In August I wrote a thesis I never published. The funds I was warning were key Crossover Research clients, so I stayed quiet. Since then, ๐ฆ๐ผ๐ณ๐๐๐ฎ๐ฟ๐ฒ ๐บ๐๐น๐๐ถ๐ฝ๐น๐ฒ๐ ๐ฎ๐ฟ๐ฒ ๐ฑ๐ผ๐๐ป ๐ฑ๐ฌ%+. Salesforce $CRM, ServiceNow $NOW, Adobe $ADBE, Workday $WDAY all off 40% from highs. Thomson Reuters $TRI dropped 16% in a single session on the Anthropic legal agent launch. The SaaSpocalypse arrived. So here's the follow-up. Not commentary on what happened, but where I think this goes next.
Most vertical SaaS companies aren't underperforming because their software is bad. ๐ง๐ต๐ฒ๐'๐ฟ๐ฒ ๐๐ป๐ฑ๐ฒ๐ฟ๐ฝ๐ฒ๐ฟ๐ณ๐ผ๐ฟ๐บ๐ถ๐ป๐ด ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐๐ต๐ฒ๐ ๐ป๐ฒ๐๐ฒ๐ฟ ๐ฏ๐๐ถ๐น๐ ๐๐ต๐ฒ ๐๐ฒ๐ฐ๐ผ๐ป๐ฑ ๐ฏ๐๐๐ถ๐ป๐ฒ๐๐. And the first business is under attack. For twenty years, one of the biggest SaaS moats was engineering complexity: deep technical talent, long roadmaps, compounding codebases that were genuinely hard to replicate. ๐๐ ๐๐ฝ๐ฒ๐ป๐ฑ๐ฒ๐ฑ ๐๐ต๐ฎ๐ ๐ฎ๐น๐บ๐ผ๐๐ ๐ผ๐๐ฒ๐ฟ๐ป๐ถ๐ด๐ต๐.
Product development is democratizing to operators with no code background but strong product vision. Look at Anthropic: they've built the engine and are shipping lookalike products at a cadence that would have taken a legacy SaaS vendor three years of roadmap, with a fraction of the headcount. That pace can kill legacy businesses overnight.
๐๐ณ ๐๐ต๐ฒ ๐ฒ๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ถ๐ป๐ด ๐บ๐ผ๐ฎ๐ ๐ถ๐ ๐ด๐ผ๐ป๐ฒ, ๐ณ๐ผ๐๐ฟ ๐บ๐ผ๐ฎ๐๐ ๐ฟ๐ฒ๐บ๐ฎ๐ถ๐ป: ๐ฑ๐ถ๐๐๐ฟ๐ถ๐ฏ๐๐๐ถ๐ผ๐ป, ๐ฝ๐ฟ๐ผ๐ฝ๐ฟ๐ถ๐ฒ๐๐ฎ๐ฟ๐ ๐ฑ๐ฎ๐๐ฎ, ๐๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐ ๐ฏ๐ฟ๐ฒ๐ฎ๐ฑ๐๐ต, ๐ฎ๐ป๐ฑ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐ถ๐ป๐๐๐น๐ฎ๐๐ถ๐ผ๐ป. The first three are moats the company builds. The fourth is a moat the company captures, and it's the one most resistant to AI disruption.
๐ฅ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐ฐ๐ผ๐บ๐ฝ๐น๐ฒ๐ ๐ถ๐๐ ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ๐ ๐๐๐ถ๐๐ฐ๐ต๐ถ๐ป๐ด ๐ฐ๐ผ๐๐๐ ๐๐ต๐ฎ๐ ๐ต๐ฎ๐๐ฒ ๐ป๐ผ๐๐ต๐ถ๐ป๐ด ๐๐ผ ๐ฑ๐ผ ๐๐ถ๐๐ต ๐ฝ๐ฟ๏ฟฝ๏ฟฝ๐ฑ๐๐ฐ๐ ๐พ๐๐ฎ๐น๐ถ๐๐. Once a vendor is embedded in a compliance workflow, ripping them out means re-attesting, re-auditing, and re-certifying every downstream process. The buyer isn't paying for software, they're paying for the accumulated paper trail. Tyler Technologies ($TYL) is the clearest version of the pattern. State and local government software across courts, public safety, assessment, and ERP. Every module is married to statutory process, FIPS, CJIS, audit trails, and procurement cycles that take years. TYL is down 42% TTM and 2026 guidance came in soft, but the moat didn't break. Revenue still compounded, and government procurement runs on five-year cycles, not five-week news cycles. Veeva is the sharper version. Revenue up 16% in FY26, Q4 beat, the stock still down 25%. The market is selling execution, not weakness. Guidewire in P&C insurance, where regulatory filings and rate approvals anchor the stack, sits in the same setup: still compounding ARR, still winning cloud conversions, multiple reset anyway. Same pattern across all three: multiples compressed, fundamentals intact. The moat is the regulatory surface area itself, and it compounds because the rules get more complex, not less.
๐ ๐๐ฎ๐ ๐น๐ผ๐ป๐ด ๐ฃ๐ฎ๐น๐ฎ๐ป๐๐ถ๐ฟ ๐ฎ๐ $๐ญ๐ฏ (read that here: https://t.co/0N0oIX8N87). ๐ก๐ผ๐ ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐๐ต๐ฒ ๐บ๐ผ๐ฑ๐ฒ๐น ๐ผ๐ฟ ๐๐ต๐ฒ ๐๐ผ๐ผ๐น๐ถ๐ป๐ด. ๐๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐๐ต๐ฒ ๐ผ๐ป๐๐ผ๐น๐ผ๐ด๐. Palantir is the proprietary-data version of the regulatory thesis. Once Palantir sits between the customer and their own data, ripping it out means rebuilding the data model from scratch. Snowflake and Databricks never had that entrenchment layer. AIP bootcamps then turned the data moat into a distribution moat: 660 bootcamps in a single quarter, 94% y/y US customer deal growth, bookings at 1.9x sales. Own the data, ship functional AI on top of it, let the GTM compound. Every vertical incumbent has a version of this available. The question is whether they'll build it before a challenger does.
But regulatory insulation is necessary, not sufficient. Plenty of vendors inside regulated verticals are still getting squeezed because they never became AI-native. BlackLine ($BL) and Trintech are feeling it in close and reconciliation as Numeric, Maximor, and Stacks build AI-native from day one. nCino ($NCNO) in banking faces the same challenge. The regulatory moat buys you time. It doesn't buy you the decade.
๐ง๐ต๐ฒ ๐๐ถ๐ป๐ป๐ถ๐ป๐ด ๐ณ๐ผ๐ฟ๐บ๐๐น๐ฎ ๐ถ๐ ๐ฑ๐ฎ๐๐ฎ ๐ผ๐ฟ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐๐๐ฟ๐ณ๐ฎ๐ฐ๐ฒ ๐ฎ๐ฟ๐ฒ๐ฎ ๐ฝ๐น๐๐ ๐ณ๐๐ป๐ฐ๐๐ถ๐ผ๐ป๐ฎ๐น ๐๐, ๐ป๐ผ๐ ๐ผ๐ป๐ฒ ๐ผ๐ฟ ๐๐ต๐ฒ ๐ผ๐๐ต๐ฒ๐ฟ. Look at why Claude is winning. Anthropic isn't competing on model benchmarks, they're competing on functional workflow. Building for the user, not the leaderboard. That's the playbook vertical incumbents need to run. Take the moat you already have, whether it's regulatory or data-entrenchment, layer genuine workflow AI on top, and the challenger can't catch you. The vendors that do both win the decade. The ones that rely on inertia alone get caught. The ones that ship AI without an anchor get commoditized. You need both.
๐ง๐ต๐ฒ ๐ฏ๐๐๐ฒ๐ฟ ๐ถ๐ ๐๐ฒ๐น๐น๐ถ๐ป๐ด ๐๐ผ๐ ๐๐ต๐ถ๐ ๐ฝ๐น๐ฎ๐ถ๐ป๐น๐. A study we ran with Battery Ventures on AI adoption in the Office of the CFO (https://t.co/xBEMSF8Y72) surveyed 129 finance leaders at companies from $50M to $5B+ in revenue. 77% said they want to uplevel existing systems with AI from new vendors that layer onto existing systems. Only 15% want to replace their current system of record with an AI-native platform. The incumbent wins if they ship AI. The AI-native challenger wins only if the incumbent doesn't.
The signal shows up in our VoC data too. In regulated verticals, mission criticality scores cluster above 9, and NPS doesn't track satisfaction, it tracks switching friction. Customers will tell you the product is mediocre and still score it 9 on "would not switch" because the compliance team vetoes any alternative. ๐ง๐ต๐ฎ๐'๐ ๐๐ต๐ฒ ๐๐ถ๐ด๐ป๐ฎ๐๐๐ฟ๐ฒ ๐ผ๐ณ ๐ฎ ๐ฐ๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ฐ๐ฒ-๐ถ๐ป๐๐๐น๐ฎ๐๐ฒ๐ฑ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ, ๐ฎ๐ ๐น๐ผ๐ป๐ด ๐ฎ๐ ๐๐ต๐ฎ๐ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ ๐ถ๐ ๐ฎ๐ฐ๐๐ถ๐๐ฒ๐น๐ ๐๐ต๐ถ๐ฝ๐ฝ๐ถ๐ป๐ด ๐ฎ๐ด๐ฎ๐ถ๐ป๐๐ ๐๐ต๐ฒ ๐๐ ๐ฐ๐๐ฟ๐๐ฒ.
Which brings us back to the second business for everyone outside the regulated or data-entrenched moat. Seat ARR got them to $100M. But with the shift to agentic workforce structures, partial human capital replacement, and pricing pressure compressing margins, the traditional SaaS model has to transform fast. The next $500M comes from monetizing the installed base: marketplace rake on demand they generate for their own customers, capital products underwritten by their own transaction data, supplier monetization, brand partnerships, group buying. The assets are already sitting there. Captive SMB audience. Proprietary transaction and behavioral data. A distribution pipe (the UI itself) that delivers new products at near-zero CAC.
๐ช๐ต๐ฎ๐'๐ ๐บ๐ถ๐๐๐ถ๐ป๐ด ๐ถ๐ ๐ผ๐ฟ๐ด๐ฎ๐ป๐ถ๐๐ฎ๐๐ถ๐ผ๐ป๐ฎ๐น ๐๐ถ๐น๐น. Monetizing the installed base requires a different org than the one that got you to scale. Different GTM, P&L optics, and talent. Founders and boards under-invest because year one looks worse before it looks better, and public markets punish any SaaS multiple that starts to look like fintech or marketplace. So the second business never ships. The round prices in the optionality. The multiple compresses. The exit underwhelms.
๐ง๐ต๐ฟ๐ฒ๐ฒ ๐ฑ๐ถ๐น๐ถ๐ด๐ฒ๐ป๐ฐ๐ฒ ๐พ๐๐ฒ๐๐๐ถ๐ผ๐ป๐ ๐ป๐ผ๐ ๐ฒ๐ป๐ผ๐๐ด๐ต ๐ถ๐ป๐๐ฒ๐๐๐ผ๐ฟ๐ ๐ฎ๐ฟ๐ฒ ๐ฎ๐๐ธ๐ถ๐ป๐ด:
๐ญ. ๐ช๐ต๐ฎ๐ ๐ฝ๐ฒ๐ฟ๐ฐ๐ฒ๐ป๐ ๐ผ๐ณ ๐ฟ๐ฒ๏ฟฝ๏ฟฝ๏ฟฝ๐ฒ๐ป๐๐ฒ ๐ฐ๐ผ๐บ๐ฒ๐ ๐ณ๐ฟ๐ผ๐บ ๐๐ผ๐๐ฟ๐ฐ๐ฒ๐ ๐ผ๐๐ต๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐๐๐ฏ๐๐ฐ๐ฟ๐ถ๐ฝ๐๐ถ๐ผ๐ป ๐ฎ๐ป๐ฑ ๐ฝ๐ฎ๐๐บ๐ฒ๐ป๐ ๐ฝ๐ฟ๐ผ๐ฐ๐ฒ๐๐๐ถ๐ป๐ด? Under 5%, they haven't started. 10 to 20%, thesis is live. Over 20%, it's working.
๐ฎ. ๐๐ผ๐ ๐ต๐ฎ๐ฟ๐ฑ ๐๐ผ๐๐น๐ฑ ๐ถ๐ ๐ฏ๐ฒ ๐๐ผ ๐ฟ๐ฒ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ ๐๐ต๐ถ๐ ๐ฐ๐ผ๐บ๐ฝ๐ฎ๐ป๐ ๐ณ๐ฟ๐ผ๐บ ๐๐ฐ๐ฟ๐ฎ๐๐ฐ๐ต ๐๐ถ๐๐ต ๐๐ ๐๐ผ๐ฑ๐ฎ๐? If a well-funded team with Claude and six engineers could rebuild the functional product in nine months, the software isn't the moat. The moat has to live somewhere else: proprietary data, a network, integrations, or regulatory surface area the challenger can't clear. If you can't point to at least one, you're underwriting a melting ice cube.
