owned mortgage branches like WSBF’s scaled beautifully. But as rates climbed in 2022 and 2023, refinancing dried up, purchase demand softened, origination volumes collapsed, and hedging losses plus MSR markdowns drove results negative.

Mortgage underwriting branch: call option on falling rates

The key point is that this drag is cyclical, not structural. The mortgage arm is not broken; it is simply dormant until the rate environment improves. When 30-year mortgage rates move lower, origination activity will recover — first in refinancings, then in purchase demand. Gain-on-sale spreads typically widen when volume returns, hedging losses abate, MSR valuations improve, and scale benefits re-emerge. In this scenario, the Mortgage Banking arm pivots from a $1–2 million quarterly drag to a net contributor. Because the platform is fully owned, all of the upside accrues directly to WSBF shareholders. The equity optionality here is real: a small-cap bank with a steady earnings floor suddenly unlocks a second earnings engine when the cycle turns.

Catalysts are clear: evidence of sequential origination growth, improvement in gain-on-sale margins, and a turn back to positive mortgage results. These will coincide with lower 30-year rates and should prompt a re-rating. Risks are equally identifiable — if rates stay higher for longer, mortgage losses persist; if deposit costs spike or credit quality deteriorates, Community Banking’s stability could come under pressure. But the risk/reward remains favorable: downside is cushioned by the recurring community franchise, while upside is convex to a rate-driven mortgage rebound.

Stable community banking fundamentals provide durability, while the out-of-favor mortgage arm offers cyclical torque tied to the rate path. That blend of earnings stability and optionality is precisely the kind of mispriced setup we look to own — a small-cap financial with a defined catalyst path, limited downside, and significant upside when macro conditions shift.

When the current results from the mortgage banking segment improve, and thus turning EPS higher, we do believe the stock deserves to trade near its tangible book value.

Valuation

With the assumption that falling rates will bolster Mortgage Banking pre-tax profits to nord 6m, EPS ending fiscal 2025 will already better, and thus 2026 EPS causes the implied PE to an undervalued 7.83x, meaning price is likely to move towards high 18’s following a closing discount to the already growing TBV, so a reasonable re-rating toward tangible book would imply the stock trades into the high $18s, offering 20–25% upside from current levels

Capital Management

Waterstone capital allocation is impolute, in Q2 2017, the company had nearly 28 million shares outstanding. To close the valuation gap they’ve been aggressively buying back shares since 2020, reducing the share count to 20 million in Q3 2020 and progressively to an even lower 18m in recent days, repurchasing nearly 3% of the float last quarter (Q2 2025). Obviously, being the share price lower than the book value this makes the buybacks accretive and demonstrates that the company’s management is doing something to address some of the value gap present.

The cash generated by the mortgage origination arm, we think, should be aimed at constantly repurchasing its own shares, as the company has been doing and will continue to do.

Waterstone pays a decent 3,96% dividend, a bit lower than the past two years but still a very respectable yield.

Ownership

Insider ownership for the stock approximates 20% ($57 Million), and while they receive a considerable compensation, their ownership stake is large enough to ensure that management’s incentives are closely aligned with shareholders and that they are highly motivated to see the stock appreciate.

On the institutional side, one notable holder is Renaissance Technologies, currently the third-largest shareholder. While this may or may not carry direct implications, their position is meaningful, and the presence of such a highly regarded fund lends credibility to the stock’s ownership base.

Risks

The company has a low ROE. Not a risk per se, but something to have in mind and possibly a reason why this is a pure trade more than a long-term investment.

Although relatively well diversified portfolio, has some CRE loans, which we generally do not like (obviously it adds extra risk, but the % is sufficiently low).

The thesis is levered to a decline in mortgage rates. Should rates remain elevated for longer, refinancing volumes could not recover, and the mortgage arm could continue to break-even or even bleed some losses.

As for the percentage of variable and fixed rate loans, that does not represent an inconvenient as there is a balanced book.

Position

We are building the position with conviction, targeting upside even beyond the 2022 highs. On my valuation framework, fair value sits in the $18.5–$19.5 range, supported by both earnings recovery and tangible book value growth. With a cost basis of $14.68, we have a comfortable margin of safety and a clear path to 25–30% upside from entry levels.

The pitch is simple, if mortgage branch starts to fuel again and community banking profits stay as usual, the stock could continue to rally as we see a rerate in the multiples caused by a higher EPS and another share buyback (manifesting) with the unrestricted cash provided by the mortgage origination’s profits.

