if you enter a market that is already obviously automatable, you’re late. the accounting wave is at its peak precisely because the automatability became obvious, and obvious markets fill with consolidators until only one can win.

the real game is anticipation: pick the market that is not yet obviously automatable but will be in three to four years as the models improve, and build the distribution now. so that when capability arrives, you pass an AI layer over a network you already own and win by default. the construction rollups betting on robotics are playing this. the cleaning company filming hallways nobody photographs is playing this.

distribution built ahead of capability is the trade. the prediction economy is what the trade looks like when everyone can see it. by then, the position is closed.

the world this produces

in five years, every credible operator in a vertical will have the same automation. models commoditize. operations replicate. and over each vertical that passes the multiplier filter, a panel will exist: someone will hold the platform layer and watch that market’s economy move months before it moves on paper.

competition will look like this. two acquirers in the same vertical, same capital, same AI stack. one pays per introductory meeting to discover whether an owner might sell. the other has watched engagement, staffing, and cash behavior across the vertical for years, and sees, at pattern level, where the succession wave is forming, before anyone has made a call.

same army sizes. that is not a fair fight, and it never becomes one. because the disadvantage isn’t capital, talent, or software. it’s years, and years cannot be purchased retroactively.

vertical markets bifurcate into two kinds of players: the ones who see markets coming, and the ones who get seen coming.

the sighted originate at prices the blind can’t understand. and the operators who spent the decade optimizing level-two margins without instrumenting the panel discover their true role in the story: they built the telescope, node by node, for someone else to look through.

the state is the last buyer

every buyer named so far is private. run the era forward and the largest buyer class arrives last.

states are blind at exactly the layer that decides their crises. a government’s sensor grid is its statistical apparatus, built for accuracy, not speed: surveys, filings, revisions, publication lags measured in months. every recent crisis proved the same thing. when states subsidized their SMB economies through the last shock, they subsidized blind: relief flowing into sectors nobody could see in real time, arriving late where it was needed and generously where it was not.

this has already happened once. during covid, economists built a real-time economic tracker from private-sector data, payroll processors, card spend, because official statistics were too slow for policy. the state’s blindness in crisis, and the private panel that filled it, is not a hypothesis. it is a documented precedent.

the era this thesis describes makes that blindness strictly more dangerous. a state whose institutions reason with abundant intelligence over stale ground truth is not a smarter state. it is a faster-moving blind one. the gap between the sighted and the substrate is not only a gap between acquirers. it is a gap between jurisdictions.

the US is where this gets built first. the largest SMB economy in the world, the deepest pool of multiplier verticals, the most active buyer class for vertical software, and the most capital stacked behind AI rollups.

the state’s own incumbent cannot build this. the statistical office holds the mandate and cannot spend it. built for accuracy and delay, bound to methods that make its numbers comparable across decades. which is why the state arrives the way every actor arrives in this thesis: as a buyer.

the next crisis is not a hypothesis, only its date is. a state that cannot see its own SMB economy in real time will, when it comes, subsidize blind again, and will buy the sight afterward, at panic prices, from whoever built it.

Subscribe now

the bet, priced

every thesis is a position, and a thesis that won’t name its bet can’t be priced.

the bet: long the multiplier layer persisting through the build decade. AI transforms the workflow, the multiplier stays. that is what this entire structure stands on.

price it honestly. the band does not hold still. every model release moves it upward, and somewhere past 2030 the legally reserved tasks, the signature only a licensed advisor can give, the placement only a broker can make, may be the last wall standing. regulatory walls are real. permission is a true moat. they are also exactly what gets renegotiated when the productivity gap becomes absurd.

so the bet is not “does the layer persist.” the bet is “does the layer persist longer than the build takes.”

the squeeze. the panel ripens on the same clock the layer may wither. five years of calibration started in 2026 matures around 2031. the band slide is plausibly 2030. the asset becomes trustworthy at roughly the moment its scaffolding becomes negotiable.

here is why that is survivable: the panel watches the SME economy through the multiplier layer, not the multiplier layer itself. the consent belongs to the end customer, not the advisor. if the advisor layer thins after density is reached, the sensor keeps reading. the multiplier is the launch mechanism, not the habitat.

the squeeze does not break the thesis. it prices it: density must be reached while the multiplier still mediates. reach it before the band slides, and the panel outlives its scaffolding. miss it, and there is no second launch. the second hand on the same clock.

three predictions

a thesis that risks nothing predicts nothing. so here are commitments. dated, checkable, falsifiable.

prediction one. by the end of 2027, serial acquirers of vertical software close at least five further acquisitions of vertical information authorities. benchmark businesses, data authorities, signal products with paying subscribers. Keypoint and RealEstateAPI were the start of a pattern, not exceptions. if the pattern stops, this thesis loses its strongest external evidence.

prediction two. by end of 2028, at least one vertical in the US economy has a live panel that institutional buyers, insurers, banks, or macro funds, are paying for as an alternative data feed. if no vertical produces a priced panel by then, the third level described in this series does not materialize at market scale.

prediction three. by 2030, at least one state institution, a central bank, a sovereign fund, or a federal agency, is purchasing a real-time SME panel as an input to economic policy, crisis allocation, or regulatory oversight. if the state never arrives as a buyer, the panel’s ceiling is set by the private buyer class alone.

the software was never the asset. the operation was never the endgame. sight is. and the window to build it is the short interval before everyone else realizes what the people buying hundreds of vertical companies a year have already figured out.

the sophisticated money is already moving.

this concludes the thesis. the full document, with every receipt, every mechanism, and every objection answered, is at predictioneconomy.ai

what follows on the site is part two: the vehicle testing whether the position is manufacturable rather than accidental. dis-tribution.com. three phases. 2026 through 2030. six dials. the worst case priced. the window is open.

thank you for reading. if this series changed how you see vertical software, AI rollups, or the consolidation wave, share it with the operator or investor who needs to read it.

the full whitepaper is a single PDF download at predictioneconomy.ai

Subscribe now

I had coffee with a Harvard dropout founder yesterday. He is building an agentic company that retrieves university data for scientists.

We had a long conversation about what his moat really is and why the labs can’t do what he does. He kept coming back to how good the product was, and argued that the labs wouldn't want to gather university data manually.

“Is that the reason someone will buy your company?” I asked him.

He looked at me as if I had just asked him if I could have his firstborn child.

“Someone will buy us because we are building great software and we will have stable revenue.”

Building software nowadays has become easier than filing your taxes. It also means software is easy to copy, which raises a question: If not the tech, why would a company building the application layer on top of foundation models get bought?