๐ฏ. ๐ช๐ต๐ฎ๐ ๐ฝ๐ฒ๐ฟ๐ฐ๐ฒ๐ป๐ ๐ผ๐ณ ๐๐ต๐ฒ ๏ฟฝ๏ฟฝ๏ฟฝ๐๐๐ฒ๐ฟ'๐ ๐๐๐ถ๐ฐ๐ธ๐ถ๐ป๐ฒ๐๐ ๐ถ๐ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐, ๐ฎ๐ป๐ฑ ๐๐ต๐ถ๐ฐ๐ต ๐๐ฎ๐ ๐ถ๐ ๐๐ต๐ฒ ๐ฟ๐๐น๐ฒ ๐๐ฒ๐ ๐บ๐ผ๐๐ถ๐ป๐ด? A regulatory moat evaporates if the regulation simplifies. Underwrite the direction of travel, not just the current state.
๐๐ป๐ฑ ๐๐ต๐ฒ ๐ฐ๐น๐ผ๐ฐ๐ธ ๐ถ๐ ๐๐ถ๐ด๐ต๐๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐บ๐ผ๐๐ ๐ฟ๐ฒ๐ฎ๐น๐ถ๐๐ฒ. Retention in enterprise SaaS has largely been defined by the pain of systems replacement, not genuine moat. If the stickiness isn't backed by proprietary data, a harvesting flywheel, or regulatory surface area, those vendors are about to get disrupted. Pure seat-based pricing is dying unless vendors embrace agent-seat models, and LLM providers have been subsidizing the market on token cost, with recent pricing shifts signaling cash reserves aren't infinite.
๐๐ฒ๐ฟ๐ฒ'๐ ๐๐ต๐ฒ ๐๐ป๐ฑ๐ฒ๐ฟ๐ฎ๐ฝ๐ฝ๐ฟ๐ฒ๐ฐ๐ถ๐ฎ๐๐ฒ๐ฑ ๐ฝ๐ผ๐ถ๐ป๐: ๐๐-๐ป๐ฎ๐๐ถ๐๐ฒ ๐ฐ๐ผ๐บ๐ฝ๐ฒ๐๐ถ๐๐ผ๐ฟ๐ ๐ต๐ฎ๐๐ฒ ๐๐ผ๐ฟ๐๐ฒ ๐ด๐ฟ๐ผ๐๐ ๐บ๐ฎ๐ฟ๐ด๐ถ๐ป๐ ๐๐ต๐ฎ๐ป ๐ฆ๐ฎ๐ฎ๐ฆ ๐ถ๐ป๐ฐ๐๐บ๐ฏ๐ฒ๐ป๐๐, ๐ป๐ผ๐ ๐ฏ๐ฒ๐๐๐ฒ๐ฟ. Inference costs haven't collapsed, and burning VC cash to subsidize unit economics is a bridge, not a business model. The incumbents should be winning on P&L. They're losing on product velocity and AI-readiness. That's a solvable problem if the board has the will to ship. Vendors without a second business, without a data moat, and without regulatory insulation will still lose, despite having better margins than their AI-native challengers. Customers switch on features and speed, not on unit economics.
๐๐ป๐๐ฒ๐ฟ๐ฝ๐ฟ๐ถ๐๐ฒ ๐ฎ๐ป๐ฑ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ฒ๐ฑ ๐๐ฒ๐ฟ๐๐ถ๐ฐ๐ฎ๐น๐ ๐ฎ๐ฟ๐ฒ ๐๐ต๐ฒ ๐น๐ฎ๐๐ ๐๐ฎ๐ณ๐ฒ ๏ฟฝ๏ฟฝ๐ฎ๐ฟ๐ฏ๐ผ๐ฟ, ๐ฎ๐ป๐ฑ ๐ผ๐ป๐น๐ ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐ฑ๐ฎ๐๐ฎ ๐ฏ๐ฟ๐ฒ๐ฎ๐ฑ๐๐ต ๐ฎ๐ป๐ฑ ๐ฐ๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ฐ๐ฒ. Everywhere else, the premium is about to get competed away. Any fund underwriting vertical SaaS exposure right now should be asking the second-business question before the next check clears. DM me, email me [email protected], or let's chat about your portfolio/underwriting process (https://t.co/muMNtk6ssk).
https://t.co/ElZm7vjalx
Crazy how well this trade has worked out.
Sept 25, 2023: โgrid constraints forcing their hands to use nuclear.โ
The basket since:
โข Talen +670%
โข Vistra +370%
โข Oklo +360% (post-IPO)
โข Constellation +170%
โข NuScale +140%
โข Cameco +100%
S&P 500 same period: +73%.
@marketplunger1 Believe so, heard the same intentions from another big player as well - grid constraints forcing their hands to use nuclear. Here was project I received:
$NTDOY - was cycling through old charts I've done and the Nintendo price action I mapped out in 2023 panned out exactly as technicals were suggesting. Pairing fundamentals + with chart framework = a good way to improve win rate.
$NTDOY - been talking about this one for about 12 months; the technical positioning and fundamental moat of Nintendo is too attractive to overlook at these levels. Multi-bagger setup > I'm projecting a move to $22+ by 2H'25 as momentum comes into the picture.
Stock tends to have massive expansion cycles (see 3 prior moves). 1P content slate, new hardware, margin expansion opportunities.
Off my initial post we got a 25% move to primary resistance with expected pullback as I show on the chart. On a retest, this should start to expand in a much more meaningful capacity.
Revenue segmentation is the whole game. ARR isnโt ARR if 40% sits in the exposed bucket, and the multiple shouldnโt treat it like it is.
The cut that matters for any software name you own:
Non mission critical (engagement layer): massive trouble. Easiest to displace with an agent or thin wrapper. No data gravity, no compliance moat. Net retention is the tell, watch it quarterly.
Enterprise systems of record: insulated, in some cases growing with AI add-ons. Rip-and-replace cycles run 18 to 36 months, and CIOs prefer AI from the incumbent over a new vendor relationship. Pricing power holds.
Mid market: price chasing. AI-native challengers undercut on seat price, and customers use those quotes to negotiate incumbents down even when they donโt switch. Margin compresses either way, and consensus models arenโt pricing it.
SMB: huge risk. Lowest switching costs, no procurement friction, free or cheap AI tools clear the bar. Logo churn shows up first, then ARPU.
System of record vs system of engagement is the frame. Engagement layer is exposed. Record layer is sticky.
This is what bears get wrong. Public software is heavily biased towards the best companies.
There are thousands of little tools that youโve never heard of, but they rarely make it to IPO. When CIOs say theyโre replacing software, theyโre almost always talking about the latter.
The AI Investment Supercycle Hypothesis
Here is my original post from August 11, 2025. Releasing the full version after multiple requests.
Since I wrote this the math has only gotten worse.
The thesis in one line: the current AI boom cannot clear its capital stack. We are heading into a severe crunch and a mass consolidation event. Growth equity will absorb most of the damage.
In slightly longer form:
Too many companies for the available spend.
Too much capital chasing too small a market.
Too much dependency on unprofitable infrastructure.
Something has to give. What gives is the middle of the cap table.
This is not a thesis against AI. The technology is revolutionary. The mistake is assuming every AI company is.
Growth equity is going to take the worst of it. The investors who wrote the largest checks at the highest valuations will be the ones marking down hardest. Most LPs do not yet know how exposed they are.
The era of indiscriminate AI hype investment is winding down. What follows will separate the enduring players from the rest.
The math has the final say. It always does.
And here is the case, broken down...
1. Too many companies, not enough market
> ~70,000 AI startups globally. Most are "AI-enabled" software vendors riding the same handful of foundation models.
> The total AI hardware and software market is forecast at $780B to $990B by 2027.
> Venture and corporate investors have already deployed $500B to $600B into AI companies.
> To clear typical 10x return hurdles, those companies would need to generate $5T to $6T in revenue, roughly 6x the entire expected market.
> Sequoia's David Cahn pegged the gap between AI capex and AI revenue at a "staggering $500 billion" annual hole that has to be filled to justify the spend.
> AI-powered is the new .com. The structural setup looks like 1999.
The math doesn't add up. There is not a pie large enough to justify the cap table.
2. Funding has surged into a market that cannot absorb it
> Generative AI startups raised a record $56B from VCs in 2024 across 885 deals, nearly double the prior year.
~33% of all global VC funding in 2024 went to AI.
> Most of those rounds priced in future dominance with little proven revenue.
> One mid-2025 read: 90% of "AI companies" are just expensive wrappers around the same 5 foundational models.
> Too many startups, same customers, undifferentiated tech, all assuming each captures outsized share.
3. The economics are inverted by subsidy
The core platforms are running at massive operating losses to grab share, and that is what is keeping the rest of the stack alive.
> OpenAI: ~$4B revenue in 2024 against ~$9B in spend. Around $5B in losses for the year.
> One read on the unit economics: OpenAI spends $2.25 for every $1 it earns.
> Anthropic: $5.6B cash burn in 2024 against well under $1B in revenue. Projected ~$3B loss on ~$3.7B revenue in 2025.
> API prices are artificially low. Investors are subsidizing AI compute to drive adoption.
When the subsidy ends, the entire downstream collapses with it.
4. The wrapper economy is built on sand
> An estimated 30,000+ companies are "AI wrappers" that call foundation model APIs, repackage the output, and resell it.
> Same engines means interchangeable products. No IP, no moat, just a well-structured API call, some markup, and marketing.
> Many charge $50 to $100 per month for what a power user could replicate with direct API calls for a few dollars.
> Every token sent through a wrapper, paid or not, earns OpenAI money.
> Wrapper startups are effectively unpaid distribution arms, subsidizing OpenAI's growth while bleeding out.
The house always wins. Until the wrappers run out of cash.
5. The commoditization spiral
The vicious circle for everyone except the foundation model leaders:
> Same models everywhere means undifferentiated products.
> No real differentiation means pricing wars. GPT-4 price cuts already undercut Anthropic by up to 7x on cost per token in some configurations, forcing the industry to match.
> Lower price per customer means collapsed gross margins for any startup reselling AI.
> Lower revenue makes it impossible to support previous valuations or absorb still-rising compute and energy costs.
> Unit economics flip upside down. More users actually means more losses.
> These companies cannot operate without continual investor subsidy.
A race to the bottom is great for end users in the short term. It is lethal for thousands of me-too vendors.
6. Growth equity is the most exposed pool of capital
This is the part the LP letters are not yet ready to write.
> Growth funds wrote $25M to $100M checks at $400M+ valuations on the assumption of 1 or 2 failures out of 10.
> Most companies funded during the 2021 to 2023 boom had 18 to 36 months of capital. Many will run dry by late 2025 or early 2026.
> Industry observers now predict 90%+ of AI startups fail within 5 years.
> Conservative scenario, 60% to 70% failure: ~$400B of the ~$600B deployed gets wiped.
> Realistic scenario, 80% to 90% failure: $500B+ in losses.
> PitchBook already shows AI deal count down 42% in 2024 as reality starts to filter in.
> A growth fund with 40% of its book in AI could see an 80% write-down on that slice. That would be historic.
Early-stage VC tolerates 6 or 7 losses out of 10. Growth equity does not. The model breaks.
7. The exit window is closing too
> The 2025 IPO window for tech is selective at best, and the listings that have priced are not delivering exit multiples that clear late-stage entry.
> M&A is the realistic outcome, but acquirers will wait until valuations crumble.
> A $50M check at a $500M post will recoup pennies in a fire sale.
8. Timeline: from bubble to shakeout
2021 to 2023, build-up: GPT-3, DALL-E, generative AI breakthroughs trigger a flood of startup creation. AI is the new electricity. FOMO drives indiscriminate funding.
2024, peak froth: $56B into genAI. Headlines everywhere. Cracks underneath: GPU bottlenecks, startups with negligible traction, and Big Tech racing ahead. Even the leaders are unprofitable. OpenAI doubles ARR from $6B to $12B in H1 2025, still expects ~$14B loss for the year.
2025, saturation: Every vertical is crowded. Sales cycles lengthen as CIOs get fatigued by the 50th similar pitch. CAC climbs. Big Tech bundles AI into existing platforms, often free, undercutting standalones. Investors get selective.
Late 2025 to 2026, the crunch: Boom-era runways exhaust. Funding environment is harder. LPs are nervous about AI overexposure. Sharp pullback for everyone except the top 5%. Down rounds and outright failures pile up. Sentiment flips from FOMO to caution. The mere mention of AI no longer secures a premium.
2027, mass extinction: The global liquidity squeeze hits. Growth equity and crossover capital largely retreat. Thousands of AI startups fold in a 12 to 18 month window. Direct analog to the 2000 to 2001 dot-com collapse.
2028 to 2029, reset: Survivors fall into two camps. Foundation model and infrastructure leaders. Specialists with truly defensible domain or data moats. With less crazy competition, pricing power returns. API rates rise to profitable levels. A few big winners emerge with public-market validation. Many VC funds report poor returns on the bubble cohort.
9. The endgame
Three things hit at once:
> Valuation collapse. Multiples compress dramatically. AI startups that raised at 100x forward revenue trade at 5 to 10x if they survive at all. Private valuations could fall 70% to 90% before finding a floor.
> Mass closures. Not dozens. Thousands of companies disappear in a short window. Talent and IP get absorbed by larger players.
> Industry reset. Survivors capture larger shares. Pricing rises as subsidies fade. Focus narrows from gimmick AI features to core uses that deliver real ROI. A handful of mega-winners dominate and finally make money on AI.
The truly useful AI applications and companies remain and thrive under more rational economics. The middle of the cap table does not.