But why hasn’t the stock possibly priced a fairly simple approach? Possibly because the consensus trade is focused on bigger companies, but the path to outperformance, we think is to invest in these types of situations, where it seems the company, which already traded at an attractive valuation, has flown below the radar.

Potential Sale of 250 Water Street: Size, Valuation, and Market Cap Impact

One of the most promising catalysts we identify for SEG is the potential sale or joint venture of 250 Water Street, a one-acre vacant lot in Lower Manhattan's Seaport District. We note that this site spans 547,000 zoning square feet and is fully permitted for a mixed-use development, including 399 multifamily units (with affordable housing components), 219,000 square feet of commercial space, office, retail, and community gathering areas. We confirm that foundation work began in Q2 2022, with environmental remediation completed in Q4 2023, positioning it as a shovel-ready project.

We emphasize that the site's attractiveness stems from its eligibility under New York's legacy 421-a property tax abatement program, which provides up to 35 years of tax exemptions for developments including affordable housing. As a grandfathered "legacy project" under the program (which expired in June 2022 but allows extensions for approved sites), we point out that developers can benefit from significant tax savings, including full exemptions during construction and phased benefits post-completion. We also note that additional incentives like Section 1031 deferrals and potential 485-x incentives for affordable housing further enhance its appeal for sales or joint ventures. However, we observe that the program's completion deadline for legacy projects was extended to June 2031 via recent legislation, creating a time-sensitive window to capitalize on these benefits before potential revocation or non-compliance penalties.

We base our valuation estimates for 250 Water Street on recent comparable sales in Lower Manhattan, as detailed in the attached comps table. These include nine transactions from October 2024 to April 2025, with sale amounts ranging from $60 million (20,000 sq ft vacant land) to $200 million (400,000 sq ft office building), yielding an average price per square foot (PSF) of approximately $1,011 and a median of $722. For commercial office-specific comps, we calculate the average PSF at $485. Applying these to 250 Water Street's 547,000 sq ft:

• Average PSF ($1,011): We estimate the value at ~$553 million.

• Median PSF ($722): We estimate the value at ~$395 million.

• Commercial Office PSF ($485): We estimate the value at ~$265 million.

We note that SEG's invested capital in the site is approximately $283 million, per the portfolio analysis. We project that a sale at the median valuation could generate $112 million in net proceeds after costs, while higher-end estimates suggest up to $270 million net. Recent news indicates to us that SEG is actively mulling offers, with analysts estimating the sale could add $17 per share in value—potentially boosting the current ~$25 share price by 68% and increasing market cap by over $215 million (based on 12.7 million shares outstanding). We believe this unlock could reduce debt, fund developments, or return capital to shareholders, significantly enhancing SEG's balance sheet and investor appeal.

In our equity valuation model, a base-case DCF yields an equity value of $331 million (~$26 per share), but the "250 Water Sale" scenario adjusts cash flows (e.g., adding back tax benefits, reducing capex from $55 million to $50 million in early years) to imply a $45 per share valuation. We observe that SEG's Q1 2025 balance sheet lists developments at $145 million, underscoring the site's material contribution to net investment in real estate ($479 million).

Severe Cost-Cutting in the Tin Building and Jean-Georges Partnership Restructuring

We identify the Tin Building, a 54,000 sq ft culinary marketplace in the Seaport opened in September 2022 at a $200 million cost, as SEG's most challenging asset, often labeled a "toxic" component due to persistent losses ($33 million in 2024) from high operating costs and underwhelming foot traffic. However, we note that aggressive cost-cutting measures implemented in early 2025, including scaling back operations, reducing staff, and optimizing inventory, have begun to stem the bleed. Q2 2025 results show hospitality costs down, with segment EBITDA surging 95% year-over-year.

We see a key turnaround driver in the restructuring of the Jean-Georges (JG) partnership. In Q2 2025, we confirm that SEG completed a corporate restructuring, collapsing the Tin Building joint venture and acquiring 100% of JG's interest, internalizing food and beverage operations across the Seaport. We believe this move eliminates JV inefficiencies, such as misaligned incentives and shared losses, and is expected to boost productivity through streamlined management, better cost controls, and integrated marketing. We point out that SEG retains a 25% minority stake in the broader JG empire, providing ongoing revenue streams from over 40 global restaurants. Consolidated Tin Building revenues contributed to an 18.2% increase in total Q2 revenues to $39.8 million, with management targeting positive cash flow by year-end 2025. We conclude that these changes address years of underperformance, positioning the Tin Building as a high-margin contributor amid rising Seaport visitations.