The Old vs the New World of Software Multiples

2021 Software Exits were at their all-time high. 17X revenues were expected. An acquirer could buy a software company and be confident that revenue would stay roughly flat for the next 10 years with minimal input. They could run it as a cash cow or upsell to expand ACVs, and with them the value of each customer.

The Revenue Multiple was a good metric to value a company, because the revenue used to be:

A) Sticky

B) Stable (you could cut costs and it held steady)

2026, that’s different. Buying a 10X revenue company today gives a buyer no guarantee that the labs won't replicate the product within two years. Churn has risen sharply because building is easier than ever, which has created a red ocean.

Additionally, switching costs have fallen from a 30-day implementation and database rebuild into a 10-second automatic upload and automatic retrieval of your data. Those changes made software customers less sticky and easier to replace, and revenue harder to forecast..

Software businesses need a new way to be valued.

The Meat

The business model of most of Silicon Valleys AI-native companies:

Revenue = Token Spend forwarded to Customer + 15% Margin for application layer + (sometimes) implementation fee / service fee.

Whether it's called per-seat or usage-based pricing, it comes down to the same thing: the application layer passes most revenue to the labs and keeps a thin margin.

Labs can raise or lower their prices and the willingness of end customers to pay a premium for the application layer varies which makes revenue unpredictable in the long run.

And since the software takes minutes to replicate, revenue is a useless metric for M&A. So if not for revenue, why would anyone buy the company?

The answer is distribution and relationships.

The New Paradigm

I believe that in the near future it will get cheaper in some industries to buy a customer base than to grow organically. What we see today as vertical application layers will become time machines that let consolidators enter markets faster.

Companies will be acquired solely for their distribution and relationships. M&A will become a CAC-arbitrage game, which means it will be cheaper to buy a competitor’s users than to acquire them organically.

The two factors that will define how much a software company is worth at exit are:

1. “Cleanness of the Ideal Customer Profile (ICP)”: How tightly defined are your customers industry, company size, and decision-maker, and how hard is that market to reach?

2. Customer Acquisition Costs (CAC): What it typically costs to acquire those customers

Target groups that are hard to reach (scientists, lawyers, Enterprises) will be valued higher than target groups that usually have a cheap CAC (students, early adopters, founders, etc.).

The formula of how companies will get bought: Bending Spoons built a whole company on that theory and we will see their playbook more and more in the AI era: buy a product with a big user base, cut most of the team, keep the users. HubSpot did the same thing with The Hustle, and we saw it with Miro and Airtable as well.

What this means for your business

Revenue will no longer be the defining factor. I believe we will see more and more companies bought solely for their customer base, with the software sunset afterward.

Most founders are still optimizing for building and shipping. It won't be long before they realize that software quality adds little to nothing to their exit valuation.

Stop building. Start selling. Focus on the target group. That’s going to be the value driver.

The ex-Harvard founder already had the answer in his hands: the university access and network he's building is the value driver. He just wasn't optimizing for it. Now he is.

Shield is an MSP rollup. managed service providers. the companies that run IT for small and mid-size businesses.

read that transaction carefully because it is the tell.

OpenAI is not buying MSP margin. they are buying telemetry into every SMB that runs their IT through an MSP.

meanwhile, General Catalyst led a $74M round in Titan, another AI-enabled MSP holding company, from GC’s $1.5B creation fund. roughly half earmarked for AI-driven consolidation in traditional services.

the surface argument for both plays is level two: buy MSPs, automate the manual work, capture the margin expansion. that is not what the smart money is actually paying for.

the read that’s missing

MSPs are the last delivery layer of AI into the 95% of the economy that will never adopt AI directly.

no small manufacturing firm is going to hire an AI transformation consultant. no regional accounting firm is going to build their own agent stack. they are going to call their MSP. and their MSP is going to install whatever their consolidated parent company tells them to install.

whoever owns the MSP layer owns AI adoption for SMB. not through better marketing. through the delivery pipe itself.

and every SMB an MSP serves generates operational data. every ticket. every network event. every incident. every deployment. every security posture check. at one MSP that is noise. at 200 MSPs across a national market that is a real-time panel on SMB operations at a level no incumbent tech vendor holds.

which sectors are hiring. which are laying off. which industries are getting hit by ransomware. which companies are consolidating. all of it visible before it appears in any published statistic.

the operators most bullish on MSP consolidation are explicitly not optimizing for near-term profitability. they are maximizing coverage. maximizing telemetry surface area. maximizing the panel density that only exists once you cross a threshold.

margin expansion is the story they tell the LPs. telemetry acquisition is the actual play.

^{Up close, a Chuck Close painting is meaningless blobs. Step back, and a face snaps into existence.}

the pattern is not just MSPs

Metropolis is not a parking company. it is a real-time telemetry of urban movement. every vehicle in every lot generates data. entry time, dwell time, payment behavior, revisit frequency. the parking business pays for the sensor grid. the telemetry is the asset.

the same pattern runs through every vertical where a service operation crosses the density threshold at which its byproducts become more valuable than its outputs. a consolidated tax advisor network is a telemetry of the SMB economy. a consolidated property manager is a telemetry of residential real estate cash flow. a consolidated insurance broker network is a telemetry of SMB risk.

this is the inversion most operators miss. in the old model the software is the product. in the emerging model the software is the sensor and the panel is the product. the most valuable companies of the next decade will be indistinguishable from telemetries of the economies they operate in.

who else holds sensors

the panel will not be the only sensor grid pointed at the SMB economy. so the question that decides everything: what does it see that the others don’t?

the bank. sees its own book, not the market. a $50k inflow could be revenue, a loan, or an owner’s deposit. the classification lives one layer up, with the bookkeeper. and most structurally, the bank is a counterparty. the owner negotiates against the bank. the pre-event signal is systematically withheld from it. broad but blind.

the payment processor. real-time sight of the card-and-online fragment of revenue, in an economy that runs on invoices and bank transfer. no cost side. no payroll. no P&L. fast and shallow.

the payroll provider. ADP moves markets with one axis and decades of calibration. proof of concept, not competition.

the labs’ agents. every AI agent inside an SMB is itself a sensor. but the agents see how a company works, not what it earns, owes, and pays. workflow, not meaning.

everyone else sees movements. the multiplier layer, the tax advisors and brokers and property managers who actually do the books, sees meaning. classified, reconciled. and meaning plus calibration is the layer that turns out to be time-locked.