In August I wrote a thesis I never published. The funds I was warning were key Crossover Research clients, so I stayed quiet. Since then, ๐ฆ๐ผ๐ณ๐๐๐ฎ๐ฟ๐ฒ ๐บ๐๐น๐๐ถ๐ฝ๐น๐ฒ๐ ๐ฎ๐ฟ๐ฒ ๐ฑ๐ผ๐๐ป ๐ฑ๐ฌ%+. Salesforce $CRM, ServiceNow $NOW, Adobe $ADBE, Workday $WDAY all off 40% from highs. Thomson Reuters $TRI dropped 16% in a single session on the Anthropic legal agent launch. The SaaSpocalypse arrived. So here's the follow-up. Not commentary on what happened, but where I think this goes next.
Most vertical SaaS companies aren't underperforming because their software is bad. ๐ง๐ต๐ฒ๐'๐ฟ๐ฒ ๐๐ป๐ฑ๐ฒ๐ฟ๐ฝ๐ฒ๐ฟ๐ณ๐ผ๐ฟ๐บ๐ถ๐ป๐ด ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐๐ต๐ฒ๐ ๐ป๐ฒ๐๐ฒ๐ฟ ๐ฏ๐๐ถ๐น๐ ๐๐ต๐ฒ ๐๐ฒ๐ฐ๐ผ๐ป๐ฑ ๐ฏ๐๐๐ถ๐ป๐ฒ๐๐. And the first business is under attack. For twenty years, one of the biggest SaaS moats was engineering complexity: deep technical talent, long roadmaps, compounding codebases that were genuinely hard to replicate. ๐๐ ๐๐ฝ๐ฒ๐ป๐ฑ๐ฒ๐ฑ ๐๐ต๐ฎ๐ ๐ฎ๐น๐บ๐ผ๐๐ ๐ผ๐๐ฒ๐ฟ๐ป๐ถ๐ด๐ต๐.
Product development is democratizing to operators with no code background but strong product vision. Look at Anthropic: they've built the engine and are shipping lookalike products at a cadence that would have taken a legacy SaaS vendor three years of roadmap, with a fraction of the headcount. That pace can kill legacy businesses overnight.
๐๐ณ ๐๐ต๐ฒ ๐ฒ๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ถ๐ป๐ด ๐บ๐ผ๐ฎ๐ ๐ถ๏ฟฝ๏ฟฝ๏ฟฝ ๐ด๐ผ๐ป๐ฒ, ๐ณ๐ผ๐๐ฟ ๐บ๐ผ๐ฎ๐๐ ๐ฟ๐ฒ๐บ๐ฎ๐ถ๐ป: ๐ฑ๐ถ๐๐๐ฟ๐ถ๐ฏ๐๐๐ถ๐ผ๐ป, ๐ฝ๐ฟ๐ผ๐ฝ๐ฟ๐ถ๐ฒ๐๐ฎ๐ฟ๐ ๐ฑ๐ฎ๐๐ฎ, ๐๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐ ๐ฏ๐ฟ๐ฒ๐ฎ๐ฑ๐๐ต, ๐ฎ๐ป๐ฑ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐ถ๐ป๐๐๐น๐ฎ๐๐ถ๐ผ๐ป. The first three are moats the company builds. The fourth is a moat the company captures, and it's the one most resistant to AI disruption.
๐ฅ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐ฐ๐ผ๐บ๐ฝ๐น๐ฒ๐ ๐ถ๐๐ ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ๐ ๐๐๐ถ๐๐ฐ๐ต๐ถ๐ป๐ด ๐ฐ๐ผ๐๐๐ ๐๐ต๐ฎ๐ ๐ต๐ฎ๐๐ฒ ๐ป๐ผ๐๐ต๐ถ๐ป๐ด ๐๐ผ ๐ฑ๐ผ ๐๐ถ๐๐ต ๐ฝ๐ฟ๐ผ๐ฑ๐๐ฐ๐ ๐พ๐๐ฎ๐น๐ถ๐๐. Once a vendor is embedded in a compliance workflow, ripping them out means re-attesting, re-auditing, and re-certifying every downstream process. The buyer isn't paying for software, they're paying for the accumulated paper trail. Tyler Technologies ($TYL) is the clearest version of the pattern. State and local government software across courts, public safety, assessment, and ERP. Every module is married to statutory process, FIPS, CJIS, audit trails, and procurement cycles that take years. TYL is down 42% TTM and 2026 guidance came in soft, but the moat didn't break. Revenue still compounded, and government procurement runs on five-year cycles, not five-week news cycles. Veeva is the sharper version. Revenue up 16% in FY26, Q4 beat, the stock still down 25%. The market is selling execution, not weakness. Guidewire in P&C insurance, where regulatory filings and rate approvals anchor the stack, sits in the same setup: still compounding ARR, still winning cloud conversions, multiple reset anyway. Same pattern across all three: multiples compressed, fundamentals intact. The moat is the regulatory surface area itself, and it compounds because the rules get more complex, not less.
๐ ๐๐ฎ๐ ๐น๐ผ๐ป๐ด ๐ฃ๐ฎ๐น๐ฎ๐ป๐๐ถ๐ฟ ๐ฎ๐ $๐ญ๐ฏ (read that here: https://t.co/0N0oIX8N87). ๐ก๐ผ๐ ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐๐ต๐ฒ ๐บ๐ผ๐ฑ๐ฒ๐น ๐ผ๐ฟ ๐๐ต๐ฒ ๐๐ผ๐ผ๐น๐ถ๐ป๐ด. ๐๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐๐ต๐ฒ ๐ผ๐ป๐๐ผ๐น๐ผ๐ด๐. Palantir is the proprietary-data version of the regulatory thesis. Once Palantir sits between the customer and their own data, ripping it out means rebuilding the data model from scratch. Snowflake and Databricks never had that entrenchment layer. AIP bootcamps then turned the data moat into a distribution moat: 660 bootcamps in a single quarter, 94% y/y US customer deal growth, bookings at 1.9x sales. Own the data, ship functional AI on top of it, let the GTM compound. Every vertical incumbent has a version of this available. The question is whether they'll build it before a challenger does.
But regulatory insulation is necessary, not sufficient. Plenty of vendors inside regulated verticals are still getting squeezed because they never became AI-native. BlackLine ($BL) and Trintech are feeling it in close and reconciliation as Numeric, Maximor, and Stacks build AI-native from day one. nCino ($NCNO) in banking faces the same challenge. The regulatory moat buys you time. It doesn't buy you the decade.
๐ง๐ต๐ฒ ๐๐ถ๐ป๐ป๐ถ๐ป๐ด ๐ณ๐ผ๐ฟ๐บ๐๐น๐ฎ ๐ถ๐ ๐ฑ๐ฎ๐๐ฎ ๐ผ๐ฟ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐๐๐ฟ๐ณ๐ฎ๐ฐ๐ฒ ๐ฎ๐ฟ๐ฒ๐ฎ ๐ฝ๐น๐๐ ๐ณ๐๐ป๐ฐ๐๐ถ๐ผ๐ป๐ฎ๐น ๐๐, ๐ป๐ผ๐ ๐ผ๐ป๐ฒ ๐ผ๐ฟ ๐๐ต๐ฒ ๐ผ๐๐ต๐ฒ๐ฟ. Look at why Claude is winning. Anthropic isn't competing on model benchmarks, they're competing on functional workflow. Building for the user, not the leaderboard. That's the playbook vertical incumbents need to run. Take the moat you already have, whether it's regulatory or data-entrenchment, layer genuine workflow AI on top, and the challenger can't catch you. The vendors that do both win the decade. The ones that rely on inertia alone get caught. The ones that ship AI without an anchor get commoditized. You need both.
๐ง๐ต๐ฒ ๐ฏ๐๐๐ฒ๐ฟ ๐ถ๐ ๐๐ฒ๐น๐น๐ถ๐ป๐ด ๐๐ผ๐ ๐๐ต๐ถ๐ ๐ฝ๐น๐ฎ๐ถ๐ป๐น๐. A study we ran with Battery Ventures on AI adoption in the Office of the CFO (https://t.co/xBEMSF8Y72) surveyed 129 finance leaders at companies from $50M to $5B+ in revenue. 77% said they want to uplevel existing systems with AI from new vendors that layer onto existing systems. Only 15% want to replace their current system of record with an AI-native platform. The incumbent wins if they ship AI. The AI-native challenger wins only if the incumbent doesn't.
The signal shows up in our VoC data too. In regulated verticals, mission criticality scores cluster above 9, and NPS doesn't track satisfaction, it tracks switching friction. Customers will tell you the product is mediocre and still score it 9 on "would not switch" because the compliance team vetoes any alternative. ๐ง๐ต๐ฎ๐'๐ ๐๐ต๐ฒ ๐๐ถ๐ด๐ป๐ฎ๐๐๐ฟ๐ฒ ๐ผ๐ณ ๐ฎ ๐ฐ๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ฐ๐ฒ-๐ถ๐ป๐๐๐น๐ฎ๐๐ฒ๐ฑ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ, ๐ฎ๐ ๐น๐ผ๐ป๐ด ๐ฎ๐ ๐๐ต๐ฎ๐ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ ๐ถ๐ ๐ฎ๐ฐ๐๐ถ๐๐ฒ๐น๐ ๐๐ต๐ถ๐ฝ๐ฝ๐ถ๐ป๐ด ๐ฎ๐ด๐ฎ๐ถ๐ป๐๐ ๐๐ต๐ฒ ๐๐ ๐ฐ๐๐ฟ๐๐ฒ.
Which brings us back to the second business for everyone outside the regulated or data-entrenched moat. Seat ARR got them to $100M. But with the shift to agentic workforce structures, partial human capital replacement, and pricing pressure compressing margins, the traditional SaaS model has to transform fast. The next $500M comes from monetizing the installed base: marketplace rake on demand they generate for their own customers, capital products underwritten by their own transaction data, supplier monetization, brand partnerships, group buying. The assets are already sitting there. Captive SMB audience. Proprietary transaction and behavioral data. A distribution pipe (the UI itself) that delivers new products at near-zero CAC.
๐ช๐ต๐ฎ๐'๐ ๐บ๐ถ๐๐๐ถ๐ป๐ด ๐ถ๐ ๐ผ๐ฟ๐ด๐ฎ๐ป๐ถ๐๐ฎ๐๐ถ๐ผ๐ป๐ฎ๐น ๐๐ถ๐น๐น. Monetizing the installed base requires a different org than the one that got you to scale. Different GTM, P&L optics, and talent. Founders and boards under-invest because year one looks worse before it looks better, and public markets punish any SaaS multiple that starts to look like fintech or marketplace. So the second business never ships. The round prices in the optionality. The multiple compresses. The exit underwhelms.
๐ง๐ต๐ฟ๐ฒ๐ฒ ๐ฑ๐ถ๐น๐ถ๐ด๐ฒ๐ป๐ฐ๐ฒ ๐พ๐๐ฒ๐๐๐ถ๐ผ๐ป๐ ๏ฟฝ๏ฟฝ๏ฟฝ๐ผ๐ ๐ฒ๐ป๐ผ๐๐ด๐ต ๐ถ๐ป๐๐ฒ๐๐๐ผ๐ฟ๐ ๐ฎ๐ฟ๐ฒ ๐ฎ๐๐ธ๐ถ๐ป๐ด:
๐ญ. ๐ช๐ต๐ฎ๐ ๐ฝ๐ฒ๐ฟ๐ฐ๐ฒ๐ป๐ ๐ผ๐ณ ๐ฟ๐ฒ๐๐ฒ๐ป๐๐ฒ ๐ฐ๐ผ๐บ๐ฒ๐ ๐ณ๐ฟ๐ผ๐บ ๐๐ผ๐๐ฟ๐ฐ๐ฒ๐ ๐ผ๐๐ต๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐๐๐ฏ๐๐ฐ๐ฟ๐ถ๐ฝ๐๐ถ๐ผ๐ป ๐ฎ๐ป๐ฑ ๐ฝ๐ฎ๐๐บ๐ฒ๐ป๐ ๐ฝ๐ฟ๐ผ๐ฐ๐ฒ๐๐๐ถ๐ป๐ด? Under 5%, they haven't started. 10 to 20%, thesis is live. Over 20%, it's working.
๐ฎ. ๐๐ผ๐ ๐ต๐ฎ๐ฟ๐ฑ ๐๐ผ๐๐น๐ฑ ๐ถ๐ ๐ฏ๐ฒ ๐๐ผ ๐ฟ๐ฒ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ ๐๐ต๐ถ๐ ๐ฐ๐ผ๐บ๐ฝ๐ฎ๐ป๐ ๐ณ๐ฟ๐ผ๐บ ๐๐ฐ๐ฟ๐ฎ๐๐ฐ๐ต ๐๐ถ๐๐ต ๐๐ ๐๐ผ๐ฑ๐ฎ๐? If a well-funded team with Claude and six engineers could rebuild the functional product in nine months, the software isn't the moat. The moat has to live somewhere else: proprietary data, a network, integrations, or regulatory surface area the challenger can't clear. If you can't point to at least one, you're underwriting a melting ice cube.