Events at Pier 17 and Entertainment Portfolio: Boosting Visitor Capacity and Tin Building Turnaround

We observe that SEG's entertainment assets are driving increased foot traffic to the Seaport District, which attracted over 6 million visitors in 2024. Pier 17, the district's anchor, hosted over 200 public and private events in 2024, including 60 concerts at The Rooftop at Pier 17 (ranked #6 worldwide by Pollstar), selling 180,000 tickets. We highlight recent developments, including a 20-year lease with immersive art company Meow Wolf for experiential exhibits and a proposed glass rooftop enclosure to enable year-round concerts starting Q4 2025, potentially adding 1 million annual visitors.

We note that broader entertainment initiatives, such as Seaport Cinema, Summer Restaurant Week, and Racquet Club events, synergize with the portfolio to enhance dwell time and spending. We believe this influx is critical for the Tin Building's turnaround, as higher visitation (projected to rise 15-20% in 2025) addresses years of low traffic, enabling cross-promotions like concert-dinner packages and turning the asset profitable. Q2 2025 entertainment EBITDA jumped 122%, driven by lower per-show costs and higher sponsorships.

Las Vegas Aviators, Air Rights, and Recent News

We view SEG's Las Vegas assets as providing stable cash flows and upside. The Las Vegas Aviators, a Triple-A affiliate of the Oakland Athletics, play at the 10,000-capacity Las Vegas Ballpark (built 2019, valued at $133 million gross), hosting 75 games and 30+ events annually. Q2 2025 saw entertainment costs rise slightly to $22.5 million, but we note that revenues benefited from strong attendance amid Las Vegas's tourism boom. We observe that the MLB license expires after 2030, but renewal is likely given the team's popularity.

We highlight that SEG holds an 80% interest in air rights above the 2 million sq ft Fashion Show Mall (25th largest U.S. mall), with potential for casino-hotel development amid Las Vegas's expansion. Recent quarters highlight synergies to us, with Q1 2025 filings noting $19.5 million in unconsolidated ventures, including air rights. News from Q2 emphasizes growth from events, supporting overall portfolio diversification.

Cost-Cutting Measures, Management Incentives, Debt Structure, and Ackman's Influence

We note that SEG's new management team, led by CEO Anton Nikodemus since the spin-off, has implemented company-wide cost-cutting, including $30 million in annual savings targets through reduced capex ($62 million in 2024) and operational efficiencies. We emphasize that management incentives are strongly aligned with shareholder interests through performance-based compensation structures. Under the 2024 Equity Incentive Plan, executives receive performance-vesting restricted stock units (RSUs), with vesting tied to rigorous metrics. For instance, grants to key executives like CFO Matthew Partridge (target: 20,482 units) and General Counsel Lucy Fato (target: 4,824 units) vest on March 7, 2028, based on performance through December 31, 2027. Fifty percent of vesting depends on SEG's three-year annualized total shareholder return (TSR) relative to the Russell 2000 Index, with thresholds at the 35th percentile (50% vesting), 60th (100%), and 85th (200%). The other 50% is based on 2027 consolidated and unconsolidated asset-level EBITDA, with targets of $10 million (50%), $20 million (100%), and $30 million (200%), plus a 1.5x multiplier if non-GAAP adjusted net income is positive that year. Overall vesting ranges from 0% to 200% of target, subject to continued employment, with accelerated vesting provisions for death, disability, qualifying termination, or change in control. Annual bonuses are determined by the compensation committee's discretion, often linked to company performance. These incentives directly tie executive pay to EBITDA targets, asset monetization, profitability, and stock performance, fostering alignment with long-term value creation.

Ackman's Pershing Square holds ~50% stake (5.02 million shares bought in Q1 2025), driving the full restructuring post-spin-off, including board changes and focus on undervalued assets.

We confirm that debt totals ~$102 million as of Q1 2025 (4.92% senior note for Ballpark, variable-rate Mizuho loan for 250 Water), with $130 million cash providing net cash position. Covenants include minimum liquidity, but we project low leverage (~0.3x net debt/EBITDA) supports flexibility.

We conclude that SEG's combination of asset unlocks, operational improvements, and strong backing makes it undervalued at current levels, with potential for 50-80% upside.