When Microsoft first showed up, it offered the community the same things everyone in the industry offered at that time: some jobs, tax revenue, educational opportunities, and water replenishment down the road. It also brought FarmBeats, sensors and data that help a farmer read a field. A novel offer the farmers could use.

It wasn’t enough. None of what we were offering would repair the land. It wouldn’t give back the dark sky, or clean the air, or return the water. We led with what we make. The place needed us to account for what we take.

So what came back was yes, if — and the if was the soil, the flora and fauna, the view across the field, the water.

Subscribe now

The people who live there gave us the full picture. Through community meetings and farmer outreach run by our people on the ground in Middenmeer — people who work for Microsoft but are of the place. They knew what the sky looked like before. They knew which field floods first. They knew which farmer would take a meeting and which one wouldn’t, and why, going back a generation.

What came out of that work is in the ground now. The if they gave us changed the design, and it changed what we understood the job to be.

You cannot learn that from Redmond. You also cannot learn it by flying Redmond people in for a week of listening sessions.

Structures are stronger than principles

For most of this industry’s history, the Community Relations team arrived after the building was up and running. During my time, it moved earlier into the beginning of a project instead of the end. People fought for that and it mattered — h/t Utaukwa Allen, who is now at Google. But community relations were still not Operations or Design.

Three teams. Three sets of metrics. None of them able to move the others, because no one in the org and nothing in the structure required it.

What closed the gap wasn’t a reorg. It was a fourth team — my team. An innovation team whose whole job was crossing those boundaries. We moved independently of operations, design, and community relations, then carried what each one knew into the rooms where the others decided, and held the full picture long enough for it to become a solution.

That’s the repeatable part. You don’t have to collapse the silos. You have to staff the boundaries.

But we all know that the Regenerative Data Center is not the basis of design at Microsoft today. Pieces and parts are making their way into the basis, but the system solution is still on the shelf. Why?

Because the regenerative design principles stayed principles. And principles leave with the people who hold them.

Most of the people in Redmond were working from the chip out — making sure the revenue-generating asset had what it needed and came online fast. That is a rational place to stand and the company had built its structures around it. Capacity arriving late is a number everyone can see and feel. A data center neighbor’s night sky is not a number at all. When one worry is instrumented and the other isn’t, nobody chooses where their attention goes. It goes to the number.

My team was in Redmond, be we were not beholden to the numbers. We were free to challenge the org’s — the industry’s — assumptions.

Every project we ran started with intentional discovery, meaning we intentionally found new perspectives. We pushed the channels outward — to the sites, to the communities — until we could hear what the metrics had been filtering. Some of the sharpest problems anyone ever handed us came from people at the sites. Literal boundary crossers who work inside the walls all day, and live outside them the rest of the time.

Connecting with them is how we found what had been missed. And we repeated it at locations — neighborhoods — around the world. That was step one in designing something radically different from what the industry had been building for decades.

Narrowing the gap

Last week I wrote that ugliness in this industry isn’t a taste problem, it’s a distance problem: the people deciding what a data center is and what it does never have to live near it. This is the other half of that. The people who can see it clearly are already on the payroll. There just isn’t a path from them to the design decisions that survive leadership changes.

A large campus employs a few dozen people permanently. Measured by headcount they’re a rounding error. Measured by what they can see, they’re the most valuable instrument the company owns.

The fix isn’t a survey. It isn’t a satisfaction score, either — I’ve already written about that kind of number. It gets gamed inside a quarter.

The fix is a boundary-crossing team, but it’s a fragile one. It works while it exists, and while someone with standing keeps insisting on it. I’ve built that team more than once, in more than one industry. My track record is 0 of 4 outlasting me in companies I didn’t build. The structure has to carry what the team was carrying, or the whole thing rests on the people not leaving.

Or, what I’ve done is accept that most of the time I won’t be able to change the structure of the org. So I change its vision — or at least the vision of enough people that good work has a way forward. Pieces and parts. Moving with people as they move up in influence or move on to new places. More often than not, the work emerges in new companies that sensed the early signals too and started with their own intentional discovery of what’s already been solved.

I knew my team was on the latter path at Microsoft. We weren’t in a position to change the structure. That is why we pushed hard to get to something real as quickly as possible. Something buildable and bankable. A system solve for a problem so large that we knew there would be others looking to capitalize on it.

That’s something we can be intentional about too.

What it looks like when principles leave

This January, Microsoft was granted an environmental permit for a second data center in Hollands Kroon, over local objections that included a lack of consultation. In July, LTO — the same agricultural organization we worked with in 2020 — filed an appeal against that permit.

Same municipality. Same objection our work was built to answer.

Nobody reversed a decision. Nothing was undone. There was simply nothing in the structure moving the practice forward or even holding it in place. So when the people went, the practice went with them, and the conversation reset to where it started.

Leave a comment

From The Insider’s Guide to Innovation at Microsoft

• Feedback Loops (#62) — VS Code’s practice has a name: zero distance to the user. The data center industry practices zero distance to the chip.

• Fit People with Phase — Boundary Crossers (#68) — three teams with three sets of metrics drop batons by default. Staffing the crossing is the fix, and it’s the most portable thing here.

• Intentional Discovery (#17) — the channel didn’t exist, so discovery built one. Going wide and expanding your perspective is not a phase you can skip and recover later.

• Needs and Offers (#43) — FarmBeats was an offer. The soil, the sky, and the water were the needs. Leading with your offers and finding the needs later is the chip-out reflex in miniature.

AI Disclosure: This piece was created with the help of AI as an early listener and judgemental reader. The ideas and words are mine.

Say it ugly. Build it better. Onward!

In a data center it’s ten to twenty percent. One industry cost guide says it outright: the mechanical and electrical systems are the project, and the building is just the box they go in.

Fifteen percent of the budget. One hundred percent of what the neighbors experience.

Ask someone inside the industry — someone who has spent a career packaging megawatts of compute, storage, and memory into smaller and smaller footprints, only to string them together into bigger and bigger ones — and you’ll find someone who sees a thing of beauty. A technical marvel. A life’s work.

They’re not wrong.

Ask almost anyone else and you’ll find someone who hates the big windowless boxes. The 24/7 noise and light. The construction that never ends.

Beauty on one side of the wall. Ugliness on the other.

The people deciding what it looks like never have to look at it. So that isn’t a taste problem. It’s a distance problem.

Subscribe now

The true cost

Total cost of ownership (TCO) is a number every CFO in this industry is trying to drive down. Most of it isn’t theirs to touch. The silicon costs what NVIDIA says it costs. Electrical systems and the redundancy standards behind them account for most of the rest.

The shell and grounds are the line items they can cut, so they do. Cheaper and uglier win, because 15% of a big number is still a big number — $20M/MW across 100MW is $2 billion, and nobody walks past $300 million.

But value engineering never asks whether the wall has a function. It asks how to build this wall cheaper. A decade of that gets you a point or two of project cost, thinner and uglier every pass.

Here’s what the point or two bought.

Seventy-five data centers in the first three months of this year — worth $130 billion — blocked or delayed, matching the whole of 2025 in a single quarter.

It has not slowed. New York enacted the first statewide moratorium in July. In Maryland, thirteen counties have passed their own — that’s more than half the state. The mayor of Chicago and the governor of Texas, who agree on almost, have both called for a pause this summer.

A point or two saved. All that lost.

What we built instead

Two weeks ago I shared the design: no dumb walls, ground improvement instead of slab, permeable surfaces instead of pavement, and a mound that reads as land because it is.

What I didn’t do last time was put money on it.

Here’s the money. My team took the walls out rather than thinning them. That isn’t shaving a point from the line — it’s deleting most of it, and the foundations underneath it, and the years of on-site construction that go with both.

We used the displaced soil for berms instead of hauling it off. Planted in native species, the perimeter became local ecology instead of steel fences. Fences stop a person; berms stop trucks.

And the cost? Pennies on the dollar of today’s approach.

The beautiful version isn’t the expensive version with landscaping on top. It’s the version where the expensive, ugly thing never gets built.

That’s why we didn’t try to improve the existing design. You can’t get there with incremental — every gain is a fraction of a line item nobody should be defending in the first place. We went back to blank sheet of paper with 20 years of lessons and technology advancements in hand.

When we added it up a few years ago, just before the ChatGPT moment of 2022, the number was in the billions. It’s not hard to justify 10X-ing that today.

The walls and fence are only the start. Water, land use, jobs, supply chain, capital timing, employee satisfaction, end of life — ask the same cost vs value creation question of each and the pattern holds: the regenerative version doesn’t only cost less, it makes the money the extractive version loses. More on those parts over the next several weeks.

Leave a comment

From The Insider’s Guide to Innovation at Microsoft

• Think-Act-Feel (#27) — In reality, we Feel, Act, Think. And feel is the channel the big ugly wall broadcasts on, all day, to everyone who drives past.

• Sustainable by Design (#25) — embedding sustainability from inception. Inception is the whole argument; there is no later.

• Metrics (#34) — teams prioritize what the incentive system makes measurable. TCO didn’t account for beauty and that has turned out to be an expensive mistake.

AI was used as a drafting partner, research assistant, and judgemental reader. All the ideas, words, and mistakes are mine. Say it ugly. Build it beautiful. Onward!

It digs down six feet or more, hollows out a central chamber, and runs tunnels radiating off it toward the surface. The mud it excavates gets carried up and piled into a mound just as high as it is deep. The tunnels are the plumbing and the ventilation. The mound is the roof. It extends above high-tide and creates niches and landing places for other creatures.

When the builder moves on, the mangrove grow through the mound. Reclaiming the soil, which is why mud lobsters are sometimes called friends of the mangrove.

The mud lobster is one of my favorite models brought to the design table by the biomimicry team supporting my team at Microsoft.

What a stark contrast it provides to how we build data centers!

It uses the materials it has. It builds with the ecosystem and natural cycles. It creates more than it needs in the course of creating what it needs. Abundance in action.

Subscribe now

Data centers, on the other hand, remove the ecosystem — scraping the earth flat is the first step. Then the ground is sealed in concrete so that a 30-year structure can be built to house equipment that changes over every 3 to 5 years. And that was before AI accelerated everything.

Memories of the mud lobster came flooding back last week when a friend in the industry said, “we’ve got to move a lot of dirt.”

Dirt. They have to move an enormous amount of it. Grade it, haul it, truck it off, truck other dirt in. Before anything else happens on a site, the land gets rearranged.

Let’s be clear about something though — not all land is dirt. Most of it is soil. Soil is alive. Dirt isn’t. And when you’re cutting down a forest or taking over farmland, as is often the case with data centers, you can be sure that you are moving a lot of soil, not dirt.

What if we didn’t have to move dirt or soil though? What if we used it? What if we were more like the mud lobster?

Why is there a building?

One of the first assumptions challenged at Microsoft is that data centers need to be in buildings. It’s actually the founding story of the team I joined, predating me by 10 years. They ran servers in a tent through a wet Pacific Northwest winter. Nothing failed. The results of their experiments led to major efficiency improvements. Before their work, we treated data centers like super-cooled clean rooms.

But for all the energy improvements gained at that time, Microsoft held onto the buildings. So my team asked the question again. Why do data centers need to be in buildings?