Subscribe now

why the panel appreciates while everything else decays

now the objection. signals decay. official statistics catch up. any specific data point is worth a lot in july and worth nothing in eighteen months. so how does the panel appreciate?

the value was never the archive. an archive is a museum. the value is the freshest feed read against the longest calibration history.

one quarter of SMB cash-cycle data is noise. you can’t distinguish seasonality from structural shift. you don’t know what a normal june looks like so you can’t spot an abnormal one. five years of data plus this morning’s feed is a different asset entirely. panels are valued by length and freshness together. this is how alternative data has always been priced.

ADP has run the panel for decades and calibrated its signals against BLS revisions thousands of times. ADP’s asset is not this month’s employment data. it is the calibration history that makes this month’s data readable.

Bloomberg knows this. $30,000 per user per year. what a subscriber pays for is not the current price of every asset. every price is public. what they pay for is decades of aligned, calibrated, timestamped history that lets them price relative to something coherent.

Constellation proved it happens in verticals too. in may 2026 they acquired Keypoint Intelligence, the benchmark authority of the digital-imaging industry. every fact Keypoint produced in 2024 was stale by 2026. that was never the point. the calibration position had accumulated. that is what Constellation paid for.

nobody has ADP for SMB tax filings. nobody has Bloomberg for property-management operations. nobody has Keypoint for MSP incident data. the panels that would make those markets legible do not exist. the multipliers that would generate them are being consolidated for the first time now with ai.

time-locked

a competitor with unlimited capital cannot buy the years back. they can outspend today. they cannot replay 2026 through 2031 to build the calibration history the position requires.

this is what “time-locked” actually means. not “hard to build.” mechanically impossible to compress. the tool depreciates because AI writes it. the operation replicates because a competitor imitates it. the panel appreciates because history only runs forward.

the position is available in a way it will not be in five years. the panel that exists in 2032 will be the one someone started building in 2026. there is no shortcut.

and be honest about what each layer sees, because the exclusivities decay at different speeds. workflow exhaust is already contested. the labs are buying it. assume zero exclusivity within a few years. meaning is different. the labs’ agents see how a company works. they do not see what it earns, owes, and pays. and they cannot backfill the five winters of baselines that make a fresh number legible.

above meaning sits intent. the advisor hears “i’m thinking about succession.” no telemetry hears that. today. but run the trendline forward honestly: owners are beginning to tell their AI things they tell no human. intent’s half-life is shorter than meaning’s. it is a wasting advantage: real now, decisive now, and to be banked now.

meaning appreciates. intent burns. that is this thesis in four words.

the full thesis, with the complete evidence and every rival sensor analyzed, is at predictioneconomy.ai

Subscribe now

they couldn’t source it through brokers. they couldn’t generate it through outbound. they couldn’t wait for it to accrete organically. so they paid someone to hand-build it and transfer it.

that transaction taught me something the rest of this series is about: the buyer didn’t want my software. they didn’t want my operation. they wanted what the network could see. which owners were ready, which weren’t, and how to tell the difference before anyone else could.

the network was the sensor. the signal was the product. and the buyer was paying for something most operators don’t realize they’re sitting on.

software went through three pricing eras. sell the tool: 2-5% value capture. sell the outcome: 30-60% value capture. both are already commoditizing. a competitor with a coding agent rebuilds the tool in weeks. a funded competitor replicates the operation in years. neither is permanent.

there is a third level. it monetizes a different object entirely. not the work, but what the work lets you see.

when you hold the platform layer across a meaningful share of a vertical’s connective tissue, you stop being a software business. you are watching an entire end-customer economy evolve in near-real time. which sectors are heating up. where margins compress. where cash cycles stretch. months before any of it appears in a published statistic.

the object is not the software. not the operation. it is the aggregate anonymized signal your reach generates about the economy behind it.

the tool and the operation exist to feed it. in level one and level two the software is the product. in level three the software is the sensor and the panel is the product.

the ladder

what does the telemetry actually give you? five rungs, building in sequence.

selection. which firm to buy, which market to enter. real, but weakest. selection without the next rungs is just a better watchlist.

pricing. what it’s worth before anyone else can compute it. you know the practice’s true cash trajectory while the seller’s own bank is reading stale statements. the blind acquirer and you bid on the same asset. you’re not outbidding him. you’re out-knowing the price.

timing. not which, not what it’s worth. when. the telemetry tells you which market becomes automatable in three years, which succession wave forms in eighteen months. you enter before the re-rating, before the chorus.

position. in liquid markets, the first three rungs decay. knowledge arbitrages away. in the markets this thesis targets, illiquid, private, small, knowledge converts to ownership before it can leak. signal, allocation, network, sharper signal. the advantage stops being something you spend and becomes something that compounds.

reference. at sufficient density, something qualitative happens. you stop extracting edge from the market and the market starts coordinating through you. banks price SME credit against your feed. insurers set premiums against your baselines. your information isn’t an advantage anymore. advantages decay. it’s infrastructure. infrastructure earns rent.

selection tells you which. pricing tells you what it’s worth. timing tells you when. position makes it compound. reference makes it permanent.

the player becomes the house.

and there is a sixth rung. at density, in small markets, the arrow reverses. you stop predicting the futures and start scheduling them. you trigger the succession wave, three nodes at a time, matching aging books to young advisors through your own network, and watch the effects propagate through your own sensor. foresight becomes agency.

two paths once you hold the panel

sell the signal. anonymized, aggregated, structured for the third party who cannot build it themselves. hedge funds, banks, insurers, macro desks. the ceiling is set by what data buyers pay per year. the exit has a proven shape: Keypoint Intelligence, RealEstateAPI. bounded, defensible, predictable.

deploy the signal. keep the panel internal. use it to make your own allocation decisions. which SMEs to buy before their bank knows. which broker books are approaching succession before the owner has made a call. signal, allocation, network, signal. the ceiling is set by the size of the vertical you’re building through.

both paths prove level three monetizes a different object than level two. neither monetizes the operation. both monetize the information advantage the operation generates as a byproduct. that is what makes level three a level.