๐ฏ. ๐ช๐ต๐ฎ๐ ๐ฝ๐ฒ๐ฟ๐ฐ๐ฒ๐ป๐ ๐ผ๐ณ ๐๐ต๐ฒ ๐ฏ๐๐๐ฒ๐ฟ'๐ ๐๐๐ถ๐ฐ๐ธ๐ถ๐ป๐ฒ๐๐ ๐ถ๐ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐, ๐ฎ๐ป๐ฑ ๐๐ต๐ถ๐ฐ๐ต ๐๐ฎ๐ ๐ถ๐ ๐๐ต๐ฒ ๐ฟ๐๐น๐ฒ ๐๐ฒ๐ ๐บ๐ผ๐๐ถ๐ป๐ด? A regulatory moat evaporates if the regulation simplifies. Underwrite the direction of travel, not just the current state.
๐๐ป๐ฑ ๐๐ต๐ฒ ๐ฐ๐น๐ผ๐ฐ๐ธ ๐ถ๐ ๐๐ถ๐ด๐ต๐๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐บ๐ผ๐๐ ๐ฟ๐ฒ๐ฎ๐น๐ถ๐๐ฒ. Retention in enterprise SaaS has largely been defined by the pain of systems replacement, not genuine moat. If the stickiness isn't backed by proprietary data, a harvesting flywheel, or regulatory surface area, those vendors are about to get disrupted. Pure seat-based pricing is dying unless vendors embrace agent-seat models, and LLM providers have been subsidizing the market on token cost, with recent pricing shifts signaling cash reserves aren't infinite.
๐๐ฒ๐ฟ๐ฒ'๐ ๐๐ต๐ฒ ๐๐ป๐ฑ๐ฒ๐ฟ๐ฎ๐ฝ๐ฝ๐ฟ๐ฒ๐ฐ๐ถ๐ฎ๐๐ฒ๐ฑ ๐ฝ๐ผ๐ถ๐ป๐: ๐๐-๐ป๐ฎ๐๐ถ๐๐ฒ ๐ฐ๐ผ๐บ๐ฝ๐ฒ๐๐ถ๐๐ผ๐ฟ๐ ๐ต๐ฎ๐๐ฒ ๐๐ผ๐ฟ๐๐ฒ ๐ด๐ฟ๐ผ๐๐ ๐บ๐ฎ๐ฟ๐ด๐ถ๐ป๐ ๐๐ต๐ฎ๐ป ๐ฆ๐ฎ๐ฎ๐ฆ ๐ถ๐ป๐ฐ๐๐บ๐ฏ๐ฒ๐ป๐๐, ๐ป๐ผ๐ ๐ฏ๐ฒ๐๐๐ฒ๐ฟ. Inference costs haven't collapsed, and burning VC cash to subsidize unit economics is a bridge, not a business model. The incumbents should be winning on P&L. They're losing on product velocity and AI-readiness. That's a solvable problem if the board has the will to ship. Vendors without a second business, without a data moat, and without regulatory insulation will still lose, despite having better margins than their AI-native challengers. Customers switch on features and speed, not on unit economics.
๐๐ป๐๐ฒ๐ฟ๐ฝ๐ฟ๐ถ๐๐ฒ ๐ฎ๐ป๐ฑ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ฒ๐ฑ ๐๐ฒ๐ฟ๐๐ถ๐ฐ๐ฎ๐น๐ ๐ฎ๐ฟ๐ฒ ๐๐ต๐ฒ ๐น๐ฎ๐๐ ๐๐ฎ๐ณ๐ฒ ๐ต๐ฎ๐ฟ๐ฏ๐ผ๐ฟ, ๐ฎ๐ป๐ฑ ๐ผ๐ป๐น๐ ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐ฑ๐ฎ๐๐ฎ ๐ฏ๐ฟ๐ฒ๐ฎ๐ฑ๐๐ต ๐ฎ๐ป๐ฑ ๐ฐ๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ฐ๐ฒ. Everywhere else, the premium is about to get competed away. Any fund underwriting vertical SaaS exposure right now should be asking the second-business question before the next check clears. DM me, email me [email protected], or let's chat about your portfolio/underwriting process (https://t.co/muMNtk6ssk).
https://t.co/ElZm7vjalx
Closer than 9-18 months. Most of what youโre describing is running on our side today: continuous driver research, internal data ingest (transcripts, alt data, expert calls), source-level click-through validation, autonomous estimates and R/R, dashboard view with flags. The Excel ingress/egress is the piece that looks hardest, and it collapses the moment you stop building it as Excel. Headless spreadsheet on Postgres.
Launching this as a platform in the next month. DM and Iโll send you a few links
@garrytan@dharmesh
I just wrote something where I suggested that CEOs not leaning into AI are uninvestable. The pace is compounding weekly and if you delegate understanding you are ultimately delegating strategy. The CEO = visionary and in my view confident decisions require firsthand fluency. Anything secondhand is already behind. Curious where you both land on this.
https://t.co/NskzqNNlAc
In August I wrote a thesis I never published. The funds I was warning were key Crossover Research clients, so I stayed quiet. Since then, ๐ฆ๐ผ๐ณ๐๐๐ฎ๐ฟ๐ฒ ๐บ๐๐น๐๐ถ๐ฝ๐น๐ฒ๐ ๐ฎ๐ฟ๐ฒ ๐ฑ๐ผ๐๐ป ๐ฑ๐ฌ%+. Salesforce $CRM, ServiceNow $NOW, Adobe $ADBE, Workday $WDAY all off 40% from highs. Thomson Reuters $TRI dropped 16% in a single session on the Anthropic legal agent launch. The SaaSpocalypse arrived. So here's the follow-up. Not commentary on what happened, but where I think this goes next.
Most vertical SaaS companies aren't underperforming because their software is bad. ๐ง๐ต๐ฒ๐'๐ฟ๐ฒ ๐๐ป๐ฑ๐ฒ๐ฟ๐ฝ๐ฒ๐ฟ๐ณ๐ผ๐ฟ๐บ๐ถ๐ป๐ด ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐๐ต๐ฒ๐ ๐ป๐ฒ๐๐ฒ๐ฟ ๐ฏ๐๐ถ๐น๐ ๐๐ต๐ฒ ๐๐ฒ๐ฐ๐ผ๐ป๐ฑ ๐ฏ๐๐๐ถ๐ป๐ฒ๐๐. And the first business is under attack. For twenty years, one of the biggest SaaS moats was engineering complexity: deep technical talent, long roadmaps, compounding codebases that were genuinely hard to replicate. ๐๐ ๐๐ฝ๐ฒ๐ป๐ฑ๐ฒ๐ฑ ๐๐ต๐ฎ๐ ๐ฎ๐น๐บ๐ผ๐๐ ๐ผ๐๐ฒ๐ฟ๐ป๐ถ๐ด๐ต๐.
Product development is democratizing to operators with no code background but strong product vision. Look at Anthropic: they've built the engine and are shipping lookalike products at a cadence that would have taken a legacy SaaS vendor three years of roadmap, with a fraction of the headcount. That pace can kill legacy businesses overnight.
๐๐ณ ๐๐ต๐ฒ ๐ฒ๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ถ๐ป๐ด ๐บ๐ผ๐ฎ๏ฟฝ๏ฟฝ ๐ถ๐ ๐ด๐ผ๐ป๐ฒ, ๐ณ๐ผ๐๐ฟ ๐บ๐ผ๐ฎ๐๐ ๐ฟ๐ฒ๐บ๐ฎ๐ถ๐ป: ๐ฑ๐ถ๐๐๐ฟ๐ถ๐ฏ๐๐๐ถ๐ผ๐ป, ๐ฝ๐ฟ๐ผ๐ฝ๐ฟ๐ถ๐ฒ๐๐ฎ๐ฟ๐ ๐ฑ๐ฎ๐๐ฎ, ๐๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐ ๐ฏ๐ฟ๐ฒ๐ฎ๐ฑ๐๐ต, ๐ฎ๐ป๐ฑ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐ถ๐ป๐๐๐น๐ฎ๐๐ถ๐ผ๐ป. The first three are moats the company builds. The fourth is a moat the company captures, and it's the one most resistant to AI disruption.
๐ฅ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐ฐ๐ผ๐บ๐ฝ๐น๐ฒ๐ ๐ถ๐๐ ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ๐ ๐๐๐ถ๐๐ฐ๐ต๐ถ๐ป๐ด ๐ฐ๐ผ๐๐๐ ๐๐ต๐ฎ๐ ๐ต๐ฎ๐๐ฒ ๐ป๐ผ๐๐ต๐ถ๐ป๐ด ๐๐ผ ๐ฑ๐ผ ๐๐ถ๐๐ต ๐ฝ๐ฟ๐ผ๐ฑ๐๐ฐ๐ ๐พ๐๐ฎ๐น๐ถ๐๐. Once a vendor is embedded in a compliance workflow, ripping them out means re-attesting, re-auditing, and re-certifying every downstream process. The buyer isn't paying for software, they're paying for the accumulated paper trail. Tyler Technologies ($TYL) is the clearest version of the pattern. State and local government software across courts, public safety, assessment, and ERP. Every module is married to statutory process, FIPS, CJIS, audit trails, and procurement cycles that take years. TYL is down 42% TTM and 2026 guidance came in soft, but the moat didn't break. Revenue still compounded, and government procurement runs on five-year cycles, not five-week news cycles. Veeva is the sharper version. Revenue up 16% in FY26, Q4 beat, the stock still down 25%. The market is selling execution, not weakness. Guidewire in P&C insurance, where regulatory filings and rate approvals anchor the stack, sits in the same setup: still compounding ARR, still winning cloud conversions, multiple reset anyway. Same pattern across all three: multiples compressed, fundamentals intact. The moat is the regulatory surface area itself, and it compounds because the rules get more complex, not less.
๐ ๐๐ฎ๐ ๐น๐ผ๐ป๐ด ๐ฃ๐ฎ๐น๐ฎ๐ป๐๐ถ๐ฟ ๐ฎ๐ $๐ญ๐ฏ (read that here: https://t.co/0N0oIX8N87). ๐ก๐ผ๐ ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐๐ต๐ฒ ๐บ๐ผ๐ฑ๐ฒ๐น ๐ผ๐ฟ ๐๐ต๐ฒ ๐๐ผ๐ผ๐น๐ถ๐ป๐ด. ๐๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐๐ต๐ฒ ๐ผ๐ป๐๐ผ๐น๐ผ๐ด๐. Palantir is the proprietary-data version of the regulatory thesis. Once Palantir sits between the customer and their own data, ripping it out means rebuilding the data model from scratch. Snowflake and Databricks never had that entrenchment layer. AIP bootcamps then turned the data moat into a distribution moat: 660 bootcamps in a single quarter, 94% y/y US customer deal growth, bookings at 1.9x sales. Own the data, ship functional AI on top of it, let the GTM compound. Every vertical incumbent has a version of this available. The question is whether they'll build it before a challenger does.
But regulatory insulation is necessary, not sufficient. Plenty of vendors inside regulated verticals are still getting squeezed because they never became AI-native. BlackLine ($BL) and Trintech are feeling it in close and reconciliation as Numeric, Maximor, and Stacks build AI-native from day one. nCino ($NCNO) in banking faces the same challenge. The regulatory moat buys you time. It doesn't buy you the decade.
๐ง๐ต๐ฒ ๐๐ถ๐ป๐ป๐ถ๐ป๐ด ๐ณ๐ผ๐ฟ๐บ๐๐น๐ฎ ๐ถ๐ ๐ฑ๐ฎ๐๐ฎ ๐ผ๐ฟ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐๐๐ฟ๐ณ๐ฎ๐ฐ๐ฒ ๐ฎ๐ฟ๐ฒ๐ฎ ๐ฝ๐น๐๐ ๐ณ๐๐ป๐ฐ๐๐ถ๐ผ๐ป๐ฎ๐น ๐๐, ๐ป๐ผ๐ ๐ผ๐ป๐ฒ ๐ผ๐ฟ ๐๐ต๐ฒ ๐ผ๐๐ต๐ฒ๐ฟ. Look at why Claude is winning. Anthropic isn't competing on model benchmarks, they're competing on functional workflow. Building for the user, not the leaderboard. That's the playbook vertical incumbents need to run. Take the moat you already have, whether it's regulatory or data-entrenchment, layer genuine workflow AI on top, and the challenger can't catch you. The vendors that do both win the decade. The ones that rely on inertia alone get caught. The ones that ship AI without an anchor get commoditized. You need both.
๐ง๐ต๐ฒ ๐ฏ๐๐๐ฒ๐ฟ ๐ถ๐ ๐๐ฒ๐น๐น๐ถ๐ป๐ด ๐๐ผ๐ ๐๐ต๐ถ๐ ๐ฝ๐น๐ฎ๐ถ๐ป๐น๐. A study we ran with Battery Ventures on AI adoption in the Office of the CFO (https://t.co/xBEMSF8Y72) surveyed 129 finance leaders at companies from $50M to $5B+ in revenue. 77% said they want to uplevel existing systems with AI from new vendors that layer onto existing systems. Only 15% want to replace their current system of record with an AI-native platform. The incumbent wins if they ship AI. The AI-native challenger wins only if the incumbent doesn't.