Not rhetorically. We wanted to know the function. What is that wall doing? Holding up a roof. What’s the roof doing? Keeping the rain out and the people comfortable. Which people? Computers don’t need comfort. They need enough cooling and power. The humans who tend them are there a few hours a month. Ideally, people would not be near the servers at all.

We adopted the design principle of “no dumb walls.” Every wall had to earn its place.

Most of them couldn’t. And once we stopped assuming four walls and a roof, everything stacked on top came loose — the interior air we were conditioning for nobody, the years of on-site work to pour and frame and finish something that isn’t needed by the servers or wanted by the neighbors.

We didn’t end up with no walls. But the walls we kept were manufactured, repeatable, and able to leave when the machines leave.

That’s the distinction that matters. A wall that arrives on a truck is a wall you can take back. A wall poured in place is a wall the place tends to keep.

Why is the ground sealed?

Then somebody asked the same question one layer down. Why do we use so much concrete?

A concrete slab distributes load and holds equipment level. Fair enough. But the ground can often do that work itself. Ground improvement — compaction, aggregate piers, stone columns — stiffens soil using more soil and rock. It’s standard practice, published and proven, and it has been for decades.

Why that matters in earth terms: ground improvement can stay when everything else goes. There’s nothing to demolish and nothing to haul away. The land above it can go back to pasture, prairie, or crops, and nobody would know it was there.

The same question works on the parking and the drives. Pavement is usually there to move water somewhere else — collect it, pipe it, send it away. Away from the place that needs it. Permeable surfaces let it stay where it fell. Bioswales and constructed wetlands clean the water, cool the surroundings, beautify the campus, and provide biodiversity support. All by working with the soil, not paving over it.

None of this works everywhere. Frost depth, expansive clay, a high water table, tight tolerances on tilt — any of them can force discrete footings (minimal, precast, removable) or more. That’s a question you answer with a soil boring though, not a one-size-fits-all spec. And for as many variations of place as there are, there are variations of these solutions.

What it actually looks like

The mud lobster data center we designed was truly underground, in the mud. What follows is what that lesson looks like when it comes back up above ground.

Walk up to one and you see a green hill.*

From the secure campus side you see an open canopy. The walls are compacted earth from the site. The roof is prefab timber, arched, with soil and planting laid over it, so from the community side the shelter reads as land. Inside, the floor is compacted earth, not slab. Precast footings sit at the corners where the soil needs help — concrete, but discrete, removable, and reusable elsewhere.

By the second spring the mound has meadow on it. Birds nest in the gravel at the edges. There are frogs in the swale. Each year, more nature. Each year, more noise and light buffering, more water and air cleaning, more biodiversity and beauty.

Under the canopy, multi-megawatt clusters of containerized compute. We could build to hyperscale with the same stamp-and-repeat design, a must of the industry. Keeping the containers cool in those clusters was a true challenge when we were limited to air cooling. But AI requires liquid cooling — problem solved.

The modules arrive on flatbed trucks. Delivered and installed in days, eliminating on-going construction, matching capital allocation to demand. And every one of those pieces can leave. Nothing about the land use needs to be permanent.

Why is it a steel box?

We took the wall down. We took the slab out. And then we set steel shipping containers on the site.

The containers then were custom. Today, modular data center designs are plentiful (I shared a list of 30+ companies last week).

So ask the question again. What is the container doing? Weather, security, and enough structural strength to survive transport. That last one is why it’s steel — it’s engineered for ocean freight and nine-high stacking. Then it gets trucked once and parked for years. We build it for the worst three days of its life.

Meanwhile the evidence keeps piling up. Our founding story was a tent. Meta is now running GPU clusters under fabric structures on aluminum frames — up in about two months.

We proved the building is optional. Again.

So — be the mud lobster. No dumb walls or pavement. Let shipping containers ship. And at the end of the useful life, leave the farm, forest, field, and town better than we found it.

That was our vision for the Regenerative Data Center. It’s more achievable than ever.

Leave a comment

*The plantings are designed to match the ecology and culture of the place.

From The Insider’s Guide to Innovation at Microsoft

• Questionstorm (#45) — generating questions instead of solutions. “Why is there a building?” took down a wall, then a roof, then a slab.

• Assumption Mapping (#36) — documenting the assumptions behind a decision and revisiting them on a schedule. A permanent building is an assumption nobody wrote down.

• Single- and Double-Loop Learning (#35) — single-loop improves the building. Double-loop asks whether there should be one.

• Aim for Positive (#15) — going past harm reduction. Building so the land can go back isn’t mitigation. It’s the site keeping its options.

AI was used as a drafting partner, research assistant, and judgemental reader. All the ideas, words, and mistakes are mine. Say it ugly. Build it better. Onward!

What follows is a conversation about the psychology of the heretic — why you have to attach your self-worth to a mission instead of an institution, why real heresy begins with conscience (”this is wrong”) rather than confidence (”I’m right”), and how to bring people along a journey instead of handing them conclusions. JoAnn brings the scar tissue: Microsoft’s underground innovators, air cover, arrows, and what three years of reimagining the data center taught her about the cost of seeing the future early.

This hour is for anyone who’s ever been called a jerk (or treated worse) for seeing the future early — and anyone who has one of those people in their organization and hasn’t yet asked what they’re fighting for.

Subscribe now

Key takeaways

If you’re the heretic

• Stop waiting for permission. You will never be granted a license to innovate — Eve didn’t have one either. If you’re waiting for it, it’s already over.

• Heresy starts with conscience, not confidence. The first move isn’t “I’m right,” it’s “this is wrong.” That feeling obligates you to research — humility before conviction — and it’s the conviction people see and mistake for arrogance.

• Socialize the problem, not your answer. Keep building the solution quietly while you make the problem impossible to ignore. When the organization finally feels it, your answer comes off the shelf looking five steps ahead.

• Walk people down the road of heresy step by step. Nobody likes being told what to think. Start them at the moment of conscience and let them arrive at your conclusion as if it were their own.

• Air cover stacks. Every success story JoAnn found had a champion absorbing arrows — and when the champion leaves, the work goes back underground. Find yours, even outside the chain of command, and be one for the people behind you.

If you’re the one watching the heretic

• Before you write someone off as obnoxious, ask one question: what are they fighting for? If it’s something other than themselves, that’s not arrogance — there’s something heroic in it, even if you don’t yet understand the cause.

• If you’re senior, your gift to the system is protection. You were probably a heretic once. Keep your eye out for the young people with radical ideas and help them stick around long enough to make the changes you wish you’d made.

• Don’t ask employees to “be heretics” and expect comfortable answers. A company that invites heresy is really asking for the ideas it wishes it had thought of. Real heretics will hand you the ones you didn’t.

To carry with you

• Your job isn’t to keep your job. Your job is to do something. Untangle your self-worth from the institution and attach it to the mission.

• Learn the anger-to-joy pivot. Anger powers the fight but burns you out; joy comes from looking past the obstacle to the future you’ve already seen. Cycle between them or the burnout wins.

• Good work always finds a way forward. If you saw the problem correctly, your answer gets implemented eventually — maybe not with you, maybe not on your timeline. The better the visionary, the longer the wait.

• Emotion, not data, drives change. People don’t want to be managed by story; they want the authentic account of the moment your heart was touched. Tell that honestly and the change begins.

A final thought from Angus:

Leave a comment

Connections to The Insider’s Guide to Innovation at Microsoft

• Air cover is the book’s recurring ingredient — Angus gave it a psychology. (00:11:19–00:12:43)

• The ClearType story became Angus’s proof case for “moment of conscience.” (00:21:00–00:23:56)

• The Garage validates Angus’s claim that heretics must find each other. (00:17:12)

• AI as a Service proves the book’s point about where ideas actually live. (00:47:21–00:48:38)

• “I’m not an innovator” is the objection both the book and the episode exist to dismantle. (00:08:40–00:09:01)

Thank you for everyone who could join live and to everyone who follows us here and on LinkedIn. Please share with your friends, colleagues, champions, and fellow heretics. It takes the village to overthrow the status quo. Onward!

The first truck idles at the line with its hood higher than his head, grille caked in the same pale dust that has settled on the hedges, the mailboxes, the parked cars on both sides of the street. It is the dust he is standing in. Behind that truck, another. Then another. The line bends around the curve and out of sight, and where it reappears on the ridge it has become a thin brown seam running toward the cranes.

The engines are loud enough that he doesn’t bother talking to the dog. The ground shifts a little each time one pulls forward.

This is the scene every day for years.

Everyone is arguing about living with data centers, the finished building. Water. Electricity costs. Noise. Lighting.

The Regenerative Datacenter of the Future (RDOTF) team at Microsoft — the program I led — argued about all of it too, and we found answers. Heat leaving the building could go into greenhouses instead of the sky. Cooling could take less water than the farmland around it. Noise and light could be designed out.

Then somebody asked what the town experiences while we’re building it.

Ten years of heavy equipment. Ten years of dust, night lighting, engines idling before dawn. Ten years of workers coming and going, which is its own quiet harm — rents rise, and they do not come back down.

Every good decision we had made for the RDOTF would be invisible for years. We realized — we could not design a good neighbor and deliver it badly.

So we went back further than the building.

Subscribe now

Years or months

In 2023, a developer proposed a 1.125 million square foot data center off Tower Road in Morrow County, Oregon. The county’s public works director asked a question most counties don’t think to ask: not what happens when it’s running, but what happens while it’s being built. The engineering firm Kittelson & Associates put counters on the road for a week and modeled it out.

Twenty-four months of construction. Across that period: 113,710 vehicle trips.

Ninety-seven thousand of them are workers’ cars and pickups. Fifteen hundred cement loads. Five eight-thousand-gallon water tankers a day to keep the dust down, which it never quite does. All of it on a town road that carried 1,717 vehicles a day before anyone broke ground.

That’s one building. Campuses are master-planned for several, delivered in sequence. Five to seven buildings at twenty-four months each is at least decade.

Nobody lied about any of this. The decade simply stopped being treated as a decision. It became the cost of doing it at all — unavoidable, nobody’s fault, and therefore nobody’s to change.

That’s the part that isn’t true. The dust isn’t a consequence of building a data center. It’s a consequence of building it on site, one truck at a time. Somebody chose that.

This past March, IEEE Spectrum ran a piece on what a site looks like when the building doesn’t get built there.

You pour a concrete pad. That’s the site work. The compute arrives on a truck — manufactured somewhere else, delivered like an appliance. It gets set on the pad and networked to the ones beside it. One company is planning a campus of fifty. The CEO described the finished site as looking like school buses lined up in a parking lot, completed in six months.

My team at Microsoft, planning for hyperscale sites, estimated one year to complete the site. Berms and flats, trees and grasses, constructed wetlands and bioswales — the whole ecology established within a year. Compute capacity delivered within the first six months and then over the ten years that follow. All hidden within what looks like a community park. Instead of a decade of watching trucks, a decade of watching nature return.

The prototype is in the ground in North Holland.

Nature first

At the Middenmeer campus, we worked with local landscape architects to put 150 native trees and 2,300 square meters of shrubs, grasses and groundcovers into the edges of an operating data center — species chosen to behave like a functioning ecosystem, holding stormwater and stopping erosion, not to decorate a fence line.