Subscribe now

the AlphaGo objection

the third level rests on a bet: that data gathered from the world stays valuable in a world of abundant intelligence.

go is a closed world. perfect information. the board simulates itself. AlphaGo learned from human games, then played itself, then AlphaZero skipped humans entirely. human data gets you to human level. self-play gets past it. eventually the human data isn’t needed at all. data that teaches skills is a wasting asset.

but the economy is not a board. the economy must be observed. no amount of self-play generates the fact that layoffs started in freight forwarding last tuesday. a superintelligence in a datacenter is still blind. it cannot deduce this quarter’s margins in austrian dental practices from first principles. someone has to measure the world. compute cannot conjure ground truth.

in a world of abundant intelligence, the scarce input flips from thinking to seeing. models commoditize. sensors don’t. the rothschild edge survives even AGI. it may be the only edge that gets more valuable under it, because faster actors extract more from the same time advantage.

the defensibility hierarchy. three sentences.

the tool depreciates. AI rebuilds it in weeks. the operation scales. a funded competitor replicates it in years. the panel is time-locked. nobody can replicate it without replicating your entire network history. and history is the one thing money can’t compress.

the multiplier

so the mechanism matters. how does a network actually generate economy-scale signal? not scraping. not surveillance. not buying data.

a multiplier is a firm whose customers are themselves firms.

one tax advisor serves dozens to hundreds of small businesses. one insurance broker sits in front of dozens of policy-holding entities. one property manager holds the operational layer of many owners’ buildings. behind 1,500 advisors sits a meaningful share of a country’s economy. real-time financial data. bottom-up macro signals. cash-flow patterns across the entire small-business economy.

you don’t build reach to the network. you build reach through it. one hop, through a partner the end customer already trusts. reach to end customers adds. reach through multipliers compounds.

the selection rule. fragmentation × end-customer leverage. verticals that pass: tax advisors, insurance brokers, MSPs, property managers. verticals that fail: direct-to-consumer services, verticals with one incumbent controlling the layer.

two laws.

law one: organic pace builds footholds, not networks. austria has 3,000 tax advisors. germany has 46,000 insurance brokers. the US has 50,000 property management firms handling 20+ million doors. at aggressive organic pace, a platform reaches low single-digit penetration in two years. that is a foothold, not a network. no acquirer pays for a foothold.

law two: the multiplier can never be the one paying. the moment the advisor has to pay, they slow down. they compare. they negotiate. they wait for budget approval. the whole flywheel jams. velocity is the economics. payment is friction. the multiplier rides for free. a third party pays.

what sits behind the verticals. behind each tax advisor: the bookkeeping reality of every SME they serve. behind each insurance broker: the underwriting reality of a national SME insurance market. behind each property manager: real-time patterns on maintenance backlogs, vacancy trends, rent behavior, owner distress. the buyer for this signal isn’t hypothetical. Beacon already paid eight figures for RealEstateAPI.

why now. the ownership layer is aging. most tax advisor owners are past 55. most insurance broker owners are past 55. and the automation grade sits exactly where the third-level play requires. high enough that AI can meaningfully transform the workflow. not so high that the multiplier disappears.

one hop to an economy.

that is what the position is worth. if the signal it generates is actually worth something. which brings up the strongest objection to this entire thesis. part 3/4 answers it.

Subscribe now

the full thesis is at predictioneconomy.ai.

napoleon against wellington. the battle that will decide the fate of europe. and the outcome will decide the price of every bond in the market.

but one family has spent years preparing for this moment. the rothschilds have built the fastest private courier network in europe. riders, boats, and, so the legend goes, carrier pigeons.

napoleon lost.

nathan rothschild knew a full day before the government did. and he traded on it. the legend prices his gain in the hundreds of millions in today’s money.

the edge was never capital. it was seeing first. the couriers were not a messaging business. they were sensors, pointed at a continent, owned by the one family that understood what a day’s head start was worth.

this document is about the courier networks being built right now. and about the people building them, who mostly aren’t saying so.

intelligence is only worth what it can see.

i tracked the last fifteen acquisitions across the most sophisticated serial acquirers of vertical software. Constellation Software, OpenAI, Beacon. deal after deal, the same shape: buy the layer that serves a market, and you see that market move before anyone else does.every shift in the economy shows up in the books of the people who serve it, months before it shows up anywhere else.

they are buying the pigeons.

the moat is moving. software is commoditizing, and value migrates up a layer each era. SaaS is over. service-as-software runs the tool and sells the outcome. but extrapolate the trendline to where it terminates: models become a utility, operations become a commodity, and distribution, the network that gives intelligence its eyes, becomes the last layer standing.

i call that phase the “prediction economy”: the point at which the winners in vertical markets are no longer the ones who automate the work, but the ones who see the market through the work. and the advantage compounds: seeing first becomes a selection advantage, then a pricing advantage, then a timing advantage, until it converts into a compounding position, and finally into a terminal asset. you become the reference layer the market prices itself against.

the player becomes the house.

and the position, once built, cannot be rebuilt. because the advantage is not capital, talent, or software. it is accumulated history. a market panel that started recording in 2026 can never be caught by one that starts in 2028. history is the one thing money can’t compress.

the rothschild edge, industrialized.

from software to infrastructure

three things changed in the last eighteen months. AI collapsed the rebuild cost. the buyers moved up the stack. and capital piled up faster than credible vehicles to absorb it.

level one: sell the tool. classical SaaS. 2 to 5% of value captured. in june 2025, Wix paid $80 million for Base44, an app-building platform six months old, built by one man, with no outside funding. one person, half a year, eighty million. a competitor with a coding agent replicates most SaaS in weeks. code depreciates.

level two: sell the outcome. same software, but you run it yourself. Crosby sells legal filings. Corgi underwrites insurance policies, full-stack, $2.6B valuation eighteen months after founding. capture jumps to 30-60%.

Sam Lessin takes it further: buy the businesses you operate better. Metropolis raised $1.8 billion and took SP+ private. four thousand parking locations, four billion a year in payments, one company’s computer vision. the AI rollup wave is this argument in execution. Crescendo, Long Lake, Dwelly, Titan. all doing the same thing in different verticals.

level two works. but ownership is a capture structure, not a level. a level monetizes a different object entirely. and the defensibility question doesn’t go away: every Metropolis has a competitor two years behind it with the same thesis and a cheaper cost of capital. Long Lake and Dwelly are the proof. same band. two geographies. funded independently.

run the sequence forward. level one goes to free. HubSpot proved that a decade ago. level two goes to commodity. every credible operator will have the automation.

software stops being a product and becomes infrastructure. everyone has it. nobody wins on it.

reach, then sight.

april 2026. OpenAI makes its first media acquisition: TBPN, a daily tech talk show. fifty-eight thousand youtube subscribers. reported at hundreds of millions. run the math per subscriber and the deal is absurd. run it per room reached and it’s cheap. the company whose models everyone rents concluded the scarce asset wasn’t the model. it was the channel.

HubSpot is running the same play as a series. The Hustle in 2021. Starter Story in february 2026.

that’s reach. rooms entered.

the serial acquirers went a level higher.

Constellation bought Keypoint Intelligence in may 2026. not a tool that operates in the digital-imaging vertical: the benchmark authority the vertical measures itself against.