The signal shows up in our VoC data too. In regulated verticals, mission criticality scores cluster above 9, and NPS doesn't track satisfaction, it tracks switching friction. Customers will tell you the product is mediocre and still score it 9 on "would not switch" because the compliance team vetoes any alternative. ๐ง๐ต๐ฎ๐'๐ ๐๐ต๐ฒ ๐๐ถ๐ด๐ป๐ฎ๐๐๐ฟ๐ฒ ๐ผ๐ณ ๐ฎ ๐ฐ๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ฐ๐ฒ-๐ถ๐ป๐๐๐น๐ฎ๐๐ฒ๐ฑ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ, ๐ฎ๐ ๐น๐ผ๐ป๐ด ๐ฎ๐ ๐๐ต๐ฎ๐ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ ๐ถ๐ ๐ฎ๐ฐ๐๐ถ๐๐ฒ๐น๐ ๐๐ต๐ถ๐ฝ๐ฝ๐ถ๐ป๐ด ๐ฎ๐ด๐ฎ๐ถ๐ป๐๐ ๐๐ต๐ฒ ๐๐ ๐ฐ๐๐ฟ๐๐ฒ.
Which brings us back to the second business for everyone outside the regulated or data-entrenched moat. Seat ARR got them to $100M. But with the shift to agentic workforce structures, partial human capital replacement, and pricing pressure compressing margins, the traditional SaaS model has to transform fast. The next $500M comes from monetizing the installed base: marketplace rake on demand they generate for their own customers, capital products underwritten by their own transaction data, supplier monetization, brand partnerships, group buying. The assets are already sitting there. Captive SMB audience. Proprietary transaction and behavioral data. A distribution pipe (the UI itself) that delivers new products at near-zero CAC.
๐ช๐ต๐ฎ๐'๐ ๐บ๐ถ๐๐๐ถ๐ป๐ด ๐ถ๐ ๐ผ๐ฟ๐ด๐ฎ๐ป๐ถ๐๐ฎ๐๐ถ๐ผ๐ป๐ฎ๐น ๐๐ถ๐น๐น. Monetizing the installed base requires a different org than the one that got you to scale. Different GTM, P&L optics, and talent. Founders and boards under-invest because year one looks worse before it looks better, and public markets punish any SaaS multiple that starts to look like fintech or marketplace. So the second business never ships. The round prices in the optionality. The multiple compresses. The exit underwhelms.
๐ง๐ต๐ฟ๐ฒ๐ฒ ๐ฑ๐ถ๐น๐ถ๐ด๐ฒ๐ป๐ฐ๐ฒ ๐พ๐๐ฒ๐๐๐ถ๐ผ๐ป๐ ๐ป๐ผ๐ ๐ฒ๐ป๐ผ๐๐ด๐ต ๐ถ๐ป๐๐ฒ๐๐๐ผ๐ฟ๐ ๐ฎ๐ฟ๐ฒ ๐ฎ๐๐ธ๐ถ๐ป๐ด:
๐ญ. ๐ช๐ต๐ฎ๐ ๐ฝ๐ฒ๐ฟ๐ฐ๐ฒ๏ฟฝ๏ฟฝ๏ฟฝ๏ฟฝ๐ ๐ผ๐ณ ๐ฟ๐ฒ๐๐ฒ๐ป๐๐ฒ ๐ฐ๐ผ๐บ๐ฒ๐ ๐ณ๐ฟ๐ผ๐บ ๐๐ผ๐๐ฟ๐ฐ๐ฒ๐ ๐ผ๐๐ต๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐๐๐ฏ๐๐ฐ๐ฟ๐ถ๐ฝ๐๐ถ๐ผ๐ป ๐ฎ๐ป๐ฑ ๐ฝ๐ฎ๐๐บ๐ฒ๐ป๐ ๐ฝ๐ฟ๐ผ๐ฐ๐ฒ๐๐๐ถ๐ป๐ด? Under 5%, they haven't started. 10 to 20%, thesis is live. Over 20%, it's working.
๐ฎ. ๐๐ผ๐ ๐ต๐ฎ๐ฟ๐ฑ ๐๐ผ๐๐น๐ฑ ๐ถ๐ ๐ฏ๐ฒ ๐๐ผ ๐ฟ๐ฒ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ ๐๐ต๐ถ๐ ๐ฐ๐ผ๐บ๐ฝ๐ฎ๐ป๐ ๐ณ๐ฟ๐ผ๐บ ๐๐ฐ๐ฟ๐ฎ๐๐ฐ๐ต ๐๐ถ๐๐ต ๐๐ ๐๐ผ๐ฑ๐ฎ๐? If a well-funded team with Claude and six engineers could rebuild the functional product in nine months, the software isn't the moat. The moat has to live somewhere else: proprietary data, a network, integrations, or regulatory surface area the challenger can't clear. If you can't point to at least one, you're underwriting a melting ice cube.
๐ฏ. ๐ช๐ต๐ฎ๐ ๐ฝ๐ฒ๐ฟ๐ฐ๐ฒ๐ป๏ฟฝ๏ฟฝ๏ฟฝ ๐ผ๐ณ ๐๐ต๐ฒ ๐ฏ๐๐๐ฒ๐ฟ'๐ ๐๐๐ถ๐ฐ๐ธ๐ถ๐ป๐ฒ๐๐ ๐ถ๐ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐, ๐ฎ๐ป๐ฑ ๐๐ต๐ถ๐ฐ๐ต ๐๐ฎ๐ ๐ถ๐ ๐๐ต๐ฒ ๐ฟ๐๐น๐ฒ ๐๐ฒ๐ ๐บ๐ผ๐๐ถ๐ป๐ด? A regulatory moat evaporates if the regulation simplifies. Underwrite the direction of travel, not just the current state.
๐๐ป๐ฑ ๐๐ต๐ฒ ๐ฐ๐น๐ผ๐ฐ๐ธ ๐ถ๐ ๐๐ถ๐ด๐ต๐๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐บ๐ผ๐๐ ๐ฟ๐ฒ๐ฎ๐น๐ถ๐๐ฒ. Retention in enterprise SaaS has largely been defined by the pain of systems replacement, not genuine moat. If the stickiness isn't backed by proprietary data, a harvesting flywheel, or regulatory surface area, those vendors are about to get disrupted. Pure seat-based pricing is dying unless vendors embrace agent-seat models, and LLM providers have been subsidizing the market on token cost, with recent pricing shifts signaling cash reserves aren't infinite.
๐๐ฒ๐ฟ๐ฒ'๐ ๐๐ต๐ฒ ๐๐ป๐ฑ๐ฒ๐ฟ๐ฎ๐ฝ๐ฝ๐ฟ๐ฒ๐ฐ๐ถ๐ฎ๐๐ฒ๐ฑ ๐ฝ๐ผ๐ถ๐ป๐: ๐๐-๐ป๐ฎ๐๐ถ๐๐ฒ ๐ฐ๐ผ๐บ๐ฝ๐ฒ๐๐ถ๐๐ผ๐ฟ๐ ๐ต๐ฎ๐๐ฒ ๐๐ผ๐ฟ๐๐ฒ ๐ด๐ฟ๐ผ๐๐ ๐บ๐ฎ๐ฟ๐ด๐ถ๐ป๐ ๐๐ต๐ฎ๐ป ๐ฆ๐ฎ๐ฎ๐ฆ ๐ถ๐ป๐ฐ๐๐บ๐ฏ๐ฒ๐ป๐๐, ๐ป๐ผ๐ ๐ฏ๐ฒ๐๐๐ฒ๐ฟ. Inference costs haven't collapsed, and burning VC cash to subsidize unit economics is a bridge, not a business model. The incumbents should be winning on P&L. They're losing on product velocity and AI-readiness. That's a solvable problem if the board has the will to ship. Vendors without a second business, without a data moat, and without regulatory insulation will still lose, despite having better margins than their AI-native challengers. Customers switch on features and speed, not on unit economics.
๐๐ป๐๐ฒ๐ฟ๐ฝ๐ฟ๐ถ๐๐ฒ ๐ฎ๐ป๐ฑ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ฒ๐ฑ ๐๐ฒ๐ฟ๐๐ถ๐ฐ๐ฎ๐น๏ฟฝ๏ฟฝ ๐ฎ๐ฟ๐ฒ ๐๐ต๐ฒ ๐น๐ฎ๐๐ ๐๐ฎ๐ณ๐ฒ ๐ต๐ฎ๐ฟ๐ฏ๐ผ๐ฟ, ๐ฎ๐ป๐ฑ ๐ผ๐ป๐น๐ ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐ฑ๐ฎ๐๐ฎ ๐ฏ๐ฟ๐ฒ๐ฎ๐ฑ๐๐ต ๐ฎ๐ป๐ฑ ๐ฐ๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ฐ๐ฒ. Everywhere else, the premium is about to get competed away. Any fund underwriting vertical SaaS exposure right now should be asking the second-business question before the next check clears. DM me, email me [email protected], or let's chat about your portfolio/underwriting process (https://t.co/muMNtk6ssk).
https://t.co/ElZm7vjalx
๐ง๐๐ ๐๐ ๐ฃ๐๐๐ฌ๐๐ข๏ฟฝ๏ฟฝ๏ฟฝ๏ฟฝ๐ ๐๐ข๐ฅ ๐ข๐ฃ๐๐ฅ๐๐ง๐ข๐ฅ๐ฆ ๐๐ก๐ ๐๐ก๐ฉ๐๐ฆ๐ง๐ข๐ฅ๐ฆ
The technical stack matters more than ever. If the vendors in your stack aren't shipping aggressively, your competitors are already outpacing you. Every C-suite and CTO should be running a full evaluation of vendor product development timelines right now. ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ ๐ถ๐ ๐๐ต๐ฒ ๐ณ๐ฎ๐ฐ๐๐ผ๐ฟ ๐๐ต๐ฎ๐ ๐บ๐ฎ๐๐๐ฒ๐ฟ๐ ๐บ๐ผ๐๐, and the delta between internal tooling and your existing vendors is what determines whether you're a ๐น๐ฒ๐ฎ๐ฑ๐ฒ๐ฟ ๐ผ๐ฟ ๐ฎ ๐น๐ฎ๐ด๐ด๐ฎ๐ฟ๐ฑ.
Before I get into it, three questions every founder and C-suite team needs to answer:
1) When did you last audit your vendor stack for shipping velocity?
2) If Anthropic or OpenAI shipped your core capability tomorrow as a native feature, what would you still own?
3) Can you, the CEO, explain your product's AI architecture in under 5 minutes without deferring to your CTO?
๐๐ก๐ง๐๐ฅ๐ข๐ฃ๐๐ ๐๐๐ฆ๐๐๐๐ก๐
In 30 days: Claude Opus 4.7. Claude Design. Project Glasswing with Mythos Preview. Cowork GA on Mac and Windows. Computer use in Cowork and Claude Code. Interactive apps on mobile. Auto mode and /ultrareview in Claude Code. ๐ ๐ผ๐ฟ๐ฒ ๐ฝ๐ฟ๐ผ๐ฑ๐๐ฐ๐ ๐๐ต๐ฎ๐ป ๐บ๐ผ๐๐ ๐ฆ๐ฎ๐ฎ๐ฆ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ๐ ๐ต๐ฎ๐๐ฒ ๐๐ต๐ถ๐ฝ๐ฝ๐ฒ๐ฑ ๐ถ๐ป ๐๐๐ผ ๐๐ฒ๐ฎ๐ฟ๐. Imagine an Olympian showing up to your after-work sports league. Cool to watch, not a fair fight.
And they sit in the ๐ฟ๐ถ๐๐ธ-๐ณ๐ฟ๐ฒ๐ฒ ๐ถ๐ป๐ฐ๐๐ฏ๐ฎ๐๐ผ๐ฟ ๐๐ฒ๐ฎ๐. Every business building on Anthropic's platform to stay competitive is also showing them which vectors are worth attacking. Claude Design came for Figma the week it shipped. Claude Code came for Cursor. ๐ฌ๐ผ๐ ๐ฏ๐๐ถ๐น๐ฑ ๐๐ต๐ฒ ๐บ๐ฎ๐ฟ๐ธ๐ฒ๐. ๐๐ป๐๐ต๐ฟ๐ผ๐ฝ๐ถ๐ฐ ๐ฟ๐ฒ๐ฎ๐ฑ๐ ๐๐ต๐ฒ ๐๐ถ๐ด๐ป๐ฎ๐น. ๐ง๐ต๐ฒ๐ ๐๐ต๐ถ๐ฝ ๐ป๐ฎ๐๐ถ๐๐ฒ.
Reading the signal only works if you can act on it faster than anyone else, and that's why ๐๐ป๐๐ต๐ฟ๐ผ๐ฝ๐ถ๐ฐ'๐ ๐ฒ๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ ๐ฟ๐๐ป ๐๐ถ๐๐ต๐ผ๐๐ ๐ฎ ๐๐ผ๐ธ๐ฒ๐ป ๐ฐ๐ฒ๐ถ๐น๐ถ๐ป๐ด. The NYT reported one engineer spent over $150,000 on Claude Code in a single month. ๐ง๐ต๐ฎ๐'๐ ๐ป๐ผ๐ ๐ฎ ๐ฏ๐๐ด. ๐ง๐ต๐ฎ๐'๐ ๐๐ต๐ฒ ๐ถ๐ป๐ฐ๐ฒ๐ป๐๐ถ๐๐ฒ ๐๐๐ฟ๐๐ฐ๐๐๐ฟ๐ฒ. Uber rolled Claude Code to 5,000 engineers and ๐ฏ๐น๐ฒ๐ ๐ถ๐๐ ๐ฒ๐ป๐๐ถ๐ฟ๐ฒ ๐ฎ๐ฌ๐ฎ๐ฒ ๐๐ ๐ฏ๐๐ฑ๐ด๐ฒ๐ ๐ถ๐ป ๐ณ๐ผ๐๐ฟ ๐บ๐ผ๐ป๐๐ต๐. CTO Praveen Neppalli Naga: "I'm back to the drawing board, because the budget I thought I would need is blown away already." He's replanning upward, not pulling back. ๐ง๐ต๐ฎ๐ ๐ฎ๐๐๐บ๐บ๐ฒ๐๐ฟ๐ ๐ถ๐ ๐๐ต๐ฒ ๐ด๐ฎ๐ฝ: Anthropic ships without a cost ceiling. Every buyer ships with one.