Then, this February, on the plot set aside for the campus expansion, the first trees went in before construction started. Not alongside. Before. The community affairs director’s reasoning was that a tree is something people can see, so they can see that you’ve started.

That is the whole idea, executed at small scale on a real site.

I’ll be precise about what that does and doesn’t prove. The perimeter is a prototype of a component, not a demonstration of the whole. The full version — a site complete in a year, capacity arriving on flatbeds across the decade — is a design.

What the prototype settles is the assumption that the data center and nature cannot coexist. The engineers who previously worried about squirrels catching fire (true story) can now rest easier.

The 2nd phase of the prototype proves that nature can come first, with construction happening behind it.

Where this leaves us

I’m not going to tell you modular works everywhere — at 100MW or more, that’s yet to be proven. And communities aren’t getting nothing out of the hyperscale deals. Community benefit agreements are standard now, and some are substantial.

But no amount of the contracted benefit changes the lived experience, and ten years is a long time.

The good news: I found more than forty companies now delivering some form of modular or manufactured data center. Fourteen are purpose-built around it — ECL, Crusoe, Flexnode, Duos Edge AI, Nautilus, Soluna and others — alongside incumbents who have sold factory-built infrastructure for years: Vertiv, Schneider Electric, Huawei, Eaton. Three years ago, when we were planning our first of kind, the list was much shorter. I’ve put the whole landscape, with what each one actually builds, here for anyone who wants to check my work or hand it to a planning commission.

The major hyperscalers have already proven modular units work at the tactical edge. Microsoft, AWS and Oracle have all built ruggedized modular units for defense and remote deployment — command centers, disaster zones, places where nobody is willing to wait two years for a building. It is good enough for the Pentagon. It has not been offered to Main Street.

The fix isn’t an invention. It’s a specification — another kind of contract — and proof.

Until somebody requires it, the trucks will come through every morning and afternoon. And our little boy will have to walk in front of them and get dust on his ice cream.

Leave a comment

From The Insider’s Guide to Innovation at Microsoft

• Pattern #4: Innovating More Than Technology — The modular alternative isn’t a technical breakthrough. It’s a change in how the thing gets procured, specified and delivered. New ways of doing business, paired with technology, are what produce durable advantage.

• Core Concept #24: 10 Types of Innovation — Process and structure are innovation categories, not afterthoughts. Changing where the building is assembled is a process innovation with more community impact than most product decisions.

• Core Concept #51: Regenerative Design — Take-make-waste isn’t only a materials pattern. A construction method that consumes a neighborhood’s daily life for years and gives it back as a finished box is a line, not a loop.

Sources

• Regenerous Labs, Modular / Manufactured Data Center Provider Landscape, August 2026. [PLACEHOLDER-URL]

• Kittelson & Associates, Tower Road Traffic Volume Forecast, Technical Memorandum to Morrow County Public Works, August 7, 2023 (Exhibit 81, Morrow County Planning Commission).

• Kittelson & Associates, “Why Data Centers Don’t Fit Standard Traffic Assumptions,” April 8, 2026.

• Matthew S. Smith, “The AI Data Centers That Fit on a Truck,” IEEE Spectrum, March 30, 2026.

• Microsoft Local, “How businesses, educators, and communities found common ground in a tiny forest.”

• Microsoft Local, “Microsoft bouwt aan toekomstbestendige datacentercampus in Middenmeer,” March 2026.

• NL Times, “Microsoft data center uses 1% of all Dutch electricity,” July 14, 2026.

• Data Center Dynamics, “Microsoft acquires 50 hectares for data center expansion in the Netherlands,” June 2026.

This piece was drafted with AI assistance. The argument, the reporting decisions, and the final words are mine. Say it ugly. Build it better. Onward!

That was the list in 2019. In comment periods, county hearings, headlines in papers. The same five things from communities in different corners of the world. Singapore, the Netherlands, New Zealand, Italy. Even a few towns in the US.

Subscribe now

Notice what isn’t on the list. No carbon. No climate. No politics. Every item is local, physical, and sensory: something you can smell at the fence line, hear from a porch, see against a ridge, or fail to find in a well.

People weren’t objecting to an idea. They were itemizing damages.

For two decades the industry had an answer for this, and it worked. Tax revenue. Some jobs. Free wi-fi. A replenishment project in the region. Permission had a market rate, and companies paid it.

Then the price stopped being the point. In 2019 Amsterdam and Singapore both stopped issuing permits outright. And cloud demand was already on course to multiply the footprint tenfold. AI hadn’t arrived yet.

So we designed for yes, please.

Not for tolerance. Not for a narrow approval on a split vote. We set out to build something a community would want on its land whether or not anyone needed a data center: waste heat piped to greenhouses instead of exhausted through noisy fans, water returned cleaner than it arrived, no diesel, a site that produced more than it consumed. We designed out NIMBY to create PIMBY — Please in my backyard.

It worked the way we hoped. Show people the designs, walk them through what the building actually does for the place it sits in, and the room changes. For the communities, it changes from resignation or resentment to relief and reception.

And it changes for the builders too.

Veterans of this industry — people who had spent twenty and thirty years building the big windowless boxes — reviewed our plans and told us they would be proud to build it. Not interested. Not supportive. Proud.

Angus Fletcher writes about emotions as information, feedback. Fear tells you that you have no plan. Anger tells you that you’re stuck with only one. You don’t argue your way out of anger — you make a second plan.

Data center designers and engineers have had essentially one plan for a generation. A plan that put them in conflict with almost every town they entered. And every ounce of ingenuity in a brilliant, deeply technical field went into making that one plan marginally cheaper and faster.

The regenerative design was the first real new plan. The pride wasn’t a mood produced by pretty pictures. It was the specific relief of finally having somewhere else to go. To have a vision of the future they could be proud of with others, not despite others.

Communities had one plan too. Fight, or lose. Refusal was the only instrument they had.

The Regenerative Datacenter can be the industry’s “Joy Pivot” with communities, and the communities’ Joy Pivot with the industry. We need it to be both.

Pressure without pride doesn’t produce better buildings. It produces the same building in a friendlier county. Communities have pushed for twenty years and mostly what moved was the map.

Pride without pressure doesn’t produce buildings at all. I can tell you exactly how that goes. Hundreds of people who wanted to build something, and one no at the top.

Piece by piece, our design is making its way into the specifications at Microsoft and, more so, in new companies. Good work finds a way forward. Push and pull — that’s the balance that works.

Leave a comment

This is #30 in my weekly Innovating Out Loud series. AI was used as a research, writing, and arguing partner. All the ideas and final edits are mine. Say it Ugly, Build it Better. Onward!

That memory floated back to the surface this weekend when I picked up my 16-yr-old niece from the airport. She flew across the country on her own and navigated the trip, with connecting flights, beautifully — right up until she took home someone else’s bag.

That’s how the useful stuff actually gets built. Not from the average day. From the exception — the one that marks you. I will forever bag my toiletries even on road trips. She will forevermore triple check her luggage tag.

And that’s the next topic of discussion on Innovating Out Loud this Thursday, July 23rd with our guest, Angus Fletcher.

Subscribe now

Angus is a story scientist at Ohio State’s Project Narrative who has spent years studying this — and who started out studying brain cells in a lab. He argues human intelligence was never really logic. It’s four older powers, and every one runs on the exceptional, not the average:

• Intuition — the one strange detail everyone else walked past.

• Imagination — running tomorrow before it happens.

• Emotion — the read on whether your plan is any good, before you could say why.

• Commonsense — taking the next step when the map runs out.

Underneath all four runs storythinking: we don’t hold the world as data, we hold it as cause and effect — this, then that, then the thing after. It’s why one shampoo coated backpack teaches more than a hundred sensible warnings.

And none of it is fixed. The Army grew these powers in its Special Operators; the same drills later moved doctors, salespeople, rooms full of eight-year-olds. You build them by taking back what school trained out of you — chasing the odd detail instead of the average, letting the feeling in the room mean something, stretching past the edge of what you actually know, or not.

Now think about the thing most of us are starting to hand our judgement to. It “knows” more than any human who ever lived — and has been marked by none of it.

It’s a reflection on the water: the whole forest, every branch in place, rooted in nothing. It runs on the average of nearly everything ever written. It sounds intuitive but notices nothing exceptional. It sounds wise but doesn’t know when it doesn’t know. It sounds moved but has nothing at stake. Ask it the same thing a thousand times and it answers with the same even confidence — right or wrong, it can’t feel the difference — and it walks away from every answer exactly as it walked in.

At sixteen, my niece is just starting to collect the exceptions that will make her. At fifty, I’ve got a backpack full. The machine will still be averaging when we’re both long done growing.

So the question was never human vs. machine. It’s which work is whose — and what we hand away when we automate the parts that only a life can grow.

Join Angus and me this Thursday to continue the conversation and learn more about how he trains everyone from 8-yr-olds to Special Forces on developing the four pillars of Primal Intelligence. Register here.

Leave a comment

Keeping it short this weekend - birthday pass 🥳 Onward!

You think you have plenty of time. And then all of a sudden, the trail is grey, then the trail is just a suggestion, and then you are standing in a place you can no longer see. You reach for your flashlight or headlamp.

This is the moment.

Your thumb is on the switch. You can end this whole situation in one second.

Turn it on and ten feet of path reappears ahead of you. Past the beam and all around you is still a black wall — six inches beyond the light it seems even darker now. The forest that was around you a moment ago is receding. Something breaks low and fast off to your right. You can’t see it. And you won’t.

But what if you wait? What if you stand there in the darkness?

Subscribe now

The first few minutes aren’t great, for me at least. My heart races and I imagine all the things that are lurking in the shadows. Stalking me. I feel like I’m doing something stupid. My instincts argue with my choice.

Then a trunk emerges from the dark — the white birch always show up first. Then a few feet of the trail. Not brightly, but there, a pale rift in the darkness. Then the space between the trees gains depth. A smear of foxfire on a rotting log. A lightning bug or two (fireflies to anyone not from Jersey). Looking up, I can see a canopy that is not black but very dark blue, with holes in it. Some stars shine through.

I wouldn’t have noticed the foxfire in daylight. Lightning bugs don’t come out until dark. The forest has a light of its own that illuminates far more than my flashlight. I just have to wait for my eyes to adjust.

Great Question

In the June episode of the Innovating Out Loud webcast, Larry Robertson and I talked about questions — the subject of his new book, Great Question: The Art of the Ask and Getting More of What You Really Want. Forty minutes into our effort to question questions, Larry asked me a doozy.

What do you already know that you will realize soon?

I have made a living out of asking questions. I train people how to do it. This question did something only a handful of questions have done throughout my life — it hit my mental pause button. My mind went blank.

Except that’s not right, and I got this wrong the first three times I tried to write about it. My mind wasn’t blank. It was dark. There was a there there — you can hear my oooooo reaction in the clip — I just couldn’t see it. Still can’t.

But as I sat with it, my mind became a room with the light off. Dark, but it had depth.

And then it wasn’t a room anymore. It was the forest.

On or Off?

I have been sitting with this question for nearly three weeks. I still can’t see very far down the trail but things are starting to emerge. My mind’s eye is starting to adjust. What was at first a surprise — the sudden darkness and wanting to find an answer — has eased into a calm abiding. It’s actually become rather nice. I’m enjoying the not knowing.