Beacon paid eight figures for RealEstateAPI in march: bootstrapped, property data on 150 million american homes, three hundred customers. not a tool that operates in a market. the data layer under one. and Beacon’s pitch to founders says it plainly: as AI turns competence into a commodity, the scarce asset left is trust. trust is distribution by its human name. what took Constellation three decades, venture capital is trying to manufacture inside a single fund cycle.

the buyer of the software is advertising the data.

that’s sight. not rooms entered. rooms understood.

and the tape is longer than the AI cycle. Monsanto paid a billion for Climate Corp in 2013. Moody’s paid two billion for RMS in 2021. and Verisk, an insurer data consortium that grew into a listed company, is the terminal form: the panel outlived every tool around it.

two decades of tools. five years of reach. now sight, with deals landing months apart. the intervals shorten. the shelf doesn’t restock. most verticals never built a Keypoint. behind most of the economy there is nothing to acquire.

the tape terminates in a shortage.

what remains scarce?

General Catalyst committed $1.5 billion of a single fund to AI rollups. Thrive is raising a permanent vehicle north of a billion. Beacon has raised over half a billion across three rounds. Metropolis took another $1.6 billion in november 2025. J.P. Morgan arranging a billion-dollar term loan for a parking startup.

Peter Thiel wrote in 2014: “superior distribution alone can create a monopoly, even without a better product. a better product alone cannot.”

the capital knows. the buyers know. the question is what they’re buying toward.

the tools are free. the operations are commodity. ownership terminates at the value of the work being done.

what remains scarce is what the buyers on this page are already moving toward. not the software. not the operation. the informational advantage the distribution creates. who is churning, who is growing, whose margins are compressing, where the breaches are landing.

the smartest operators will stop monetizing the user entirely, because friction slows distribution, and distribution is the sensor grid. the user isn’t the customer. the user is the instrument.

that is the third level. part 2/4 names it precisely.

the full thesis is at predictioneconomy.ai

part 2/4: tthe mechanism that generates economy-scale signal, and the one kind of firm everyone has been mispricing

Subscribe now

this is the follow-up. because once you understand who pays, the next question is what you actually capture. and that answer has changed twice in the last decade, with a third shift already happening, quietly, at the companies operating closest to the frontier.

three levels of value capture. each one collapses the level below.

level one: sell the tool

this is classical SaaS. you build software. you charge a subscription. the software creates value for the customer, saves them time, finds them better deals, closes their books faster, and you capture a slice of that value in monthly fees.

the honest math on the slice: 2 to 5%. if a leadgen tool saves a sales org 30% of their time and closes 20% more deals, the customer keeps 95% of the value created. you get the subscription.

that was fine when software was expensive to build and hard to replace. neither is true anymore. the customer re-evaluates you every renewal cycle. AI collapses the rebuild cost. sam lessin has written about this: LTVs are finite. code depreciates.

selling the tool is a shrinking share of a shrinking pie.

level two: sell the outcome

this is what the AI rollup thesis is actually about, whether operators name it that way or not.

same software. but you don’t sell it. you run it yourself and sell the result.

the leadgen tool doesn’t get licensed to the sales org. it runs internally and you sell qualified meetings at €750 each. the tax software doesn’t get sold to CPAs. you run the CPA firm and sell filed returns. the property management system doesn’t get licensed to property managers. you become the property manager.

value capture jumps from 2-5% to 30-60%. not because you built better software. because the customer can no longer comparison-shop you against another SaaS vendor. they’d have to replace the whole function. your margin is your secret.

this is why AI rollups work. service pricing on software cost structure.

Ralph Winter, real estate investor and one of the sharpest people in this space, put it plainly last month. the property managers didn't want software. they wanted the problem solved.

but here’s the part most operators miss. L2 is not the endpoint. it’s the foundation.

level three: the prediction economy

this is where it gets interesting. and where i part ways with the people who built the frameworks i’ve been citing.

Sam Lessin’s answer to what comes after SaaS is ownership. use software to run real businesses better, then buy them. metropolis didn’t sell parking software. it bought SP+ and kept the entire uplift.

i think ownership is the right capture structure. i don’t think it’s a level.

buying the business you operate better is L2 with better economics. you’re still monetizing the same thing. the operation. equity instead of fees changes how much of the value you keep. it doesn’t change what the value is.

the third level is the asset that only exists at scale: data you gather through building access through your distribution network. this all ties back to distribution becoming the only moat, not because of the size it gives you, but the access to behavioral data that allows you to anticipate market shifts and through that leverages your next move of ownership.

and the mechanism that creates it has a name. the multiplier.

a multiplier is a firm that serves other firms. one tax advisor is one customer. but behind one tax advisor sit dozens of SMEs. behind 1,500 advisors sits the bookkeeping reality of a meaningful share of a country’s economy. you don’t build reach to the network. you build reach through it. that’s the lever that turns a 1,500-node network into an economy-scale instrument. and it’s why the vertical selection rule is not fragmentation alone. it’s fragmentation times end-customer leverage.

run the platform layer across a vertical’s multipliers and you see something that exists nowhere else. the evolution of an entire SMB economy, in near-real time. which sectors are heating up. where margins compress. where cash cycles stretch, months before any published statistic. pattern-level movement of a whole market.

the tool depreciates. AI rebuilds it in weeks. the operation scales. a funded competitor replicates it in years. the panel is time-locked. no competitor can replicate it without replicating your entire network history. and history is the one thing money can’t compress. that’s the exact reason i keep saying distribution is the only moat.

the large players have already figured this out. as a product. in may 2026, constellation software acquired keypoint intelligence: the benchmark authority of the digital-imaging industry, the lab everyone in that vertical cites. in march 2026, beacon software paid eight figures for RealEstateAPI, a bootstrapped property-data business with 300+ paying customers. the buyers of vertical software are already buying the information authority of verticals. the sellable form of a signal layer has a name and a precedent. the benchmark business with paying subscribers.

that's the long-term moat everyone is searching for. call it the prediction economy: the phase where the winners aren't the ones who automate the work, but the ones who see the market through the work.

two paths in level three

once the panel is running, the strategic question is what you do with it.

path one is mid-term cash. sell the signal. anonymized, aggregated, structured for buyers who can’t build it themselves. hedge funds paying for pre-print macro signal. insurers pricing SME credit risk. banks calibrating loan books. this is the intuit small business index model with paying subscribers instead of free thought leadership. clean revenue. no operational complexity. no conflict-of-interest exposure because you’re never a counterparty to your customer base.

the tax platform sells the SMB cash-cycle feed to three hedge funds and two macro desks. eight-figure annual revenue on a network the operating business paid to build. the panel becomes a subscription business layered on top of the service business. the acquirer, when it comes, buys a rollup with a proprietary data product bolted on. that’s real value.