๐ฆ๐ฃ๐๐๐ ๐๐ฆ ๐ง๏ฟฝ๏ฟฝ๏ฟฝ๐ ๐ ๐ข๐๐ง
Every SaaS vendor in your stack should be benchmarked against internal dev velocity. If your team with Claude can ship in weeks what your vendors take quarters to deliver, those vendors aren't buying you time anymore. ๐ง๐ต๐ฒ๐'๐ฟ๐ฒ ๐ฐ๐ผ๐๐๐ถ๐ป๐ด ๐๐ผ๐ ๐๐ถ๐บ๐ฒ. The question on every renewal is no longer "is this vendor reliable." It's ๐๐ผ๐ฟ๐๐ต ๐๐ต๐ฒ ๐๐ฒ๐น๐ผ๐ฐ๐ถ๐๐ ๐ด๐ฎ๐ฝ ๐ถ๐ ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ๐. If the answer is no, they're the thing killing you.
๐ฉ๐๐๐ ๐ฆ๐๐ข๐ฃ ๐๐ฆ ๐ ๐ฆ๐๐๐๐ ๐๐ฆ๐ฆ๐จ๐
Most people dismissing AI coding as slop are reacting to bad implementations, not AI itself. The slop they're seeing is real. ๐๐'๐ ๐ฎ ๐๐ธ๐ถ๐น๐น ๐ฎ๐ป๐ฑ ๐ฎ๐ฟ๐ฐ๐ต๐ถ๐๐ฒ๐ฐ๐๐๐ฟ๐ฒ ๐ฝ๐ฟ๐ผ๐ฏ๐น๐ฒ๐บ, ๐ป๐ผ๐ ๐ฎ ๐บ๐ผ๐ฑ๐ฒ๐น ๐ฝ๐ฟ๐ผ๐ฏ๐น๐ฒ๐บ. Good engineers put proper bounds in place: modular architecture, shared layers, fallback patterns, drift detection. ๐๐ผ๐ป๐ฒ ๐ฟ๐ถ๐ด๐ต๐, ๐๐ถ๐ฏ๐ฒ ๐ฐ๐ผ๐ฑ๐ถ๐ป๐ด ๐ถ๐ ๐ญ๐ฌ๐ ๐น๐ฒ๐๐ฒ๐ฟ๐ฎ๐ด๐ฒ. The architecture you build on determines whether AI compounds or corrupts. Proof point: Boris Cherny, head of Claude Code, said Anthropic's AI coded "pretty much all" of Cowork, ๐ฏ๐๐ถ๐น๐ ๐ถ๐ป ๐ฟ๐ผ๐๐ด๐ต๐น๐ ๐๐๐ผ ๐๐ฒ๐ฒ๐ธ๐.
๐ง๐๐ ๐ฆ๐ง๐๐ฅ๐ง๐จ๐ฃ ๐ช๐๐ก๐๐ข๐ช
Startups with no legacy architecture have a ๐ป๐ฎ๐ฟ๐ฟ๐ผ๐, ๐ฟ๐ฒ๐ฎ๐น ๐๐ถ๐ป๐ฑ๐ผ๐ ๐๐ผ ๐ต๐ถ๐ ๐๐ฅ๐ฅ ๐๐ต๐ฟ๐ฒ๐๐ต๐ผ๐น๐ฑ๐ ๐ป๐ผ ๐ผ๐ป๐ฒ ๐๐ต๐ผ๐๐ด๐ต๐ ๐ฝ๐ผ๐๐๐ถ๐ฏ๐น๐ฒ. Disadvantage: data breadth. Advantage: data capture design. ๐๐ฟ๐ฐ๐ต๐ถ๐๐ฒ๐ฐ๐ ๐๐-๐ป๐ฎ๐๐ถ๐๐ฒ ๐ณ๐ฟ๐ผ๐บ ๐ฑ๐ฎ๐ ๐ผ๐ป๐ฒ ๐ฎ๐ป๐ฑ ๐๐ผ๐ ๐ฐ๐ผ๐บ๐ฝ๐ผ๐๏ฟฝ๏ฟฝ๐ฑ ๐ฝ๐ฟ๐ผ๐ฝ๐ฟ๐ถ๐ฒ๐๐ฎ๐ฟ๐ ๐ฑ๐ฎ๐๐ฎ ๐ณ๐ฎ๐๐๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐ฒ๐ป๐๐ฒ๐ฟ๐ฝ๐ฟ๐ถ๐๐ฒ๐ ๐ฐ๐ฎ๐ป ๐ฝ๐ผ๐ฟ๐ ๐๐ต๐ฒ๐ถ๐ฟ๐.
๐๐-๐๐๐ฅ๐ฆ๐ง ๐-๐ฆ๐จ๐๐ง๐๐ฆ ๐ฆ๐๐ข๐จ๐๐ ๐๐ ๐๐ก ๐จ๐ก๐๐๐ฅ๐ช๐ฅ๐๐ง๐๐ก๐ ๐ฅ๐๐ค๐จ๐๐ฅ๐๐ ๐๐ก๐ง
๐๐ผ๐๐ป๐ฑ๐ฒ๐ฟ๐ ๐ฎ๐ป๐ฑ ๐-๐๐๐ถ๐๐ฒ๐ ๐๐ต๐ผ ๐ฑ๐ฒ๐น๐ฒ๐ด๐ฎ๐๐ฒ ๐๐ ๐ณ๐น๐๐ฒ๐ป๐ฐ๐ ๐ต๐ฎ๐๐ฒ ๐ฎ๐น๐ฟ๐ฒ๐ฎ๐ฑ๐ ๐น๐ผ๐๐. Secondhand information moves too slowly in a market where architecture decisions compound daily. C-suites set vision, and successful AI deployments come from leaders who can ๐ฐ๐ผ๐ป๐ณ๐ถ๐ด๐๐ฟ๐ฒ, ๐ป๐ผ๐ ๐ท๐๐๐ ๐ฎ๐ฝ๐ฝ๐ฟ๐ผ๐๐ฒ. If you're deferring to someone more technical than you on strategic AI calls, you're reacting to a market someone else is reading in real time.
๐ช๐๐๐ง ๐๐๐ง๐จ๐๐๐๐ฌ ๐ฆ๐๐ข๐ช๐ฆ ๐๐ก๐ง๐๐ฅ๐ฃ๐ฅ๐๐ฆ๐๐ฆ
๐ง๐ต๐ฒ ๐ฟ๐ฒ๐ฎ๐น ๐ธ๐ถ๐น๐น๐ฒ๐ฟ๐: procurement cycles, security review, SOC 2 / HIPAA / FedRAMP regimes, change management, and RAG that doesn't scale to real corpus volume. Most enterprises are managing AI like a side project inside a pre-AI codebase. ๐ง๐ต๐ฒ ๐ฝ๐ฟ๐ผ๏ฟฝ๏ฟฝ๐๐ฐ๐๐ถ๐ผ๐ป ๐ฏ๐ฟ๐ฒ๐ฎ๐ธ๐ ๐ฎ๐ฟ๐ฒ ๐๐ต๐ฒ ๐๐๐บ๐ฝ๐๐ผ๐บ, ๐ป๐ผ๐ ๐๐ต๐ฒ ๐ฑ๐ถ๐๐ฒ๐ฎ๐๐ฒ.
๐ง๐ต๐ฒ ๐ณ๐ถ๐ ๐ถ๐๐ป'๐ ๐ฎ๐ป๐ผ๐๐ต๐ฒ๐ฟ ๐ฝ๐ถ๐น๐ผ๐. Stand up a dedicated AI review track so security and procurement don't restart from zero on every vendor eval. Put a single C-suite owner on AI budget with authority to spend, not a committee. Build AI-native services parallel to the legacy codebase, not bolted onto it.
๐ง๐ต๐ถ๐ ๐ถ๐ ๐๐ต๐ฒ๐ฟ๐ฒ ๐๐ฟ๐ผ๐๐๐ผ๐๐ฒ๐ฟ ๐๐ฝ๐ฒ๐ป๐ฑ๐ ๐บ๐ผ๐๐ ๐ผ๐ณ ๐ถ๐๐ ๐๐ถ๐บ๐ฒ. We score AI Capability and AI Resilience across PE portfolios and operators, and most organizations are further behind than they think.
๐ง๐๐ ๐ช๐ฅ๐๐ฃ๐ฃ๐๐ฅ ๐๐๐ข๐ข๐๐๐๐ง๐
The majority of companies marketed as "AI-native" are repackaged base models with a UI on top. ๐๏ฟฝ๏ฟฝ๏ฟฝ๐๐ป๐ฑ๐ฒ๐ฟ๐ ๐ฎ๐ป๐ฑ ๐ถ๐ป๐๐ฒ๐๐๐ผ๐ฟ๐ ๐ป๐ฒ๐ฒ๐ฑ ๐๐ผ ๐ด๐ฒ๐ ๐บ๐๐ฐ๐ต ๐๐ต๐ฎ๐ฟ๐ฝ๐ฒ๐ฟ ๐ฎ๐ ๐๐ฒ๐น๐น๐ถ๐ป๐ด ๐๐ฟ๐๐ฒ ๐บ๐ผ๐ฎ๐๐ ๐ณ๐ฟ๐ผ๐บ ๐ฎ ๐ต๐ฒ๐ฎ๐ฑ ๐๐๐ฎ๐ฟ๐ ๐ฑ๐ฟ๐ฒ๐๐๐ฒ๐ฑ ๐๐ฝ ๐ฎ๐ ๐ฑ๐ฒ๐ณ๐ฒ๐ป๐๐ถ๐ฏ๐ถ๐น๐ถ๐๐. Some wrappers will monetize fast and sell to strategics in the next 2-4 years. Most won't. Two forces compress the field: AI participation costs (infrastructure plus token burn) keep rising, and customer wallet share for AI tooling is finite. Only so many winners emerge per category. ๐ง๐ต๐ฒ ๐ฟ๐ฒ๐๐ ๐ฟ๐ฒ๐๐๐ฟ๐ป ๐ฐ๐ฎ๐ฝ๐ถ๐๐ฎ๐น ๐ฎ๐ ๐ฏ๐ฒ๐๐, ๐๐ฒ๐ฟ๐ผ ๐ฎ๐ ๐๐ผ๐ฟ๐๐.
๐ช๐๐ข ๐ฆ๐จ๐ฅ๐ฉ๐๐ฉ๐๐ฆ
Three archetypes make it through the cycle.
Vendors with proprietary data, engineered workflows, regulated-vertical lock-in, or hardware-attached integrations deep enough that a native platform release can't replace them. SaaS incumbents that rebuild before their slot gets benchmarked out of the stack. Founders fluent enough to dictate AI strategy, not delegate it. Configured, not approved. ๐๐ณ ๐๐ผ๐ ๐ฐ๐ฎ๐ป'๐ ๐ฎ๐ฟ๐๐ถ๐ฐ๐๐น๐ฎ๐๐ฒ ๐ต๐ผ๐ ๐๐๐ ๐ ๐ณ๐ถ๐ ๐๐ผ๐๐ฟ ๐ฝ๐ฟ๐ผ๐ฑ๐๐ฐ๐ ๐ถ๐ป ๐๐ป๐ฑ๐ฒ๐ฟ ๐ฑ ๐บ๐ถ๐ป๐๐๐ฒ๐, ๐๐ผ๐'๐ฟ๐ฒ ๐ป๐ผ๐ ๐ถ๐ป๐๐ฒ๐๐๐ฎ๐ฏ๐น๐ฒ.
Email [email protected] with "investor checklist" or "founder checklist" and I'll send you the 20 questions every investor or every founder should be asking right now.
In August I wrote a thesis I never published. The funds I was warning were key Crossover Research clients, so I stayed quiet. Since then, ๐ฆ๐ผ๐ณ๐๐๐ฎ๐ฟ๐ฒ ๐บ๐๐น๐๐ถ๐ฝ๐น๐ฒ๐ ๐ฎ๐ฟ๐ฒ ๐ฑ๐ผ๐๐ป ๐ฑ๐ฌ%+. Salesforce $CRM, ServiceNow $NOW, Adobe $ADBE, Workday $WDAY all off 40% from highs. Thomson Reuters $TRI dropped 16% in a single session on the Anthropic legal agent launch. The SaaSpocalypse arrived. So here's the follow-up. Not commentary on what happened, but where I think this goes next.