I feel pretty sure that nothing is going to eat me if I just stand still for a little while. And I believe I will see some things that would have otherwise scattered.

And I think that is the point of the question.

It would be easy enough to ask AI to help me find an answer faster. I could pull together a panel of my experts to generate ideas, hundreds of them in minutes. But that would be like turning on the flashlight in the forest. Sure, I’ll see the trail. The worn path. But everything else will disappear into the shadows, maybe never to be seen again.

My refusal to use AI in this case is harder than it sounds, because I’m an AI champion. I use it constantly and find it astonishing. But there’s a distinction in how I use it that I couldn’t have articulated a month ago.

1. Expansion by exposure — transcripts, research, arguments, other people’s thinking, ten times more of the forest than my legs could ever carry me across. That’s exploration.

2. Expansion by generation — hundreds of options, on demand, right now. That’s a flashlight pointed at my own uncertainty, and what it returns is the average of everything ever said about my problem. It’s the worn path with the rest of the forest gone black behind it.

The average is in the beam. The extraordinary is in the dark, running away from the light.

You and Only You

Thought experiments have produced some of the largest breakthroughs in human history. Galileo mentally dropping rocks off the Tower of Pisa dismantled Aristotle’s theory of gravity. Einstein chased a beam of light in his head. No data. No lab. Just a person sitting with a question, adjusting their mind to it.

That is the dark adaptation of innovation — of creativity. It is the oldest and most powerful cognitive practice we have, and it works precisely because we don’t turn the light on. The answer was already there. What was missing was someone sitting with the question long enough to perceive it.

More specifically, what is missing is you. Perception cannot be delegated. Not because the other people or machines are stupid but because we do not have your perception. Your senses. Your collection of experiences. Someone didn’t come up with the theory of special relativity. Einstein did.

So the Monday morning practice is not one of patience or even courage. It’s trust. Trust in yourself. Trust that your eyes will adjust. That your mind can work its way through a darkened path.

There is a time and a place for artificial illumination. There are trails you actually need to see, deadlines you need to meet, and questions that already have answers somebody else found that are worth knowing. For those, use the light. I do, every day.

But the truly novel answers, your truly novel answers, can only be found in the deep forest of your mind. And they may only emerge if you stand still for a little while and let them.

Leave a comment

I could not ask for a better next guest to continue this conversation: join me on July 23rd at 10:30 AM PT / 1:30 PM ET with Angus Fletcher. He’s the author of Wonderworks and Primal Intelligence — books that have come up in IOL conversations frequently — a story scientist at Project Narrative, and professor at Ohio State University. Angus argues that the human mind starts with story. A process he calls Narrative Cognition. By the time we chat, I will have spent a month sitting with a question that has become a dark forest. I THINK he’ll tell me that’s exactly how it’s supposed to work.

Connections to The Insider’s Guide to Innovation at Microsoft

• Diverge-Converge-Synthesize As we say in the book, the best innovators diverge more often, and for longer, than everyone else — resisting the urge to converge early for the sake of order, control, or security. Because just like the flashlight in the forest, early convergence returns the middle of what's known, and misses the extraordinary.

• Intentional Discovery Discovery doesn’t fit between meetings. That’s the finding from the book, and it’s the one people most often nod at and then ignore — because a calendar full of open space looks, to almost everyone, like a calendar full of nothing. Three weeks of not answering a question looks like nothing too. It isn’t.

• Think Week Twice a year, Bill Gates would disappear to a remote cabin with a stack of papers and no company. The value was never in the cabin — it was in the removal of input long enough for something to surface. You can now recreate the cabin and fill it with a machine that will hand you three hundred ideas before lunch. That is not a Think Week. That’s a very expensive flashlight.

• Questionstorm The technique in the book is to generate questions instead of solutions — who, what, when, where, how often — and hold off on answering. We use it to help open up the problem space, and by extension, the solution space.

This piece was written with AI assistance. I use it to research, to think out loud, and to argue with. The judgment, the errors, and the sitting are mine. Onward!

Larry’s an innovation advisor and the author of Great Question, a book he built by talking to more than 140 people — neuroscientists, teachers, artists, CEOs. What he found is that we’re a questioning species. Our brains actually do their best work in uncertainty, not away from it. We got into why “best” isn’t a fixed answer but something you build together, why asking “why” too early kills the good stuff, and how I once flipped a data center project by asking “What if communities asked us to build one?” instead of “How do we get permission?” We closed on AI — useful, but only if you’ve already got the muscle it can’t teach you. An hour in, our assumption held. Questions really are where it starts.

Subscribe now

Five things worth carrying with you

1. Questions aren’t a technique. They’re how we’re built. Your brain does its sharpest thinking in change, complexity, and uncertainty — Larry’s neuroscientists call it “innovation mode.” So questioning isn’t something you bolt onto the work. It’s the work.

2. Quit leading with “why.” Start with “why” and you’re down a rabbit hole you won’t climb out of. Try Deb Meyer’s five instead: How do we know what we know? Is there a pattern? What if? Is there another way? Who cares? As Larry put it, a question is just going there. You’re not committing to anything yet.

3. Flip the question and watch the whole thing change. For years the industry asked, “How do we get permitted to build a data center?” We asked, “What if people were asking us to build one?” You can’t solve that the same way. What came out didn’t look, sound, or run like a data center at all. Because innovation is value creation — if it’s not changing someone’s life, it’s just a cool idea.

4. It starts with trusting yourself. Every field Larry studied circled back to the same thing: faith in your own questioning, earned from the wins and the flops. You can’t build that trust across a team if it isn’t there in you first. When you’re stuck, lean on the three things you can always trust — curiosity, play, and search.

5. AI can’t teach you to question. It can help. But AI averages toward an answer, and you can’t amplify a skill you don’t have. Build the muscle yourself, and how you use AI will change. Skip that, and you’re just handing over the thinking that would’ve made you better.

You can find Larry’s book, Great Question: The Art of the Ask and Getting More of What Your Really Want everywhere books are sold.

Next up on Innovating Out Loud: Angus Fletcher. Story Scientist and author of books that have come up in many previous episodes. Wonderworks and Primal Intelligence live on my desk. This is a not to miss session. NOTE the change of time: 10:30 AM PT / 1:30 PM ET. July 23, 2026. Register here and see you then!

Leave a comment

The Menominee retained this land in 1854 — a fraction of their homeland — and logged it from the start under one rule: don’t cut more than the forest can regrow.

They fought the agent’s refusal to Congress and, in 1908, won more than a sawmill. Their rule went in with the victory: federal law now required their forest to be managed on a sustained yield. Fifty years of practice became statute.

Today, after a hundred and seventy years of continuous harvest, the forest holds more standing timber than when they started. All around the Menominee the Great Lakes pine was falling in one of the fastest liquidations in history. You can see the difference in satellite images: a dense green rectangle with the razor-straight edges of the reservation, ringed by land that was cut once and never came back.

Subscribe now

Another Map

Washington, D.C., 1935. Government men sat with city maps and colored the neighborhoods — green for “best,” red for “hazardous.” The red tracked one thing with uncanny fidelity: where Black families lived. It was an innovation of its kind — however ignorant and cruel. A technical system for sorting mortgage risk, adopted nationwide.

The men who drew the lines died decades ago. Their lines lived on.

Mortgage capital followed them for generations. Home equity — the main way American families build wealth — compounded in the green and stalled in the red. The gap those men drew is still measurable today, in neighborhoods they never visited, in families that never heard their names.

A line drawn around a forest in 1854, lines drawn around neighborhoods in 1935 — two maps. Both still visible, one from orbit, one in the wealth data. Same mechanism, opposite effect.

The Test

America turns 250 this weekend, and innovation retrospectives will hand the credit to technologies like the assembly line and the transistor. I wanted to sort 250 years differently — into what was encoded and endured, regardless of intent. A generous billionaire doesn’t pass the test; bylaws do. A bigoted clerk doesn’t fail it; maps do. I discovered a wide range of examples.

Philadelphia, 1787. A few blocks from where the Constitution was being drafted, two Black preachers — Richard Allen and Absalom Jones — founded the Free African Society. Shut out of the new nation’s protections, they wrote bylaws of their own — dues pooled for the sick, mutual aid made structural. The model spread throughout Black America for two centuries.

Ohio, 1921. Thomas Midgley Jr. and Charles Kettering successfully tested lead, already known to be toxic, as a gasoline additive. GM and Standard Oil pushed it into the market, where it poisoned three generations before being banned.

Alabama, around 1900. Cotton had done to Southern soil what any monoculture does: mined it. Then the boll weevil arrived and took the cotton too. Enter George Washington Carver, who demonstrated how crop rotations restored fallow land, feeding the fields and the farmer. His practices are still outperforming the alternative, 125 years after he made them free to every farmer who could read.

The test is a permanence detector, not a virtue detector. Whatever we write into structure, mutual benefit or selective harm, we get for generations. The mutual-benefit lineage seems to share an author pattern though: people the system locked out. For them, reciprocity wasn’t a preference. It was the only architecture that would hold.

The Opportunity

Reconfiguration follows ruptures. And ruptures show up as cracks first. One of today’s largest cracks is in the news daily. A data center arrives in a town with plenty of cheap land and a few thousand people. Jobs are promised, tax breaks are signed. Within a few years the people pay more for power, the aquifer runs low, and the profit is booked somewhere the town will never see.

Now look around — the shape abounds.

Gig workers with no safety net. Whole regions insurers quietly decline to cover from catastrophic weather events. A generation of anxious people. Each one is the product of technologies sold to us the way leaded gasoline was — as progress, with the costs kept off the books.

Extraction is meeting its limits everywhere though. In the soil, in the water, in the markets that price risk. The Menominee and their forest show us what’s possible. An alternative, a structure, that can hold across generations despite many powerful attempts to break it (a history worth exploring).

Their story shows how structure reinforces structure: every encoding makes the next erasure harder.

The Menominee also show us that mutually-beneficial repairs, regenerative repairs, start with the people who would otherwise pay the costs. For data centers that means it starts with the people in the towns requiring peace and quiet and nature preserved, energy and wealth shared, and culture respected.

The history is clear: either the towns create the structures that require what they want or the men in faraway rooms with maps and red markers will. Either way, what we encode now will endure for decades or even centuries. For better or worse.

Leave a comment

Connections to The Insider’s Guide to Innovation at Microsoft

• Adaptive Cycle of Innovation (#22). The book borrows nature’s cycle — seedling, forest, disruption, regeneration — and names the rigidity trap: systems too fixed to survive the lightning strike. The Menominee forest is that cycle done right, and reconfiguration after rupture is the spine of this piece.

• Biomimicry / Systems Thinking (#77). Nature is the original continuous innovator; the resilient move is to build so that when disruption comes, the next growth is already underway. A forest holding more timber after 170 years of cutting is that principle made literal.

• Innovate Upstream to Succeed Downstream (#74). What we encode early decides what’s possible later. Redlining and sustained yield are both upstream choices still running downstream, generations on — which is exactly why what we sign now matters.

Written with a team of custom AI assistants as part of my Innovating Out Loud practice. The ideas, mistakes, and final words are mine. Say it Ugly, Build it Better. Onward!