path two is long-term ownership. keep the signal internal. use it to make your own investment and expansion decisions. which SMEs to acquire before their bank does. which sub-verticals to build into next. which geographies are heating up before the competition sees the shift. the panel becomes an origination engine for your own capital deployment.

the tax platform sees which regional accounting practices are quietly consolidating clients and buys them at multiples the market hasn’t priced yet. it sees which SMB segments are structurally growing and expands its multiplier network toward them first. the panel doesn’t get sold. the panel gets deployed.

the difference is what game you’re playing.

path one is a good business. clean, defensible, predictable revenue. the ceiling is set by what data buyers will pay per year, and the exit is an information-authority acquisition like keypoint or realestateAPI. good outcome. bounded outcome.

path two is a compounding position. every insight from the panel makes the next capital allocation sharper. every acquisition tightens the network. the ceiling is set by the size of the vertical you’re building through, not by what a hedge fund will pay for a data feed. bigger upside. bigger execution risk. requires capital you may not have and patience the board may not want.

most operators can only run one of these. the discipline is picking which and building the whole structure around it. path one shapes contract terms, licensing architecture, GDPR posture, customer disclosure. path two shapes capital structure, board composition, deal pipeline, retention economics.

the mistake is trying to keep both open. you can sell the signal or use the signal. running both is how you end up with hedge fund contracts that block your own acquisition activity because your customers realized you’re both selling their aggregate signal and using it to buy their competitors.

pick the game.

what to build

three revenue layers stacked, one asset that appreciates.

L1 spreads the network. the free 10x feature that gets you through the door to the multipliers.

L2 runs the operation. the service business that makes the network real and generates the panel underneath it.

L3 is where the decision lives. sell the signal for clean recurring revenue. or use the signal to compound your own ownership. one is a good business with a bounded ceiling. the other is a compounding position with a bigger ceiling and a harder execution.

pick the level three you’re actually playing. build the whole structure around it.

that's the whole play.

Subscribe now

based on conversations with AI rollup operators, property platform founders, and the ongoing question of what actually gets captured when software becomes infrastructure.

they clean your house for free. no subscription. no upsell. no fine print. you book them, they show up, they clean, they leave. no invoice.

the twist is that every cleaner wears a camera. every cleaning session is filmed. every corner, every surface, every motion.

their bet: in five years, humanoid robots will do household cleaning. the company that owns the most video data of humans actually cleaning real homes will train the best robots. and the robotics companies will pay a fortune for that data. more than any human customer would ever pay for the cleaning service itself.

so they don’t monetize the cleaning. they monetize the data.

the cleaner they send to your house is not the product. the video of that cleaner is the product. and you are not the customer. the robotics company in three years is the customer.

they play free with everyone whose money isn’t the real money.

that’s the play most AI rollup operators are missing.

your distribution network is not your customer

i had a call this week with a founder building an AI-native tax advisory platform in austria. serious operator. real product. income and expenditure calculation already at 90% automation. tax returns generated and filed automatically. the software works.

but they’re stuck in the same trap most AI-native companies are in. they’re trying to monetize the tax advisors they onboard. build a great product. charge a monthly fee. grow organically.

that math doesn’t work for the exit game they’re actually playing.

here’s the reality: austria has 3,000 tax advisors. even if this founder onboards one advisor a week - already aggressive - they’ll hit 100 tax advisors in two years. no acquirer buys a distribution network of 100 advisors in a market of 3,000. that’s 3% penetration. not a network. a foothold.

the acquirer they need - ETL, or one of the german or american CPA rollups - will only pay for meaningful distribution. 1,000 tax advisors. 1,500. the majority of the market. anything less is a nice acqui-hire, not a strategic acquisition in a small market.

so the question isn’t “how do we build a better tax advisory product.” the question is “how do we onboard 100 tax advisors a month for the next 12 months.”

and you cannot do that if the tax advisor is the one paying you.

because the moment the tax advisor has to pay, they slow down. they compare. they negotiate. they wait for budget approval. they ask their partners. they miss your emails. the whole flywheel jams.

the tax advisor is not your customer. the tax advisor is your distribution.

find the third party

the cleaning company found robotics. what’s the tax advisor equivalent?

think about what a network of 1,500 austrian tax advisors actually generates: real-time financial data on every SME in the country. bottom-up macro signals. industry-level benchmarks. cash flow patterns across the entire small business economy. anonymized, in aggregate, this is a data set that literally does not exist anywhere else.

who wants that data?

hedge funds. macro analysts. banks running SME credit strategies. insurers pricing small business risk. anyone trying to see which sectors are heating up months before it shows in public data. funds already pay millions a year for alternative data feeds: credit card panels, app usage, satellite imagery. a real-time signal on an entire SME economy is the same category. it doesn’t exist yet.

that kind of feed is a rounding error in fund economics.

but here’s the part most founders get wrong. and it’s the part that ends careers.

the data doesn’t belong to the tax advisor. it belongs to the entrepreneur. austrian tax advisors are bound by professional secrecy. they can’t pipe client financials anywhere.

so the entrepreneur has to be in the room. not a hidden data source. a user. they get the free bookkeeping experience, and in exchange they opt in to their data feeding anonymized, aggregated statistics.

a few million a year from data buyers covers your entire operation. software free for tax advisors. software free for entrepreneurs. every friction point that slowed onboarding gone.

the tax advisor gets the 10x feature for free. the entrepreneur gets a better bookkeeping experience for free and opts in on the way through. the data buyers pay for a macro signal only they know how to extract value from.

three parties. one pays. two ride for free. distribution builds at maximum velocity.

big fish in a small pond

here’s the part nobody wants to say out loud: if you’re in a small market, you’re playing an exit game. and that’s fine.

austria has 3000 tax advisors. even a dominant AI-native tax advisory platform there won’t IPO in austria. it won’t be a decacorn. the ceiling is real. everyone in the ecosystem knows it.

what most founders in this position do is pretend they’re playing the global game anyway. they raise VC money. they pitch expansion into germany and switzerland. they position for a story their market can’t support. and then they burn 18 months trying to prove the unprovable before quietly pivoting to a strategic sale, because the market is too saturated in other regions.

the smarter move is to know from day one what game you’re in. play it clean. optimize for the acquirer you know is coming.

for the austrian tax advisor play, the acquirer is a german or american CPA rollup that wants to enter the austrian market and doesn’t want to build the distribution from scratch. they’ll pay for the network. they won’t pay for the software - they have their own. they won’t pay for the customer relationships - they’ll integrate them into their existing platform.

so build for that. don’t over-invest in software the acquirer will replace anyway. don’t over-invest in features the acquirer won’t use.

