In brief
There's a number in Euclid Ventures' Vertical Report 2026 that deserves more attention than it got.
There's a number in Euclid Ventures' Vertical Report 2026 that deserves more attention than it got.
Across 4,395 VC financings worth $186 billion in 2025, vertical companies took 53% of deal volume — 2,331 deals — and $56.2 billion of capital. By Q4 that share hit 55%, the year's high.
But the number that actually matters is the one further down. In the $1–5 million tier — pre-seed and seed, where new companies get formed — vertical share went from 53% in Q1 to 60% in Q4.
That's not a market shifting. That's founders making a different decision about what to build.
What the shift is actually about
For fifteen years the default advice was to build horizontally: pick a function that every company needs, sell it to all of them, let the total addressable market do the work in the pitch deck.
Vertical software was the less glamorous choice. Smaller market, harder sales, industry knowledge required, and a ceiling you could calculate on a napkin.
Three things inverted that.
Generative AI collapsed the cost of building the boring parts. The document parsing, the classification, the routing, the summarising — the unglamorous middle of most industry workflows — stopped being years of engineering. What's left as the hard part is knowing which workflow matters, which is industry knowledge, not engineering.
The report has a striking data point on this: in Manufacturing & Industrial, 57% of founders started their companies after 2022. That's not existing companies pivoting. That's net-new company formation caused by the tools getting cheap enough that people with domain knowledge could act on it.
Horizontal AI turned into a commodity fast. If your product is a good interface to a general model, so is everyone else's, and the model provider can absorb you at any time. Depth in one industry is one of the few defensible positions left.
The buyers changed. A shop owner, a contractor, a clinic manager doesn't want a platform to configure. He wants his specific problem gone. Vertical products win that comparison every time, because they can afford assumptions a horizontal product can't.
Where the money went
The 2025 breakdown is worth reading as a map of what's considered solved and what isn't:
| Vertical | Deals | Capital |
|---|---|---|
| Healthcare | 597 | $11.8B |
| Financial Services | 501 | $11.7B |
| Manufacturing & Industrial | 110 | $8.6B |
| Legal | 97 | $3.3B |
| AEC & Trades | 106 | $2.1B |
| Energy & Climate | 60 | $965M |
Healthcare and financial services dominate on volume, which is unsurprising — they've absorbed capital for a decade.
The interesting rows are lower. Legal took $3.3B across 97 deals, including Clio at $850M and Filevine at $400M. Law firms are not famous for technological adventurousness. They got digitised anyway, and quickly, once the products stopped being generic and started being about the specific way legal work happens.
AEC & Trades — architecture, engineering, construction, and the trades — took $2.1B, and produced CompanyCam's $415M round at a $2 billion valuation. For an industry that ran on paper and phone calls until recently, that's a fast repricing.
Manufacturing grew 41% quarter over quarter.
Notice which industries these are. Not the ones with the most sophisticated buyers. The ones with the least sophisticated existing software — where the incumbent isn't a competitor's product, it's a spreadsheet, a WhatsApp group, and somebody's memory.
What investors say they're buying
Euclid's read on investor priorities: category leadership and founder-market fit. General Catalyst led 24 vertical deals ($1.8B), a16z 22 ($1.9B).
"Founder-market fit" is one of those phrases that means nothing until you look at what actually gets funded. In these categories it means something specific and slightly uncomfortable for traditional startup pattern-matching: the founder having worked in the industry beats the founder having worked at a scaled tech company.
The report describes the trades opportunity as "fragmented end markets, low technology adoption, high workflow complexity."
Read that as a description of what it takes to sell there. Fragmented means no big logos to land — you need thousands of small customers and a way to reach them cheaply. Low adoption means your competition is a paper folder, and the customer has been burned before. High complexity means you can't learn the workflow from an interview; you have to have lived in it.
That combination is why generalist founders lose these markets and why they stayed unbuilt so long.
The vertical nobody's finished
Auto repair sits in this exact quadrant and hasn't had its repricing yet.
In the US alone there are more than 230,000 independent shops. The 50 largest companies in the industry generate less than 10% of total revenue, and independents hold roughly 70% of the market. Nobody has consolidated it; nobody has digitised it.
The software that exists is a digital logbook: work order, status, note. It assumes a mechanic will stop mid-repair to type. He doesn't, which is why so much of this software gets bought and abandoned, and why the actual operating system of most shops remains a whiteboard and a group chat.
That's the problem I've spent three years on, from the operator side — I own a shop network and built MECH Orbit inside it, around the idea that the physical event should produce the record rather than a person being asked to. Car enters the bay, camera reads the plate, work order opens. Mechanic photographs and dictates from his phone. Customer approves specific lines, with the time and scope stored.
I'm not neutral about this category. But the market data isn't mine, and it says the same thing: the money is moving toward industries where the incumbent is memory.
What this means if you're building
If you're deciding what to start right now, the numbers argue for something counterintuitive.
Pick a smaller market you actually know. The $1–5M tier moved eight points toward vertical in a single year because that's where the return on domain knowledge is highest, and because AI removed the excuse that industry-specific software takes too long to build.
The uncomfortable corollary: if you don't know an industry deeply, this advantage isn't available to you, and the horizontal alternative is getting commoditised by the month.
Which is a strange and rather good moment for the people who've been running unglamorous businesses this whole time and assumed the software revolution wasn't about them.
Sources
What to check in your workshop
- Verify the primary source and date behind every figure
- Separate reported facts from the author's interpretation
- Compare the conclusion with data from your own workshop