You don’t need anyone to tell you innovation (any change) is hard. You need something more specific and far more useful: which thing, on this idea, is the one dragging it down. The fix for a story that never landed is nothing like the fix for an organization quietly rejecting the change — and treating one like the other is how a good idea gets dusty.

Last week at Innov8rs in Toronto, Taryn and thirty people used the “fast read” of Innovation Coach 2.0 to find answers in 90s. It read what each person wrote, named the most likely reason their idea had stalled, and handed them a first move. It worked. People read their results and nodded.

But agreement with the diagnosis is not the same as “okay, let me go do that.” A few people said a version of: it can’t be that simple. Too easy. Too fast.

That reaction is the whole story.

Subscribe now

The tool wasn’t wrong. It was a little too good at the part that’s easy to speed up. AI can take you from stuck to a first move in ninety seconds — that’s real, and it’s new. But getting the answer and believing the answer are two different events, and only one of them got faster.

In Pattern #3 of The Insider’s Guide to Innovation at Microsoft, we shared the B2Me journey — White Rhino’s map of how a person moves from unaware of a new idea to acting on it. For big changes and small, the pattern is clear. We don’t leap, we travel. From unaware to interest, then consideration, then decision, oscillating between what you feel and what you think the whole way. Emotion moves first; the reasoning catches up.

We usually talk about B2Me as how you bring other people along. Toronto turned it inward. Hand your own stuck idea to a tool, let it give you a verdict, and you are your own first stakeholder. You have to be brought along to believe your own diagnosis.

Skip the travel and you get results without roots — the output without the understanding that lets you trust it, carry it, or build on it. The answer was true. It just hadn’t been earned, so it didn’t hold.

Speed isn’t the problem, though. For someone who’s already walked the distance — circled this idea for months, just needs it named — ninety seconds is a gift. The trouble starts when you mistake fast for done. A quick, true answer that closes the question is worse than no answer, because now you’ll act on a shallow read you trust too much.

So we changed the tool. The fast read stays — same ninety seconds, same families of cause, same first moves. But it’s no longer the destination. It’s the on-ramp. Past it, you can dig deeper: the engine behind our original Innovation Coach now sits underneath the diagnostic and walks you the rest of the way. It asks, you answer, it sharpens, and it keeps going until what you’re holding isn’t a label but a plan you could run Monday morning.

Fast read to feel it. Deeper read to earn it.

One move matters now. You’re probably already holding a stuck idea — this piece may have surfaced one. Don’t let it slide back under. Name it while you can still feel it. Take the ninety-second read, then keep going: dig deeper, and walk it down to one thing you’ll do Monday morning. The feeling you have right now is the fuel. Use it before the scroll carries you off.

Try it here: Innovation Coach 2.0

In a very meta moment, I asked the coach to give me a plan to increase return users to the the tool.

So - if you want to know when V3 arrives, comment “Let me know” and I’ll add you to the list.

Leave a comment

From the book

• B2Me Journey — behavior change as a journey, steps not leaps; emotion before cognition. The updated tool respects the journey instead of skipping it.

• Do The Work — the knowing–doing gap. The deeper read is where knowing turns into doing.

• Minimize Uncertainty (a Trust Lever) — concrete next steps cut ambiguity and move you along the journey, instead of leaving you with a verdict and no path.

• Developing in the Open — the VS Code posture. We show you the tool and what we changed about it, not just describe it.

Developed in the open, with AI as a thinking and drafting partner. The argument, the judgment, and the final words are mine. Say it ugly, Build it better. Onward!

The beaches of Washington state — where I’m currently retreating with two of my Lab partners — have me thinking about one of my hypotheses: Everything is Loopy.

The waves, the seashells, the driftwood the size of full trees.

Even rocks loop. A cliff face wears down to sediment, the sediment packs into stone, the stone rides a plate into the heat, melts, and rises again as something new. A few hundred million years a turn — but it turns.

Nothing in nature travels in a straight line. And nothing in nature wastes. Every output is an input. The fallen leaf is the soil’s lunch.

Nature loops, and we draw lines.

A line has an open end, and value pours out the whole length, not just the tip. A line is a loop somebody cut open to take from. That’s the problem with non-regenerative business in one breath.

The spiral economy

Circularity is a loop that returns to where it started. But only from one perspective. What looks like a circle from above is a spiral from the side. Circularity across time.

Therefore a regenerative loop is a spiral. A regenerative system is a network of spirals, amplifying and balancing. Onward, not upward or downward — what does up or down mean in time anyway?

Yet, the predominant mental model for a thriving business today is a line that goes “up and to the right” over time. It’s as flat as a circle but open-ended. Which is why lines leak value. They waste.

Subscribe now

How do we stop the leaks? We just said, essentially, we can’t actually close a loop. It doesn’t return to the same spot. It spirals onward. So a regenerative business needs a different action. Don’t close it. Bend it.

This hypothetical spiral economy has another name: the fifth industrial revolution. At the 2019 World Economic Forum, IR5 wasn’t fully defined, but the vision for it was described in the best way I’ve found so far: the economic future when “technology is bent back to the service of [people and planet].”

Bending is the work. A line doesn’t curve itself. Up and to the right is business in two-dimensions — what technology achieves when no one decides where it goes over time. Bending it takes craft and intention.

Onward

Everything that lasts loops. It spirals onward, intersecting with other spirals, changing from the meeting. The question we’re asking in the Lab during this retreat — that might be worth asking of your company or your own week’s work — is whether we are building lines or loops.

And if we find lines, how do we bend them?*

Written with AI as a thinking and drafting partner. The ideas, the edits, and the final words are mine. Say it Ugly, Build it Better. Onward!

*This week’s thoughts may also be influenced by the World Cup.

JoAnn and Taryn spoke with Jayshree on the May 28th episode of IOL. Here are a few key takeaways.

Five takeaways

1. AI can accelerate the climb from data to wisdom. It can’t make the climb for you. Jayshree maps the whole journey onto data, information, knowledge, wisdom — and gives each rung a framework. TRACE your data. FAST track your information – separating facts from assumptions, scenarios from trends. Build knowledge across four ARCS, because no function is a monolith and one-size-fits-all becomes one-size-fits-none. Then go with the FLOW for wisdom: flexible not fixed, winding not vertical. Most AI failures don’t start with the model. They start with untraced data going in.

2. AI flips the pyramid — and DIKW becomes WIKD. Normally you earn your way up from data to wisdom. AI inverts it: it hands you a confident conclusion instantly, so you start at the answer before you’ve understood the question. Flip the pyramid and DIKW reads backward as WIKD — wicked. Not wicked-bad. Wicked as in the engineering sense: the solution changes the problem, there’s no single right answer, every attempt reveals new complexity. AI was supposed to solve wicked problems. In the wrong hands, it manufactures them.

3. It’s not artificial intelligence. It’s “artificial diligence.” Calling it intelligence is off-putting and is inaccurate. It’s diligent: it’ll do what you ask without complaining. Strong at breadth, brittle at precision. And in R&D, a confident wrong answer isn’t better than no answer — it’s worse. So the demand shifts from “show me the money” to “show me the source.”

4. AI narrows the access gap and widens the judgment gap. A business leader can now interrogate any technical concept, generate hypotheses, pull a patent landscape — no scientist in the room. The gap in access shrinks. But the gap in judgment grows, because everyone’s working from the same AI-generated synthesis. What separates a good decision from a bad one is whether you have the depth to know what the AI got wrong.

5. AI doesn’t just change jobs — it changes the jobs to be done. The work itself is being redefined, which makes adoption a team development challenge, not a technology deployment one. People don’t resist good tools. They resist disruption to identity, process, and power, all at once. That’s why the real work lives in change enablement, not the tech.

Closing

The technology is ready. The organization usually isn’t — and that gap is human. Jayshree’s whole toolkit points at the same quiet truth: AI is most useful to the people who’d still know what to do without it. Watch the full conversation for all four frameworks, the books behind them, and why she thinks the unlikely candidates are the ones true innovation needs.

Check out Jayshree’s books, The Heart of Science Series: Engineering Blueprint (2024); Engineering Fine Print (2022); Engineering Footprints, Fingerprints, & Imprints (2020). All sales proceeds go towards a scholarship for women in STEM administered by Society of Women Engineers (SWE).

Our next session of Innovating Out Loud is on June 25th at 9 AM PT / 12 PM ET with special guest, Larry Robertson. He’s an author, advisor, innovator, speaker, columnists and collaborative partner of the Lab! He and JoAnn will be talking about his new book, Great Question: The Art of the Ask and Getting More of What You Really Want.

Never miss a live episode. Register for the series at: www.regenerouslabs.com/innovatingoutloud

A thread through them: each is about the easy thing to do with a question — ask the obvious one, skip the inconvenient one, collapse the messy one to a single number — and what the easy move quietly costs.

1. The question we wouldn’t let them ask

This week I ran my first session with Duke Corporate Education — thirty senior leaders from a big tech company, working on the hardest part of strategy. Not setting it. Executing it, when the destination is fixed and the path keeps moving under you.

To better understand the Why of each strategy, we did a question storm. Teams of six, fifteen minutes, one rule: you may ask anything except why. Who, what, when, where, how, how often, if true, then what — all fair game. But why itself was off the table.

Round one filled the flip charts with the usual questions. In round two we stretched for weirder. The weird questions are the point — they send your brain down paths it would never have walked.

Forbidding the easy question is what produced the range. They understood their why more deeply precisely because they weren’t allowed to ask it head-on.

The constraint wasn’t a limit on the thinking. It was the method.

Subscribe now

2. Step 0

Most plans and assessments start with one question: what are you building? Output, product, roadmap, return.

In a regenerative process there’s a question that comes before it: what is it built on? The soil, the water, the stable climate — yes — and the web of relationships and infrastructure that everything else stands on. We treat that ground and our communities as free and infinite.

They aren’t. And once you make that Step 0 — the first thing you check, not the externality you discover later — the whole conversation changes. You can run it as a simple gate: is the ecosystem you’re building in regenerating, holding, or degrading? Green, amber, red.

Degradation at this step is a red light no amount of value somewhere else can turn green. And the gate has the most power before the deal is signed, while there’s still leverage — not in the report written afterward, explaining what got lost.

Step 0 asks What are you building? And what is it built on?

3. The number that makes people disappear

Every framework eventually meets the same demand: just give me one number. A single score, to rank and compare and decide.

I get the pull — a budget office can’t act on a paragraph. But there’s a trap I keep circling. Make the number too precise, and people game it: they optimize the metric and quietly lose the thing it was meant to measure. Make it too loose, and people and planet disappear — anything that doesn’t fit the formula stops counting.

Both failures come from the same wish: to collapse a living, many-sided thing into a single grade.

What if the honest move is to refuse? Not “no measurement” but a picture across dimensions that can’t be traded off against each other, plus the one or two places that genuinely are decision-ready. A portrait, not a grade. You give up the comfort of a ranking. You keep the people and places the ranking would have erased.

I’m working on whether a framework can hold that line without falling apart. More soon.

Leave a comment

Connections to The Insider’s Guide to Innovation at Microsoft

• Note 1 runs on Diverge-Converge-Synthesize and Intentional Discovery: the question storm is divergence with a rule that keeps it open instead of letting it snap shut. It also extends a thread we’ve worked before — “why” pulls you toward justifying and ruminating; “what / who / how” keeps you looking. Forbidding why is that insight, enforced.