optimize for the one thing the acquirer will pay for: distribution and access to your verticle.

the franchise model

here’s where it gets interesting.

if the monetization is the data buyer and not the tax advisor, then this exact play works in every small market that has SMEs, tax advisors, and no strong incumbent AI-native platform.

lithuania. bulgaria. slovenia. estonia. every small european market that’s too small to attract global VC attention but big enough to have a real SME economy.

in each of those markets, the same three-party model works. tax advisors onboard for free. entrepreneurs use the software for free and opt in at the door. the data buyers, possibly the same funds and banks across countries, pay for the aggregate signal.

you build a franchise of distribution networks. austria. lithuania. bulgaria. each one is a 40 million euro shareholder value creation event on its own. bundled together, you’re a 400 to 500 million euro platform play. and the acquirer is no longer a german CPA rollup buying one country - it’s a global platform paying for entry into ten markets at once.

that’s the shift. from playing a small game well to running the same well-played small game ten times in parallel.

know your game

most AI rollup operators are optimizing for the wrong thing. they’re building product for people they think are their customers, when their real customers are somewhere else entirely.

if you’re in a market where you’ll be acquired for your distribution, the tax advisor is not your customer. if you’re in a market where robotics is the future, the household is not your customer. if you’re in a market where the underlying data is the asset, the person generating the data is not your customer.

find the third party who benefits from what you’re building in a way no one else does. price them. let everyone else ride for free.

your distribution accelerates. your economics get cleaner. your acquirer sees a network that looks impossible to replicate.

play the game you’re actually in. price the party who’s actually paying. and build the distribution nobody else can catch.

that’s the whole play.

^{"Everything we see hides another thing. We always want to see what is hidden by what we see."}

based on conversations with AI rollup operators, an AI-native tax advisory founder in vienna, and a cleaning company in new york i keep thinking about.

The variable is automation grade. How much of the work in a given vertical can AI actually do. That determines whether you should grow organically, roll up, or stay away entirely. The article never applies this lens. That’s the gap.

Here’s where I agree, where I don’t, and why.

Simpler than it sounds

“Layer AI on top” is not the case. You start with a Platform Company that you know is tech receptive, and integrate custom workflow automation into their processes. The biggest waste of tokens is people building software but no one using it. AI Rollups, if led well, generally avoid that. The best case I’ve seen are General Catalyst portfolio companies. They track tech adoption and how well it works before enabling the next tranche to buy and scale.

Two different strategies get lumped together in this kind of discussion. One aggregates assets primarily for multiple arbitrage - buying at low multiples, bundling, selling the aggregate higher, with limited operational integration. The other integrates acquisitions around a platform company with shared technology, processes, and culture, where the value comes from operational improvement rather than the multiple spread. The article applies critiques of the first to argue against the second.

One point the article gets right: organic GTM after inorganic growth is hard.

“Customer acquisition” flattens the thesis

Distribution is the moat. Full stop. But that is not the only reason an AI rollup acquires.

You buy the customer segment. The market trust. The standing processes. The fulfillment. The decades of operational data that informs what you build next. Not just the customers.

The spectrum no one talks about

Below 30% automation: you can’t underwrite the value creation. The AI transformation isn’t deep enough. The margin improvement is too thin to compound.

30 to 70% automation: this is where rollups make sense. You’re buying companies with real operational substance that AI can meaningfully transform but not fully replace. CPAs. MSPs. Property managers. Call centers. You need the people, the processes, the structure. You also need AI to make them dramatically more efficient. The CPA isn’t dead weight just because she’s not AI-native. The magic is enabling her to do 10x the work she could do alone.

A niche case I just listened to: tree trimmers across America. Unbelievably fragmented market with 30% achievable automation and revenue expansion through AI. Apparently many deals get lost because people in that business regularly miss checking their emails for offers. That’s the kind of gap rollup-plus-AI was built for. Above 70% automation: rollups get nervous. Customers can switch easily. You’re buying expensive headcount you’re about to automate away. This is where Emergence’s thesis holds.

But it holds here. Not everywhere.

Harper Insure, the AI-native insurance broker in the blog’s own example, has over 80% automation. They don’t need to roll up. More inbound than they can service. Correct. That’s what above-70% looks like. But generalizing from that to “rollups are mostly wrong” ignores the entire band where the economics work differently.

The multiple question

I’ve thought about this a lot.

The multiple expansion only works if you actually do the AI transformation. Constellation, TransDigm, Danaher trade at 20-25x EBITDA because they demonstrably improve the companies they buy.

If we just aggregate assets, we deserve a PE multiple. If we genuinely double free cash flow through AI in eighteen months across a portfolio, we deserve something closer to a software multiple.

The risk looks like venture. The return profile looks like venture. The vehicle just happens to be a balance sheet instead of pure equity.

And even in the downside case, where multiples compress and the software-adjacent re-rating never comes, the AI transformation still increased the cash flow of the underlying business. You make a return on the cash flow alone. That’s the floor. Venture upside, PE downside protection.

Where the framework meets a real case

I talked to a founder this week. AI native management consulting. Automates change management and business transformation. Currently at 50% automation, heading to 90%. Off-the-shelf SaaS.

A holding company offered him $5M to acquire a consulting firm.

Wrong play. I agree. But not because rollups are bad. Because this specific founder has the wrong profile for it.

Automation level. He’s at 50% now, heading to 90%. If you can automate 80%+ of the process, acquiring consulting firms means buying expensive headcount you’re about to automate away. Organic growth is cheaper and faster.

No confirmed PMF. Still fundraising his seed. Inorganic growth before PMF is burning capital without knowing if the core technology holds. If the product doesn’t stick, you’ve just bought expensive distribution for nothing.

SaaS vs. custom workflows. AI rollups require proprietary, custom automation for each acquired company. “Stephanie” from accounting has done it one way her whole life. You don’t hand her an off-the-shelf tool. You build something that reads her current systems and automates based on how she’s actually been doing it. Beacon built several proprietary technologies in roughly a year. That’s how deep it goes. An off-the-shelf product and a rollup strategy pull in opposite directions.

Where the article breaks

The customer isn’t paying more. The article spins it into ‘then what’s the customer paying more for.’ That’s a category error. The customer is paying the same or less. We’re delivering it at a fraction of the cost. The delta is the margin. That’s the whole thesis.

The AI doesn’t have to be visible to the customer for the unit economics to flip. It has to be invisible enough that the customer doesn’t churn and effective enough that we double free cash flow in eighteen months. Those are the two metrics.