Most vertical SaaS companies aren't underperforming because their software is bad. ๐ง๐ต๐ฒ๐'๐ฟ๐ฒ ๐๐ป๐ฑ๐ฒ๐ฟ๐ฝ๐ฒ๐ฟ๐ณ๐ผ๐ฟ๐บ๐ถ๐ป๐ด ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐๐ต๐ฒ๐ ๐ป๐ฒ๐๐ฒ๐ฟ ๐ฏ๐๐ถ๐น๐ ๐๐ต๐ฒ ๐๐ฒ๐ฐ๐ผ๐ป๐ฑ ๐ฏ๐๐๐ถ๐ป๐ฒ๐๐. And the first business is under attack. For twenty years, one of the biggest SaaS moats was engineering complexity: deep technical talent, long roadmaps, compounding codebases that were genuinely hard to replicate. ๐๐ ๐๐ฝ๐ฒ๐ป๐ฑ๐ฒ๐ฑ ๐๐ต๐ฎ๐ ๐ฎ๐น๐บ๐ผ๐๐ ๐ผ๐๐ฒ๐ฟ๐ป๐ถ๐ด๐ต๐.
Product development is democratizing to operators with no code background but strong product vision. Look at Anthropic: they've built the engine and are shipping lookalike products at a cadence that would have taken a legacy SaaS vendor three years of roadmap, with a fraction of the headcount. That pace can kill legacy businesses overnight.
๐๐ณ ๐๐ต๐ฒ ๐ฒ๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ถ๐ป๐ด ๐บ๐ผ๐ฎ๐ ๐ถ๏ฟฝ๏ฟฝ๏ฟฝ ๐ด๐ผ๐ป๐ฒ, ๐ณ๐ผ๐๐ฟ ๐บ๐ผ๐ฎ๐๐ ๐ฟ๐ฒ๐บ๐ฎ๐ถ๐ป: ๐ฑ๐ถ๐๐๐ฟ๐ถ๐ฏ๐๐๐ถ๐ผ๐ป, ๐ฝ๐ฟ๐ผ๐ฝ๐ฟ๐ถ๐ฒ๐๐ฎ๐ฟ๐ ๐ฑ๐ฎ๐๐ฎ, ๐๐ผ๐ฟ๐ธ๐ณ๐น๐ผ๐ ๐ฏ๐ฟ๐ฒ๐ฎ๐ฑ๐๐ต, ๐ฎ๐ป๐ฑ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐ถ๐ป๐๐๐น๐ฎ๐๐ถ๐ผ๐ป. The first three are moats the company builds. The fourth is a moat the company captures, and it's the one most resistant to AI disruption.
๐ฅ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐ฐ๐ผ๐บ๐ฝ๐น๐ฒ๐ ๐ถ๐๐ ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ๐ ๐๐๐ถ๐๐ฐ๐ต๐ถ๐ป๐ด ๐ฐ๐ผ๐๐๐ ๐๐ต๐ฎ๐ ๐ต๐ฎ๐๐ฒ ๐ป๐ผ๐๐ต๐ถ๐ป๐ด ๐๐ผ ๐ฑ๐ผ ๐๐ถ๐๐ต ๐ฝ๐ฟ๐ผ๐ฑ๐๐ฐ๐ ๐พ๐๐ฎ๐น๐ถ๐๐. Once a vendor is embedded in a compliance workflow, ripping them out means re-attesting, re-auditing, and re-certifying every downstream process. The buyer isn't paying for software, they're paying for the accumulated paper trail. Tyler Technologies ($TYL) is the clearest version of the pattern. State and local government software across courts, public safety, assessment, and ERP. Every module is married to statutory process, FIPS, CJIS, audit trails, and procurement cycles that take years. TYL is down 42% TTM and 2026 guidance came in soft, but the moat didn't break. Revenue still compounded, and government procurement runs on five-year cycles, not five-week news cycles. Veeva is the sharper version. Revenue up 16% in FY26, Q4 beat, the stock still down 25%. The market is selling execution, not weakness. Guidewire in P&C insurance, where regulatory filings and rate approvals anchor the stack, sits in the same setup: still compounding ARR, still winning cloud conversions, multiple reset anyway. Same pattern across all three: multiples compressed, fundamentals intact. The moat is the regulatory surface area itself, and it compounds because the rules get more complex, not less.
๐ ๐๐ฎ๐ ๐น๐ผ๐ป๐ด ๐ฃ๐ฎ๐น๐ฎ๐ป๐๐ถ๐ฟ ๐ฎ๐ $๐ญ๐ฏ (read that here: https://t.co/0N0oIX8N87). ๐ก๐ผ๐ ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐๐ต๐ฒ ๐บ๐ผ๐ฑ๐ฒ๐น ๐ผ๐ฟ ๐๐ต๐ฒ ๐๐ผ๐ผ๐น๐ถ๐ป๐ด. ๐๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐๐ต๐ฒ ๐ผ๐ป๐๐ผ๐น๐ผ๐ด๐. Palantir is the proprietary-data version of the regulatory thesis. Once Palantir sits between the customer and their own data, ripping it out means rebuilding the data model from scratch. Snowflake and Databricks never had that entrenchment layer. AIP bootcamps then turned the data moat into a distribution moat: 660 bootcamps in a single quarter, 94% y/y US customer deal growth, bookings at 1.9x sales. Own the data, ship functional AI on top of it, let the GTM compound. Every vertical incumbent has a version of this available. The question is whether they'll build it before a challenger does.
But regulatory insulation is necessary, not sufficient. Plenty of vendors inside regulated verticals are still getting squeezed because they never became AI-native. BlackLine ($BL) and Trintech are feeling it in close and reconciliation as Numeric, Maximor, and Stacks build AI-native from day one. nCino ($NCNO) in banking faces the same challenge. The regulatory moat buys you time. It doesn't buy you the decade.
๐ง๐ต๐ฒ ๐๐ถ๐ป๐ป๐ถ๐ป๐ด ๐ณ๐ผ๐ฟ๐บ๐๐น๐ฎ ๐ถ๐ ๐ฑ๐ฎ๐๐ฎ ๐ผ๐ฟ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐ ๐๐๐ฟ๐ณ๐ฎ๐ฐ๐ฒ ๐ฎ๐ฟ๐ฒ๐ฎ ๐ฝ๐น๐๐ ๐ณ๐๐ป๐ฐ๐๐ถ๐ผ๐ป๐ฎ๐น ๐๐, ๐ป๐ผ๐ ๐ผ๐ป๐ฒ ๐ผ๐ฟ ๐๐ต๐ฒ ๐ผ๐๐ต๐ฒ๐ฟ. Look at why Claude is winning. Anthropic isn't competing on model benchmarks, they're competing on functional workflow. Building for the user, not the leaderboard. That's the playbook vertical incumbents need to run. Take the moat you already have, whether it's regulatory or data-entrenchment, layer genuine workflow AI on top, and the challenger can't catch you. The vendors that do both win the decade. The ones that rely on inertia alone get caught. The ones that ship AI without an anchor get commoditized. You need both.
๐ง๐ต๐ฒ ๐ฏ๐๐๐ฒ๐ฟ ๐ถ๐ ๐๐ฒ๐น๐น๐ถ๐ป๐ด ๐๐ผ๐ ๐๐ต๐ถ๐ ๐ฝ๐น๐ฎ๐ถ๐ป๐น๐. A study we ran with Battery Ventures on AI adoption in the Office of the CFO (https://t.co/xBEMSF8Y72) surveyed 129 finance leaders at companies from $50M to $5B+ in revenue. 77% said they want to uplevel existing systems with AI from new vendors that layer onto existing systems. Only 15% want to replace their current system of record with an AI-native platform. The incumbent wins if they ship AI. The AI-native challenger wins only if the incumbent doesn't.
The signal shows up in our VoC data too. In regulated verticals, mission criticality scores cluster above 9, and NPS doesn't track satisfaction, it tracks switching friction. Customers will tell you the product is mediocre and still score it 9 on "would not switch" because the compliance team vetoes any alternative. ๐ง๐ต๐ฎ๐'๐ ๐๐ต๐ฒ ๐๐ถ๐ด๐ป๐ฎ๐๐๐ฟ๐ฒ ๐ผ๐ณ ๐ฎ ๐ฐ๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ฐ๐ฒ-๐ถ๐ป๐๐๐น๐ฎ๐๐ฒ๐ฑ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ, ๐ฎ๐ ๐น๐ผ๐ป๐ด ๐ฎ๐ ๐๐ต๐ฎ๐ ๐๐ฒ๐ป๐ฑ๐ผ๐ฟ ๐ถ๐ ๐ฎ๐ฐ๐๐ถ๐๐ฒ๐น๐ ๐๐ต๐ถ๐ฝ๐ฝ๐ถ๐ป๐ด ๐ฎ๐ด๐ฎ๐ถ๐ป๐๐ ๐๐ต๐ฒ ๐๐ ๐ฐ๐๐ฟ๐๐ฒ.
Which brings us back to the second business for everyone outside the regulated or data-entrenched moat. Seat ARR got them to $100M. But with the shift to agentic workforce structures, partial human capital replacement, and pricing pressure compressing margins, the traditional SaaS model has to transform fast. The next $500M comes from monetizing the installed base: marketplace rake on demand they generate for their own customers, capital products underwritten by their own transaction data, supplier monetization, brand partnerships, group buying. The assets are already sitting there. Captive SMB audience. Proprietary transaction and behavioral data. A distribution pipe (the UI itself) that delivers new products at near-zero CAC.
๐ช๐ต๐ฎ๐'๐ ๐บ๐ถ๐๐๐ถ๐ป๐ด ๐ถ๐ ๐ผ๐ฟ๐ด๐ฎ๐ป๐ถ๐๐ฎ๐๐ถ๐ผ๐ป๐ฎ๐น ๐๐ถ๐น๐น. Monetizing the installed base requires a different org than the one that got you to scale. Different GTM, P&L optics, and talent. Founders and boards under-invest because year one looks worse before it looks better, and public markets punish any SaaS multiple that starts to look like fintech or marketplace. So the second business never ships. The round prices in the optionality. The multiple compresses. The exit underwhelms.
๐ง๐ต๐ฟ๐ฒ๐ฒ ๐ฑ๐ถ๐น๐ถ๐ด๐ฒ๐ป๐ฐ๐ฒ ๐พ๐๐ฒ๐๐๐ถ๐ผ๐ป๐ ๏ฟฝ๏ฟฝ๏ฟฝ๐ผ๐ ๐ฒ๐ป๐ผ๐๐ด๐ต ๐ถ๐ป๐๐ฒ๐๐๐ผ๐ฟ๐ ๐ฎ๐ฟ๐ฒ ๐ฎ๐๐ธ๐ถ๐ป๐ด:
๐ญ. ๐ช๐ต๐ฎ๐ ๐ฝ๐ฒ๐ฟ๐ฐ๐ฒ๐ป๐ ๐ผ๐ณ ๐ฟ๐ฒ๐๐ฒ๐ป๐๐ฒ ๐ฐ๐ผ๐บ๐ฒ๐ ๐ณ๐ฟ๐ผ๐บ ๐๐ผ๐๐ฟ๐ฐ๐ฒ๐ ๐ผ๐๐ต๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐๐๐ฏ๐๐ฐ๐ฟ๐ถ๐ฝ๐๐ถ๐ผ๐ป ๐ฎ๐ป๐ฑ ๐ฝ๐ฎ๐๐บ๐ฒ๐ป๐ ๐ฝ๐ฟ๐ผ๐ฐ๐ฒ๐๐๐ถ๐ป๐ด? Under 5%, they haven't started. 10 to 20%, thesis is live. Over 20%, it's working.
๐ฎ. ๐๐ผ๐ ๐ต๐ฎ๐ฟ๐ฑ ๐๐ผ๐๐น๐ฑ ๐ถ๐ ๐ฏ๐ฒ ๐๐ผ ๐ฟ๐ฒ๐ฐ๐ฟ๐ฒ๐ฎ๐๐ฒ ๐๐ต๐ถ๐ ๐ฐ๐ผ๐บ๐ฝ๐ฎ๐ป๐ ๐ณ๐ฟ๐ผ๐บ ๐๐ฐ๐ฟ๐ฎ๐๐ฐ๐ต ๐๐ถ๐๐ต ๐๐ ๐๐ผ๐ฑ๐ฎ๐? If a well-funded team with Claude and six engineers could rebuild the functional product in nine months, the software isn't the moat. The moat has to live somewhere else: proprietary data, a network, integrations, or regulatory surface area the challenger can't clear. If you can't point to at least one, you're underwriting a melting ice cube.
๐ฏ. ๐ช๐ต๐ฎ๐ ๐ฝ๐ฒ๐ฟ๐ฐ๐ฒ๐ป๐ ๐ผ๐ณ ๐๐ต๐ฒ ๐ฏ๐๐๐ฒ๐ฟ'๐ ๐๐๐ถ๐ฐ๐ธ๐ถ๐ป๐ฒ๐๐ ๐ถ๐ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ผ๐ฟ๐, ๐ฎ๐ป๐ฑ ๐๐ต๐ถ๐ฐ๐ต ๐๐ฎ๐ ๐ถ๐ ๐๐ต๐ฒ ๐ฟ๐๐น๐ฒ ๐๐ฒ๐ ๐บ๐ผ๐๐ถ๐ป๐ด? A regulatory moat evaporates if the regulation simplifies. Underwrite the direction of travel, not just the current state.