• Note 2 is Aim for Positive and Top-Down, Bottom-Up, Outside-In: Step 0 brings the outside-in voice — the people and place — to the table before the design sets, and the gate asks for surplus, not merely less harm.

• Note 3 is Aim for Positive seen from the other side, plus Metrics: a multi-dimensional picture you can’t trade off, rather than a single grade. Signals over scores.

AI Disclosure: this piece was written with the help of my AI friends and editors. All ideas, choices, and final words are mine. Mistakes and all. Say it Ugly, Build it Better. Onward!

Your backyard is where this story really starts, but not how it was told this week at Microsoft Build.

Satya Nadella shared the great news during his keynote: the new AI data center design, Fairwater, would use about as much water in a year as one neighborhood restaurant. Fill the cooling loop once, recycle the same water continuously, barely a top-off needed from that point on. The room cheered. And they were right to — the water win is real.

The hum is real too, though. And it’s only going to grow worse as the site grows bigger.

Unfortunately, Microsoft closed one loop and left others wide open. That hum is heat being thrown away. The same heat blowing over your fence in Wisconsin is the kind Microsoft captures elsewhere and pipes straight into people’s homes. In Espoo, Finland, recovered data-center heat warms a quarter of a million people. So picture what leaves the building here, every night: efficiency Microsoft could have banked, warmth and quiet you’re owed, and energy made at real cost to the planet doing one job instead of two. Three kinds of value, drifting off into the cold.

Subscribe now

Quiet doesn’t scale. Reuse does.

Low-frequency noise is very good at moving around and through walls, and the only way to really kill it is to make less of it. In a data center, that leaves two levers: quieter fans and capturing the waste heat instead of exhausting it.

There’s a whole field dedicated to quieter fans, much of it borrowed straight from animals that have moved through air and water silently for millions of years — the serrated edge of an owl’s wing, the ribbed flipper of a humpback, the grooved skin of a shark. None of it is exotic. All of it sitting on the shelf today, cheap to design in before the steel goes up and a costly afterthought once it’s standing.

But quieter fans only help so much. Stack enough of them to cool 3 GWs of compute — Fairwater’s projected size — and the noise climbs right back.

Heat capture is the better first lever, because it does two things at once. It cuts the drone — less heat to shed, less air through fans, less noise made — and it creates value instead of destroying it. Closing the water loop makes closing the heat loop easier than ever. Water-to-water heat exchangers have been standard equipment since the early 1900s, ubiquitous in power generation, chemical processing, and modern district energy.

Microsoft already runs the largest data-center heat-recovery scheme in the world. The whole industry knows the move. This isn’t a technology challenge. It’s a business-model challenge.

Build with the place, not around it

The water achievement is something operators can claim alone, from a stage. Reusing the heat asks for more — a partner, a contract, a real conversation with the place you’ve landed in. Which is the same conversation neighbors keep saying never happened. Because it was optional, it waited. And collaboration that stays optional is collaboration that doesn’t happen.

No single user can absorb 3GW of heat, so reuse at this scale becomes a coalition — a paper mill, a year-round food line, a greenhouse sized to the stream. Done well, it reclaims a real share of the heat and takes that much load off the fans. Not silent and solved. But a stream of value sent back into the place instead of thrown over our fences.

Timing is the rest of the challenge. If a fraction of the load is already emptying backyards, the full build — more power, no heat capture easing the fans — carries further. Low-frequency sound is notoriously hard to block, so the ring of homes that can hear it will widen as the site does. The same math runs the other way too though. Every MW added is more heat worth reusing, and one more reason to design the loop now, while the site and the protests are relatively small.

External pressure creates internal permission

I’m going to say it until it gets boring — external pressure creates internal permission.

When Satya listed what a new data center build has to protect — electricity prices, jobs, the tax base — the list was real and worth honoring. But look at what it leaves off. The reality felt in our backyards. The heat loop hasn’t been closed in Wisconsin for exactly one reason: nothing and no one required it. I’ve asked people I know in the industry, and plenty of them want to build it.

Known solution. Real business case. Real people who’d say yes.

The water, the heat, the noise — three faces of the same thing, and not the only ones. There’s the light washing the stars from the sky, the years of heavy construction rolling through town, the diesel generators that outlast every climate pledge they contradict. Each one has options. Each is cheapest to address with the community and nature in the room on day one. That’s the whole of it. Bring every dimension to the table before the design sets, and build with the place instead of around it.

Cheer the step. Then ask for the rest.

The Mount Pleasant campus is mostly still unbuilt. The heat is right there. The fix that quiets the fans is the same one that reuses the heat. And the water already proved Microsoft can close a loop the moment they decide it matters.

So cheer the water. Then ask for the rest — the heat reused, the fans quieted, the neighbors counted in, not sacrificed. Not a different project. Not a trade-off. The same instinct carried all the way through, so the value keeps circulating through the system instead of leaking out of it. The company, the community, and the nature that supports us all will come out stronger.

The first step deserves a cheer. It just isn’t the place to stop.

Leave a comment

Connections to The Insider’s Guide to Innovation at Microsoft:

• Pattern #4: Innovating More Than Technology — The piece’s spine: “not a technology challenge, a business model challenge.” Every component already exists — heat exchangers since the early 1900s, quieter-fan designs sitting on the shelf. History is clear: the decisive innovation is rarely the device but the business model and value chain around it.

• Pattern #3: Innovating With Everyone — “Build with the place, not around it.” Bring community and nature to the table before the design sets. The Regenerative Datacenter team’s own precedent in the book — 200 conversations inside and outside before a single concept was drawn — is the proof that the early, optional conversation is the one that changes the outcome.

• Aim for Positive / Regenerative Design — Closing the loop to send a stream of value back into the place instead of throwing waste heat over the fence: surplus by design, not harm reduction. The book quotes Satya asking whether, at the core of the business model, “are you creating a surplus around you?” This piece asks him to finish the question.

• Top-down, Bottom-up, Outside-in — All three are in the room and none are wired together: top-down names what to protect (electricity, jobs, tax base); bottom-up is the engineers who already want to build the heat loop; outside-in is the backyard neighbors going unheard. Durable change needs the three connected — which is exactly what “external pressure creates internal permission” names: the outside-in voice unlocking what the bottom-up already wants to build.

This piece was developed in the open, with AI as a thinking and drafting partner. The argument, the judgment, and the final words are mine. Say it Ugly, Build it Better! Onward!

Subscribe now

Earlier this week, Pope Leo XIV called for all of humanity to respond to the start of the AI age just as Leo XIII had before him at the opening of the machine age. Our Leo reached past our windowless hall to an older building to make his appeal: a tower on a plain, raised by people who set out to make a name for themselves. People who believed they needed nothing and no one beyond their own kind and their own materials. That story ends the way pride stories always end. In ruin. The language breaks. The work scatters. The people walk away from each other.

Leo XIV told another story. A ruined city and a man named Nehemiah who refused to rebuild it on his own. Nehemiah called his neighbors together and asked each to rebuild a length of the fallen walls. He listened to their fears and learned what they needed. He matched the work to the people. He rebuilt the relationships before they rebuilt the walls. Together, they rebuilt the city. And because every family built a piece of it, they then cared for it together.

Between those two stories sits one idea: technology is never neutral. It takes on the character of whoever designs it, builds it, and runs it. The tower and the city are built with the same stone. What changes is who decided and who participated.

We’ve Been Here Before

Pride in the old story isn’t wealth or arrogance. It’s self-sufficiency — the conviction that you can raise the whole thing on your own terms, owing nothing, needing no one, pushing every cost outward onto the plain. The windowless hall runs on that same conviction. It’s designed to take what it needs from the place it stands in, and to give relatively nothing back. A building like that makes a name for some. But it doesn’t make neighbors.

We have seen this play out before. In 1891, Leo XIII watched factories swallow farmland and families whole. In his encyclical he wrote that workers carried a dignity the market kept forgetting.

We let the towers climb for another forty years until we fell into a depression and a war. Only out of that rubble did we build what Leo had asked for: the fair wage, the union, the weekend, the plain idea that an economy serves people not the other way around.

That is the part worth acknowledging at this moment. Last time it took a collapse to make us build together. Our response was late, partial, and reversible — we have spent the last few decades letting it slip away.

Today’s Towers of Babel

The strange windowless halls going up across the countryside faster than the towns around them can vote reflect the image of a few dozen people in the industry. They decide what data centers are or are not. People who are so far up the tower they can no longer see the ground clearly. They can’t hear the townspeople over the din of the diesel engines.

Being on the edges of the industry though, closer to the ground than the clouds and the gods, my team and I could see and hear our neighbors. Thirty of us, from seven companies, asked what Nehemiah might ask: what does a data center need to be to be welcomed by each place?

We asked that question of hundreds of people — and other life — all over the world. Then designed an answer.

It runs small and close instead of vast and remote. Its heat doesn’t vanish into the sky — it warms homes, feeds greenhouses, and helps the water of its place flow. Its power speaks to the grid instead of straining against it. The people who live beside it sit at the table before the siting is decided, not after the permits are filed. Each neighbor, old and new, builds a section of the wall. Not one of them can finish it alone.

That is what we call the Regenerative Datacenter of the Future. It’s the product of mutual benefit by design.

One Stone for Our Home

Leo wrote his letter to everyone — believer or not, the whole human family. He asks a human question under all the theology: will we build to make a name, or to make a home? I can’t answer that for the entire AI industry. I can only answer for my stretch of the wall. Energy, water, ecology and the concrete and steel halls we keep raising on top of them. For whatever reason, that’s where I landed, so that’s the section I’ll help build a better way — as though the dignity of all life were true, because it is.

The tower builders say we need them more than they need us. The city builders — community builders — say we need each other. History has shown us which one collapses under its own weight, and which holds a place and its people together.

I am just one stone holder, but I’ve got friends and neighbors. As do you. Together, maybe we can heed the call of this Leo, skip the depression and war this time, and lay the foundations of an age that is golden for us all.

Leave a comment

Connections to The Insider’s Guide to Innovation at Microsoft

This piece connects directly to four frameworks from the book:

• Innovating With Everyone — Nehemiah didn’t hold a stakeholder meeting — he called his neighbors together and gave each family a section to own. The seven-company coalition that designed the Regenerative Data Center operated on Pattern #3’s central premise: engage early, engage widely, engage with empathy. Meet people where they’re at and move forward together.

• Top-Down, Bottom-Up, Outside-In — The windowless hall is a pure top-down artifact — decisions made in rooms far from the plain, ignoring the ground-level feedback and community signals coming in. The Regenerative Data Center inverted this deliberately: designed from the bottom up, with hundreds of outside-in voices, human and otherwise.

• Aim for Positive — A data center that takes from the county’s water, pulls from the county’s grid, and returns a number on a tax form is aiming for less bad. The Regenerative Data Center aimed for abundance — heat returned to homes, power that steadies the grid, ecology restored, economy supported — mutual benefit by design, not compliance.

• Language as Strategic Tool — Leo XIV’s encyclical is a masterclass in strategic language: naming “Babel” and “Nehemiah” rewrites what builders optimize for before they pick up a stone. “Mutual benefit by design” and “each neighbor builds a section” work the same way — deliberate terminology that reframes what a coalition believes is possible, and then makes possible.