The blogs ‘that customers won’t switch unless the AI is visibly better’ framing assumes we’re trying to win on visible AI features. We’re not. We’re trying to win on outcome and price. The customer experiences “this works better and costs less.” Not “wow, this is AI-native.”

Apples to oranges. Of course a team of 10 is cheaper than a team of 150. But you can’t automate 100% of what an MSP does. You need technicians to fix the router and people checking how well tickets have been sorted. The degree of automation possible in the vertical determines the strategy. Not some universal preference for organic growth.

The pre-PMF pivot. The article argues that if you need a rollup for distribution, your product isn’t good enough. That’s an argument against pre-PMF companies using rollups as a distribution shortcut. That was never the idea behind AI Rollups. You don’t create a new SaaS and roll up an industry. You build on top of their processes. The article is arguing against a version of AI rollups that serious operators aren’t running.

What actually holds customers

The article treats relationship stickiness as the primary retention mechanism and then argues it’s weaker than it looks.

But relationship stickiness isn’t even the primary mechanism.

What holds customers after acquisition, in order:

→ Processes get better, issues resolved faster → Seniority of combined group and knowledge → Greater options and support for customer → Existing relationship

Outcome improvement is the moat. Relationships are fourth.

That’s why earn outs exist. Not the solution to all problems. But they bridge the transition while outcome improvement takes hold.

Per outcome pricing is the future. But you don’t start there. You start with the existing contracts and improve delivery within them.

And “organic then inorganic” is not a universal sequence. It depends on the vertical, the automation grade, and whether a strong platform company exists to build on.

The data exists

The article frames ‘$50M+ ARR, retention north of 90% on the acquired book, margins expanding rather than capped’ like nobody has it.

General Catalyst has it.They published the playbook. Crescendo acquired PartnerHero and is rolling out AI customer service to its 200 customers, taking gross margins to 60-65%. Long Lake has acquired 18 businesses with 25-30% productivity gains in HOA Management and a 10x increase in new customer pipeline. Dwelly acquired 6 property management agencies and doubled EBITDA margins where the tech is fully deployed. Titan acquired RFA and reduced user onboarding from weeks to minutes.

That’s not a thesis. Those are transactions that closed.

The data exists. The question isn’t whether AI rollups can work. It’s whether you’re running the right playbook for the right vertical at the right automation grade.

The founder I talked to this week made the right call. He read his position on the spectrum correctly. High automation trajectory. Off-the-shelf product. No PMF. He turned down $5M because the framework told him to.

In that case, Emergence Capital’s argument would land right - for the wrong reasons. The blog argues not to roll up because rollups are mostly wrong. The actual answer is don’t roll up because your automation grade, product architecture, and PMF status put you above the 70% line.

One is ideology. The other is diagnosis.

Based on conversations with AI rollup operators, PE advisors, and founders across the automation spectrum. Special thanks to Kate Bender, Partner at General Catalyst, for the exchange.

Companies used to acquire for technology. For talent. For IP. Now, increasingly, they’re acquiring for one thing: the user base.

Clay bought blogs. HubSpot bought newsletters. PE firms are rolling up accounting practices, MSPs, and property managers - not for their systems, but for their client relationships. The pattern is quiet, but it’s accelerating.

The question worth asking: who exactly is writing these checks, and what does that reveal about where value actually lives?

Tier 1: AI Rollup Platforms

General Catalyst has committed $1.5 billion to incubating AI-native companies in specific verticals, then using those companies as acquisition vehicles to buy established firms, and their customers, in the same sectors.

The key phrase is “and their customers.”

Thrive Capital, Lightspeed, and 8VC are running similar playbooks. They’ve built the AI. They’ve raised the capital. What they need is distribution, fast.

These platforms aren’t acquiring for technology - the acquired company’s tech stack usually gets retired. They’re not acquiring for talent - most of the team churns within 18 months. They’re acquiring for one thing: access to a validated user base in a specific vertical.

The economics are stark. A platform like Crete or Shield Technology Partners might pay €1.5M for a 10-person accounting firm. After integration, what survives? The client list. Everything else is overhead.

Tier 2: PE-Backed Buy-and-Build

In Q3 2025, 78% of PE deals with AI targets were add-ons - reflecting a broad preference for creating value through buy-and-build strategies.

These are mid-market PE firms running vertical software rollups or services consolidation. They’ve acquired 3-7 companies, they’re trying to cross-sell across the combined customer base, and they’re hitting the ceiling on organic growth.

The pattern is the same. They acquire companies primarily to get access to the user base. The product gets consolidated. The team gets rationalized. The customers, if they stay, are the asset that justifies the multiple.

The pain point here is acute: integration takes longer than expected, user churn during migration is higher than modeled, and the cost per retained customer often exceeds what the original acquisition math assumed.

Tier 3: Vertical SaaS at the Inflection Point

The third buyer category is vertical SaaS companies at the $5-30M ARR inflection point.

These are companies that have hit product-market fit in one segment. The board is pushing them to expand into adjacent verticals or geographies. Their options are limited: build a sales team (12+ months to productivity), acquire a competitor (€1-5M plus integration risk), or find another path.

The companies most actively seeking distribution are those in concentrated enterprise workflows: construction, logistics, legal ops, healthcare admin, compliance. The user base is identifiable. The ICP is specific. And organic growth in new segments is painfully slow.

Who’s Not Buying

A few categories are notably absent from this trend.

Early-stage startups don’t have the budget or the board pressure. They’re still figuring out product-market fit.

Horizontal SaaS companies: Slack, Notion, broad-use tools grow through network effects and product-led growth. They don’t need to acquire vertical-specific user bases.

Companies where the product is the moat. If competitive advantage comes from proprietary technology that users can’t get elsewhere, distribution takes care of itself. The buyer of distribution is the company whose product is good but not so differentiated that users come automatically.

The Behavioral Pattern

The clearest signal isn’t firmographic: it’s behavioral.

The companies most actively seeking distribution are those that have acquired at least one business in the last 18 months where the user base was the primary value driver, and where integration took longer than expected or resulted in meaningful user churn.

They’ve felt the pain of buying an entire company just to get the customers. They’ve lived through the integration debt, the migration churn, the slow realization that the asset they paid for was leaking value from day one.

These buyers know what distribution costs through traditional M&A. They know how long it takes. And they’re looking for a faster path.

What This Reveals

The shift toward acquiring for distribution rather than technology tells us something important about where value lives in the current market.

Product is increasingly commoditized. AI is making it easier for anyone to build what you built. The defensible asset isn’t the software - it’s the relationship with a specific user segment.

The companies consolidating markets right now understand this. They’re not paying for code. They’re paying for access.

And the gap between what they’re paying and what that access is actually worth is where the next wave of value creation will happen