Sources

1. Leo XIII. “Rerum Novarum: Encyclical of Pope Leo XIII on Capital and Labor.” Vatican, May 15, 1891. https://www.vatican.va/content/leo-xiii/en/encyclicals/documents/hf_l-xiii_enc_15051891_rerum-novarum.html

2. Leo XIV. “Magnifica Humanitas.” Vatican, May 15, 2026. https://www.vatican.va/content/leo-xiv/en/encyclicals/documents/20260515-magnifica-humanitas.html

Written with AI assistance. The thinking — and the stretch of wall — are mine.

That December, the head of data center design heard through the grapevine that I had been asking questions (apparently forgetting that he was one of the first people we interviewed). Specifically, he heard a misremembered account that I had suggested Microsoft’s design was no different than Amazon’s. I hadn’t. But the question itself was a big problem to him.

He spent the next two weeks setting up meetings to discuss my behavior. Two weeks, during the first Christmas of Covid. He brought one-line diagrams to set me straight. He shut me down in meetings. And he made it clear that he did not believe I was qualified to question the design. His team’s design. He didn’t see asking questions as part of my job. He saw it as a problem to be managed. He saw it as my arrogance, hubris, and insubordination.

Like I said. People really hate intentional discovery.

Subscribe now

While that was going down, the team pressed on. By February we had the vision, the guiding principles, the first video in a Why–What–How series, and a roadmap to 2025. We had CVP approval for a demo project that solved an urgent, current-state problem — permits being blocked in the Netherlands. That demo project got funded, went into action, and is still highlighted in annual reports today. The Why video eventually became the org’s public sustainability video.

The February timing was not an accident. The next fiscal year’s budgeting started in March. I had built the rhythm of the program around it. Every February, we had tangible results to put in front of the people deciding what to invest in next. Discovery never stopped. It just did not have to look like discovery to the people we needed something from. That was my job as the orchestrator: to keep the discovery open while feeding the organization on its own terms.

With the Netherlands win under our belts, we went back to that same DC design leader with the Why video.

He teared up. He said, “This is the future. We’ll get you the money.”

The same person who had spent December trying to censure me for asking questions teared up in February at the result of those questions.

He did’t hate the work, exactly. He hated the visibility of the work — the part where it looked like nothing was getting built, where the team was talking to too many people, where the answer was not yet defensible in one-line-diagram form. Where it looked like his team’s work was being second-guessed.

By the time our work converged, he loved it. He just didn’t want to see how it was made.

The dominant pattern in this AI moment is the opposite of what we did. Give AI your question, AI will give you an answer. You receive the output without building the understanding. You cannot reliably build on top of it, because there is nothing underneath. You have a result. But discovery’s most valuable outputs aren’t results. They are often better questions. Shared vocabulary. Reduced scope. Architectural clarity. Trust between collaborators who have looked at the same hard thing together.

None of those show up on a status report. All of them determine whether what gets built will hold. The data center work — five months of intentional discovery in 2020 — produced a solution that is more viable in 2026, more feasible and more desirable, even through the disruption of AI’s emergence.

That is why we “say it ugly” — so that we can “build it better.” It’s the discipline the lab is built on. Intentional discovery is not a phase we move through. It is a mode we hold open — in parallel with design and development, narrowing as the work hardens, never fully closing while something genuinely new is being built.

Innovation is not only loopy. It’s layered.

The people who have worked with me inside that mode have told me, more than once, that it was the hardest and most rewarding work of their careers. Four of my DOTF teammates are now my partners in the Lab. Many of our coalition colleagues are frequent collaborators with the Lab.

Discovery is where dormancy meets expansion.

Most organizations have engineered out both. When you intentionally make space for both, the work and the people change together. Roots form. They hold up everything built on top of them, and compound over time.

The Regenerative Datacenter of the Future work my team and I did between 2020 and 2023 is some of the most exciting, interesting, and important work I’ve done so far. But then, so was the work I did before that. I don’t know where intentional discovery will take us this time, but I’ve been through enough loops to know it’ll be amazing.

Onward!

Leave a comment

Don’t miss this week’s live Innovating Out Loud webcast with special guest Jayshree Seth — Corporate Scientist and Chief Science Advocate of 3M. She’ll be sharing her experience and frameworks bridging the gaps between research and practice, business and technology, and data and wisdom.

Thursday, May 28th at 9 AM PT / 12 PM ET.

Connections to The Insider's Guide to Innovation at Microsoft

Intentional Discovery: purposefully exploring with the goal of ultimately having an impact on the world, resides in Pasteur’s quadrant (of the Stokes framework by the same name). The 200 interviews, the catalog of parts, the journey map from dirt to data — that is intentional discovery in practice.

Because intentional discovery is a foundational practice of repeat innovation, it connects to many other frameworks in the book. Here’s a short list:

• Pasteur’s Quadrant is the motivational state operating underneath the work. The DOTF team was not doing pure research (Bohr’s quadrant — no practical application in mind) and not doing applied research (Edison’s quadrant — scaling a pre-specified problem with a known solution). They were holding both simultaneously: genuine curiosity about what the data center of the future could be, driven by clear awareness that it had to work in the real world at scale. That specific orientation — use-inspired basic research — is what made the work generative rather than just exploratory or just incremental.

• The Pasteur-Pisano Innovation Configuration — my framework combining the work of Stokes and Pisano — explains why the solution is still more viable in 2026 than it was in 2020. The DOTF team explored both the technical dimension (the physical and systems design of a regenerative data center) and the business model dimension (the stakeholders told them early that the existing business structure would not support the future). Exploring both dimensions places the work in the architectural quadrant of the Innovation Configuration — the highest-complexity, highest-value space. Most teams explore only one dimension, which is why they produce incremental results even when their technical work is strong.

• Discover-Design-Develop — the three-phase innovation process — is the structural spine of the DOTF program. But the piece makes explicit something the book describes and the DOTF proved: the phases are not sequential. Discovery did not end when Design began. A subset of the team continued exploring the business model dimension while the physical design was being developed. The work was loopy and layered — both qualities named in the book, and the second word extended by the piece’s new aphorism: Innovation is not only loopy. It’s layered.

• Diverge-Converge-Synthesize is the internal discipline of each phase, and the December 2020 story is a case study in what happens when someone outside the process mistakes divergence for incompetence. The head of data center design could not see the value of the asking phase because he was waiting for the convergence phase. The book is direct on this: “Often, though, out of a conditioned need for order, control, simplicity, or a sense of security, we converge far too early.” His two weeks of meetings to discuss JoAnn’s behavior were, in effect, an institutional demand to converge before the work was ready.

• Pattern #1: Innovating Every Day — Step back and complete Discovery before moving to Design and Develop — is the operating instruction the DOTF team followed and the DC design leader tried to prevent. The February rhythm — tangible results timed to the budget cycle — is how the orchestrator of discovery makes Pattern #1 survivable inside a large organization. Discovery needs a strategy for persisting. The February rhythm was that strategy.

• Fit People with Phase appears in one sentence of the piece: “We had identified the people on the team who were constitutively suited for discovery — the ones who could keep asking questions when the room wanted to move on.” That is the Pioneers category in the Pioneers-Settlers-Town Planners-Boundary Crossers taxonomy. Knowing which people belong in which phase, and making those assignments deliberately, is one of the practices the book identifies as common to the most successful innovation teams.

• BXT / Lead with X is the invisible architecture of the coalition. The head of data center design was leading with T — his team had the technical design and defended it. The DOTF team led with X — they started with the people (200 interviews, personas, journey maps) and built the technical and business case inward from that understanding. By February, the Why video was X-led communication: it did not argue the technical case first. It made the case for why the future mattered to people. The tears came after the X, not after the diagrams.

• Say it Ugly is named in the piece and is the lab’s direct inheritance from the book. In the DOTF context, it meant running the Why-What-How video series even before the answers were fully formed — committing ideas to video while the design was still in progress. The book frames it as prioritizing speed of communication and iteration over perfection, requiring psychological safety to do so. The December 2020 story is a reminder of what happens when the environment does not provide that safety, and the team commits to the work anyway.

AI Disclosure: as usual, AI helped me research, write, and produce this piece. All ideas, memories, observations, and conclusions are my own. Mistakes too.

If you are wondering about the bat in this week’s image:

Microsoft Insiders

Boundary Crossers: The Innovation Roles We Don’t Talk About Enough

Innovation is not a linear pipeline—it’s a loopy living system. Like nature, it moves through cycles of emergence, growth, integration, and renewal. It requires ecosystem thinking, adaptive capacity, and res…

Read more

a year ago · 4 likes · JoAnn Garbin

Say it Ugly. Build it Better.

Onward!

I get terrible seasickness. I hate small planes. Antarctica has never appeared on any version of my bucket list.

I signed up anyway.

A few days later, my friend Larry and I were deep into nerding-out about innovation when I mentioned my second voice in passing. He stopped me.

Wait. Explain that. What’s your second voice?

Subscribe now

Innovation culture talks constantly about the courage to change. It almost never talks about the voice that tells you not to.

That’s the first voice. We all know it. It’s the one running continuously — the inner monologue, the risk register, the status checker. It evolved over human history to do a specific job: keep us alive, keep us connected, keep us from doing things that get us excluded from the group. It’s good at that job. In most contexts, it’s indispensable.

In innovation, it’s a problem.

Because the first voice’s entire threat library is made of things that innovation requires. Don’t challenge the status quo. Don’t bet on something unproven. Don’t be the person who disrupts what’s working for everyone else. Don’t make people uncomfortable. Don’t be wrong publicly. Don’t be disliked.

Changemakers get disliked.¹ ² ³

Not universally. Not always. But often enough that pretending otherwise is a form of dishonesty about the work. People hate change. And we are human about it — we translate the discomfort of change into feelings about the person associated with it. This isn’t pathology. It’s social cohesion doing exactly what it evolved to do.

The first voice knows this. It has millions of years of evidence. It would like you to fast-follow, please. Let someone else go first. Wait until it’s safe.

The second voice — mine, anyway — is not a cheerleader.

It doesn’t say you’ve got this or it’ll all work out. It’s not an alter ego with a name and a costume. It’s quieter than that. It’s a voice I’ve developed — or maybe it developed while I wasn’t paying attention — that shows up in the same instant as the first voice.

Mine specializes in recovery and patience. It tells the first voice to take a nap. Eat something. Take a walk. Set it aside and think about it tomorrow.

It shows up as often as the first, but especially at the end of the day, in reply to my evening reflection — if today was my last, was it a day well spent? Two nights ago the first voice made its case: are we even doing anything worth doing, or are we just goofing off? The second voice arrived at the same instant: not every day is a breakthrough. We worked. We learned. Go to sleep. It’s usually more Jersey than that (don’t be silly, go to bed).

That’s not dismissal. It’s a specific practice. The second voice has accumulated enough evidence over years — enough cycles of we bet on this, we did the work, we figured it out — that it can interrupt the fear response before it closes off the option space. It doesn’t know how things will go. But it knows the first voice is not the right decision-maker in this moment.

Leave a comment

The speed-and-output culture of innovation often treat recovery and patience as weakness or delay.

They are not. They are the infrastructure.

The ambiguity that innovation requires isn’t a phase you pass through on the way to certainty. It’s a permanent condition. The best practitioners I know aren’t comfortable with uncertainty because they’ve figured something out — they’re comfortable because they’ve learned to function while not knowing. That’s a trained capacity. It’s not a personality type. And it doesn’t make them superhuman. They have the first voice too. We all do. That’s not a flaw to fix. That’s human nature.

The same is true of the social cost. Being the person who sees something others aren’t ready to see — who names the constraint the organization doesn’t want named, who bets on the opportunity that looks foolish at current visibility — carries a real relational weight. You can manage that weight or not. But you cannot pretend it isn’t there without distorting how you make decisions.

The first voice processes that weight as danger. It’s not wrong. The second voice processes it as the cost of the work — not a reason to stop, but a variable to account for, to recover from, to be patient with.

Over nineteen weeks, this series has mapped a lot of terrain: systems, tools, collaboration architecture, AI as thinking partner, expert perception, incentive structures, the ecology of innovation at scale.

I’ve spent most of that time one level above the person doing the work.

This week I’m inside. Not to get philosophical — to get practical. The gap between knowing what good innovation practice looks like and being able to sustain it under real conditions lives right here. In the space where either one or both voices show up.

The second voice doesn’t make you different in kind. It makes space. Space to breathe, to sleep, to not look at it directly, to remember that life is more interesting with the work than without it. It doesn’t cheer you on. It doesn’t promise outcomes. It just keeps the option open long enough for you to act.

That’s not a cheerleader. That’s the infrastructure.

I’ve had my second voice long enough that I sometimes forget it required development. So my question is what deliberate practice of that capacity looks like — for the person who can see the opportunity and can’t yet quiet the voice that’s telling them to wait for it to be safe. That question is where we’re headed next.

¹ Ignaz Semmelweis. Hungarian obstetrician, 1840s Vienna. Discovered that doctors washing their hands between autopsies and deliveries dramatically reduced maternal mortality from childbed fever. His colleagues took offense at the implication. Semmelweis was dismissed from his post, eventually committed to an asylum, and died there in 1865 after being beaten by guards — two weeks after admission. Germ theory, which would have explained his findings, arrived a generation later.

² Alfred Wegener. German meteorologist. Proposed in 1915 that the continents had once been a single landmass. The geological establishment dismissed his theory as “delirious ravings” and “Germanic pseudo-science.” At one Royal Geographical Society meeting, an audience member thanked the speaker for “having blown Wegener’s theory to bits” and then thanked the absent Wegener “for offering himself for the explosion.” Wegener died on a Greenland expedition in 1930. His theory was accepted as plate tectonics in the 1960s.

³ Barbara McClintock. American geneticist. Discovered transposable elements — “jumping genes” — in maize in 1948. When she presented her findings in 1951, she was met with what she described as “puzzlement, and even hostility.” She stopped publishing on the subject for years. She won the Nobel Prize in Physiology or Medicine in 1983 at age 81 — more than thirty years after the discovery, and the first woman to win unshared in that category.

To name a few…

Connections to The Insider’s Guide to Innovation at Microsoft

• Behavior is the Barrier: The barrier isn’t always in the organization — sometimes it’s in the practitioner. The first voice is Behavior is the Barrier operating internally. The innovator who can’t quiet it becomes the bottleneck in their own work.

• Cognitive Inertia: The first voice is cognitive inertia in first person. It doesn’t resist change in others — it resists change in you. The second voice is the compelling force that doesn’t come from policy, incentive, or peer pressure. It comes from accumulated internal evidence.

• The B2Me Journey — Applied Inward: B2Me, applied outward, is the journey a stakeholder takes from unawareness to advocacy, guided by the innovator. The second voice runs that journey internally. When the first voice fires the fear response — are we even doing anything worth doing? — the second voice doesn’t lead with logic. It leads with emotional reset: not every day is a breakthrough. We worked. We learned. Recovery before rationale. The internal B2Me journey is the same architecture as the external one.

IOL is produced with the help of AI, specifically Perplexity, Gemini, Claude, and a team of custom personas developed by Regenerous Labs. All insights, editorial choices, and final content are mine. Mistakes and all.

Say It Ugly, Build It Better. Onward!

The work has been moving fast lately — fast enough that I keep asking whether the team around it is wide enough for the moment we’re in. Most of us sense that kind of gap before we can name it.

These last few weeks I’ve been sharing some of the experiments we’re running in the Lab. Tools built on our CORTX behavioral intelligence engine.

But around the things we’re building is a bigger build. An ecosystem. At its core is the Lab — a network of longtime collaborative partners. Supporting that core is a system of AI-powered tools we can call on at any hour and expand as the work demands. We’ve been building that out for over a year. More recently, we started building an advisory board — a network of people operating at strategic altitudes most of us only read about.

What we’re doing is a practice I wrote about a few years ago: expanding the range of domains and perspectives you receive signals from, and surrounding yourself with people who’ve done the same. Repeat innovators build arrays. That’s how they detect faint signals earlier and act on them sooner. That’s how they turn change into opportunities.

Subscribe now

Our Array

Each layer does something the others can’t.

The AI expands the array’s reach. We can interpret what we’re hearing across far more domains than we could otherwise access. We can pull a perspective from biomimicry, incentive architecture, or historical systems analysis into a working session in minutes.

The advisory board expands the array in a different direction. There are rooms none of us in the Lab have been in. Conversations at the Fortune 100 CEO level, on the global policy stage, at the tables where heads of state make decisions. Those rooms are receiving signals we’re not. The advisory layer is about finding people positioned differently, whose vantage point is higher and wider than ours.

The Lab is where signal becomes something real. The other two layers are perception infrastructure. This one is the synthesis.

Our Bet

Three layers, one question: what does the moment require?

Major technical shifts have sometimes enabled major economic shifts — not by accelerating what already exists, but by making a different set of ideas structurally possible. We in the Lab believe this is one of those times and have an idea of what that shift might look like: an economy less organized around consumption and more organized around builder verbs. Not just producing new things. Maintaining what exists. Tending what’s growing. Clearing what no longer serves.

If something like that shift is underway, the people building toward it now — rather than waiting for confirmation — will have years of compounded perception and preparation that others won’t. We’ll have had a hand in creating it.

The question isn’t whether we’re certain. It’s whether the bet is worth making.

Seizing the Opportunity

Faint signals, received alone, are easy to misread. Combined across a diverse array, they become clearer.

What is the breadth and depth of your current view, and is it enough for the moment you think you’re in?

Leave a comment

Connections to The Insider’s Guide to Innovation at Microsoft

• Network of Known Collaborators: Build the relationships before you need them. The lab and the advisory board are pre-assembled networks, activated when the work demands.

• Customer Hypothesis Instead of Idea: DevDiv’s reframe applies here: the bet on what this moment is isn’t a conclusion to defend. It’s a hypothesis to test.

• The Regenerative Datacenter of the Future case study: A core team of six flexed to dozens on any given day through an intentionally built ecosystem of experts — the same architectural move, applied to a different problem.

Innovating Out Loud is a weekly series developing ideas in the open. As with all pieces in this series, AI was used as a research and writing partner throughout. All conclusions, opinions, and editorial decisions are mine. Say it ugly, Build it better. Onward!

The post-mortem is honest. Nothing was wrong with the vision. The argument was sound. The slides were clean. The case was the right one to make.

And still.

Most of us have a private theory about why this happens, and most of us are partly right. Maybe the room was off. Maybe the timing was wrong. Maybe the buyer was three-quarters out the door before we walked in. All true, sometimes. But there’s another pattern underneath, and it shows up often enough to name.

We were solving for what we wanted to say. They were solving for what they could hear.

Those sound like the same problem. They’re not. In any audience, several “what they could hear” problems happen at once — each shaped by the listeners’ risk style, what they already trust, what they’re ready for, and the kind of evidence they accept.

Most of us know this. We’ve read the books. We’ve sat through the trainings. We tell our junior colleagues to know your audience and we mean it.

Then, often, we sit down to craft our presentation and write it for ourselves anyway.

This isn’t a knowledge gap. It’s another knowing-doing gap. And that gap is structural. Holding three or four audience models in your head while you decide what to put on a slide is expensive cognitive work. So we shortcut. We make the deck we’d want to see. We assume the audience will translate.

Sometimes they do. Often they don’t.

Subscribe now

Enter this week’s experiment, CORTX Reframer. It’s built around a single architectural principle: the tool serves the audience, not the presenter. It doesn’t just polish what you already wrote, it asks, on the audience’s behalf, whether your story can be heard.

You upload a deck. You pick the listener you’re trying to reach — a relationship-builder, a pragmatist, a visionary, or a skeptical analyst. Each is a real expert worldview, drawn from research on how different people make decisions: trust formation, readiness for change, identity and meaning, mechanism and evidence. The tool reads your deck through all four lenses at once and shows how each one scores it. Then it reframes by reorganizing the story behind your vision for the listener you are actually trying to reach. Story science and behavioral science, doing the work together. Then it scores the reframed version again, through all four lenses.

The lenses moving is the proof.

You can watch this happen in about sixty seconds. Upload a real deck, pick the skeptical analyst, and the tool may say something like: this lands at four out of ten for the analyst; the evidence beat is weak. Hit reframe and the same content reappears organized around mechanism and data, with the proof front-loaded and the ask calibrated to the kind of decision an analyst actually makes. Then the re-score. Six points stronger.

The vision didn’t change. The story did.

This is a practice my partners and I have been using for decades.

This is the third experiment in the Showing Instead of Telling arc — Innovation Coach in March, CANOPY last week, Reframer this week. Each one is small. Each one is built on CORTX the Lab’s behavioral intelligence engine.

Try it. Take a deck or script you’ve been working on, run it through, and tell me what you find.

www.regenerouslabs.com/reframer

Say it ugly, build it better. Onward!

Leave a comment

Connections to The Insider’s Guide to Innovation at Microsoft

• Behavior is the Barrier — Seventy percent of transformations fail for human reasons, not strategic ones. This piece is the pattern made personal: every presentation that flopped, every pitch that didn’t land, every meeting where the right argument got the wrong response — the gap between knowing your audience and writing for your audience is where most of it lives.

• The Knowing-Doing Gap — Knowing what to do and doing it are not the same. Holding three or four audience models in your head while you decide what goes on a slide is expensive cognitive work, so we shortcut and write for ourselves. Reframer holds the complexity so you can act on all that you already know.

• Language as a Strategic Tool — Same vision, different register, different decision. Reframer doesn’t change the case. It changes to which version of the case the listener’s brain can actually run. The scores moving is proof of this pattern in practice.

• Start With Who — Reframer’s first move is to ask who is in the room before it touches the content. The pattern most often applied to successfully assembling the team applies just as much to addressing the audience. Different listeners need different stories — Start With Who, then build the case.

• Innovate Upstream to Succeed Downstream — The decision happens in the room. The conditions for it are made before. Reframer is upstream work — done before the pitch, the renewal call, the board meeting — that enables what happens downstream.

AI Disclosure: This piece was built with AI, specifically Perplexity, Claude, Gemini and a team of custom expert personas built by Regenerous Labs. All ideas, judgements, and final edits are mine.