๐๐ป๐ฑ ๐๐ต๐ฒ ๐ฐ๐น๐ผ๐ฐ๐ธ ๐ถ๐ ๐๐ถ๐ด๐ต๐๐ฒ๐ฟ ๐๐ต๐ฎ๐ป ๐บ๐ผ๐๐ ๐ฟ๐ฒ๐ฎ๐น๐ถ๐๐ฒ. Retention in enterprise SaaS has largely been defined by the pain of systems replacement, not genuine moat. If the stickiness isn't backed by proprietary data, a harvesting flywheel, or regulatory surface area, those vendors are about to get disrupted. Pure seat-based pricing is dying unless vendors embrace agent-seat models, and LLM providers have been subsidizing the market on token cost, with recent pricing shifts signaling cash reserves aren't infinite.
๐๐ฒ๐ฟ๐ฒ'๐ ๐๐ต๐ฒ ๐๐ป๐ฑ๐ฒ๐ฟ๐ฎ๐ฝ๐ฝ๐ฟ๐ฒ๐ฐ๐ถ๐ฎ๐๐ฒ๐ฑ ๐ฝ๐ผ๐ถ๐ป๐: ๐๐-๐ป๐ฎ๐๐ถ๐๐ฒ ๐ฐ๐ผ๐บ๐ฝ๐ฒ๐๐ถ๐๐ผ๐ฟ๐ ๐ต๐ฎ๐๐ฒ ๐๐ผ๐ฟ๐๐ฒ ๐ด๐ฟ๐ผ๐๐ ๐บ๐ฎ๐ฟ๐ด๐ถ๐ป๐ ๐๐ต๐ฎ๐ป ๐ฆ๐ฎ๐ฎ๐ฆ ๐ถ๐ป๐ฐ๐๐บ๐ฏ๐ฒ๐ป๐๐, ๐ป๐ผ๐ ๐ฏ๐ฒ๐๐๐ฒ๐ฟ. Inference costs haven't collapsed, and burning VC cash to subsidize unit economics is a bridge, not a business model. The incumbents should be winning on P&L. They're losing on product velocity and AI-readiness. That's a solvable problem if the board has the will to ship. Vendors without a second business, without a data moat, and without regulatory insulation will still lose, despite having better margins than their AI-native challengers. Customers switch on features and speed, not on unit economics.
๐๐ป๐๐ฒ๐ฟ๐ฝ๐ฟ๐ถ๐๐ฒ ๐ฎ๐ป๐ฑ ๐ฟ๐ฒ๐ด๐๐น๐ฎ๐๐ฒ๐ฑ ๐๐ฒ๐ฟ๐๐ถ๐ฐ๐ฎ๐น๐ ๐ฎ๐ฟ๐ฒ ๐๐ต๐ฒ ๐น๐ฎ๐๐ ๐๐ฎ๐ณ๐ฒ ๐ต๐ฎ๐ฟ๐ฏ๐ผ๐ฟ, ๐ฎ๐ป๐ฑ ๐ผ๐ป๐น๐ ๐ฏ๐ฒ๐ฐ๐ฎ๐๐๐ฒ ๐ผ๐ณ ๐ฑ๐ฎ๐๐ฎ ๐ฏ๐ฟ๐ฒ๐ฎ๐ฑ๐๐ต ๐ฎ๐ป๐ฑ ๐ฐ๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ฐ๐ฒ. Everywhere else, the premium is about to get competed away. Any fund underwriting vertical SaaS exposure right now should be asking the second-business question before the next check clears. DM me, email me [email protected], or let's chat about your portfolio/underwriting process (https://t.co/muMNtk6ssk).
https://t.co/ElZm7vjalx
@TMTBreakout Good to be back. I've been in the lab building some things that are going to transform institutional research/diligence across private/public markets. Will get you a login when I release that in the next few weeks.
Including the full August 2025 tweet that I never posted below.
TLDR: I argued that the AI boom was structurally unsustainable: too much capital, too many undifferentiated โwrappers,โ and foundation models burning billions to subsidize usage. Since then, the bubble has inflated even further - AI firms pulled in the majority of global VC in 2025, while leaders like OpenAI blew past a $10โ12B revenue run-rate and still lost staggering amounts of money. Yet instead of a disciplined reset, investors have largely doubled down: lateโstage checks are still flowing into capitalโintensive platforms, mediocre appโlayer companies are limping through downโrounds, and only a thin slice of obviously broken wrappers is being culled. The early shakeout is real, but the capital allocation remains sloppy - far more driven by FOMO around a few brandโname winners than by rigorous views on unit economics or durable moats.
The AI Investment Supercycle Hypothesis - Mon, Aug 11, 2025
Hypothesis: The current boom in AI startups and funding is unsustainable. It will likely culminate in a severe capital crunch and mass consolidation or failure of AI companies, especially affecting growth-stage investors. Below we break down the thesis with supporting data and consider counterarguments.
Excess Company Count vs. Finite Market
Explosion of AI Startups: There are tens of thousands of AI-focused companies worldwide (~70,000 AI startups globally). Many of these are โAI-enabledโ software vendors whose products often rely on the same handful of AI models (e.g. OpenAIโs GPT). This surge echoes the dot-com era: โAI-powered is the new .com,โ with countless lookalike startups pitching similar ideas.
Limited Revenue Pool: The total AI hardware/software market is projected at $780โ$990โฏbillion by 2027. Yet venture capital and corporate investors have already poured well over $500โ600โฏbillion into AI companies to date . To justify these investments with typical 10ร returns, AI startups would need to generate on the order of $5โ6 trillion in revenue โ about 6ร the entire expected market size. In other words, the math doesnโt add up: there simply isnโt a large enough revenue pie for all these companies at their current lofty valuations. As Sequoia Capitalโs David Cahn noted, the gap between AI investment and revenue has ballooned into a โstaggering $500 billion annual revenue gapโ that must be filled to justify the spending .
Sky-High Valuations Assume Unprecedented Growth: Despite the limited market, funding has surged. Generative AI startups raised a record $56 billion from VCs in 2024 (885 deals) โ nearly double the prior year . Close to one-third of all VC funding worldwide went to AI in 2024 . These financings often came at inflated valuations (hundreds of millions or even billions), implying future dominance in their niches. Yet many have little proven revenue. An analysis in mid-2025 observed that โ90% of these โAI companiesโ are just expensive wrappers around the same five foundational models.โ In short, too many startups are chasing the same customers with undifferentiated tech, all while assuming theyโll each capture outsized revenue. This is structurally reminiscent of past bubbles.
Economics Inverted by Subsidies
Foundation Models Operating at Massive Loss: The core AI platforms (OpenAI, Anthropic, etc.) heavily subsidize AI compute costs to drive adoption. OpenAI, for example, generated about $4 billion revenue in 2024 but spent ~$9 billion to do so โ losing around $5 billion for the year. By one estimate OpenAI currently โspends $2.25 for every $1 it earnsโ. Similarly, Anthropic burned $5.6 billion in cash in 2024 while making well under $1 billion revenue . (Anthropic projects improving in 2025, but still expects to lose ~$3 billion on ~$3.7 billion revenue.) These eye-popping losses mean AI API prices are artificially low โ essentially subsidized by investors. The big model providers are keeping prices down to grab market share, even as they incur billions in operating losses.
Downstream โWrappersโ with No Moat: An estimated 30,000+ software companies are โAI wrappersโ that simply call these foundation-model APIs and repackage the output with a pretty interface . Because they all rely on the same underlying AI engines, their products are often interchangeable โ โno IP, no moatโฆ just a well-structured API call, some markup, and marketing.โ Many charge high subscription fees (say, $50โ100/month) for services that a savvy user could replicate with direct API calls for a few dollars . This works only as long as OpenAI/Anthropic keep API costs low. When the subsidies inevitably end (i.e. prices normalize upward to cover real costs), these downstream startupsโ economics collapse. Their entire model is built on thin margins. As one analysis put it, โevery token sent through a wrapper โ paid or not โ earns OpenAI moneyโฆ startups become unpaid distribution arms, subsidizing OpenAIโs growth while bleeding out.โ In other words, the house (OpenAI) always wins โ until the โwrappersโ run out of cash. At that point, thousands of these dependent products will either have to raise prices (driving away customers) or shut down.
The Commoditization Spiral: The combination of ubiquitous tech and underpriced service creates a vicious circle for most AI vendors: (1) If everyone uses the same few AI models, products become undifferentiated. (2) Competing on features is hard, so pricing wars ensue โ indeed, OpenAIโs latest GPT-4 price cuts undercut Anthropic by up to 7ร on cost per token, forcing others to match or lose business. (3) As prices per customer plummet, so do gross margins for any startup reselling AI. (4) Lower revenues make it impossible to support the previous valuations or to cover the still-high infrastructure costs (AI compute remains expensive, and energy/compute costs are not falling as fast as pricing). (5) With unit economics turned upside-down (more users actually increase losses), these companies cannot sustain operations without continual investor subsidies. This โrace to the bottomโ on price is great for end-users in the short term, but itโs lethal for the thousands of me-too vendors. Itโs analogous to the dot-com era of free services: eventually the money runs out.
Growth-Stage Capital at Extreme Risk
Late-Stage Funding Frenzy: Unlike early-stage VCs who place many small bets, growth equity investors have been writing big checks (often $25โ100 million each) into mid-stage AI companies at $400 million+ valuations. These rounds (Series B, C, D etc.) were justified by lofty growth assumptions and the fear of missing the โnext big thing.โ However, many of these startups have 8โ12 month runways due to high burn rates (expensive ML talent and cloud bills) โ meaning they will need another funding round by 2024โ2025. For example, in 2023โ24 numerous generative AI startups raised funds at unicorn valuations despite minimal revenue, and immediately ramped spending on AI infrastructure. โMost companies funded during the 2021โ2023 boom had 18โ36 months of capital,โ and many will run dry by late 2025 or early 2026 if they canโt refinance . The growth investors who led these big rounds will be left holding the bag if valuations reset.
High Failure Rates = High Write-Downs: Early-stage venture firms expect e.g. 6 or 7 out of 10 startups to fail โ their model tolerates it. Growth equity, in contrast, bets on a much lower loss rate (maybe 1 or 2 failures out of 10) because they deploy larger sums per deal. The current AI cycle is likely to betray those expectations. Industry observers predict over 90% of AI startups will fail within five years. Even before the recent frenzy, tech market indices showed sharp private valuation declines in 2022โ23 , and many AI firms that raised in 2024 have since missed milestones. If 80โ90% of funded AI companies ultimately go under, growth-stage funds with heavy AI exposure could see well over half their portfolio by value written off. In effect, billions in late-stage capital could evaporate. Estimates of the โdead moneyโ vary, but even a conservative scenario of 60โ70% startup failure would wipe out ~$400 billion (out of ~$600B invested), and a more realistic 80โ90% failure rate implies $500B+ lost. Indeed, PitchBook data show fundraising for AI has already dropped in 2024 (deal count down 42%) as reality sets in . Growth investors are slamming on the brakes, but it may be too late โ the capital is already in these companies, and many are running on fumes.
Compression of Exit Options: Another challenge for growth equity: who will buy or IPO these companies to provide an exit? The IPO window for tech is cautious in 2025, and the few public listings (e.g. Cloud, enterprise AI) have not delivered the kind of multiples needed. M&A is an option โ and indeed we may see rapid consolidation in 2025โ2027, with stronger players acquiring distressed startups for pennies on the dollar. But most acquirers will wait until valuations crumble. Funds that put $50M into a โnext-generation AI SaaSโ at a $500M valuation may recoup only a fraction in a fire sale. The timeline looks grim: 2025 will likely still see some aggressive fundraising and peak company counts, but by mid-2026 signs of saturation (slowing growth, rising customer-acquisition costs) will be undeniable. By 2027, as startups exhaust their last cash, we could witness a mass shutdown wave โ potentially thousands of AI companies closing within 12โ18 months . Growth equity portfolios will be forced to mark down failing investments (60โ90% losses in the worst cases). As one industry veteran wryly noted, โFunds with 40% of their book in AI might experience a 80% write-down in that slice โ itโll be historic.โ
Timeline: From Bubble to Shakeout
2021โ2023 โ Build-Up: Breakthroughs in generative AI (GPT-3, DALL-E, etc.) trigger a flood of startup creation and funding. Valuations skyrocket on hype. Investors cite โAI is the new electricity,โ and fear of missing out leads to overfunding of very early-stage projects . Many companies launch with little more than a demo or a fine-tuned model wrapper.
2024 โ Peak Froth: Funding reaches record levels (as noted, $56B VC dollars into genAI in 2024 ). By late 2024 and early 2025, AI headlines dominate tech. But underneath, cracks appear: infrastructure bottlenecks (GPUs), first reports of AI startups with negligible traction, and Big Tech (OpenAI, Microsoft, Google) racing ahead of the pack. The largest AI firms themselves remain unprofitable despite fast-growing revenue โ e.g. OpenAI doubled its ARR from $6B to $12B in the first half of 2025 (annualized run-rate), yet it continues to burn cash ($14B loss expected in 2025). This suggests even market leaders havenโt found efficient economics yet.
2025 โ Early Signs of Saturation: By mid-2025, the number of AI products on the market has exploded. Every sub-sector (coding assistants, AI content generators, chatbots for support, etc.) is crowded. Customer adoption, while real, cannot keep up with the supply of solutions. Anecdotally, sales cycles for B2B AI software start to lengthen as CIOs get fatigued by thousands of similar pitches. Customer acquisition cost (CAC) rises โ more effort needed to convince users who have already tried 5 different AI copywriters or coding copilots. Big Tech enters aggressively, bundling AI features into their platforms (often free or at low cost), undercutting standalone startups. Investors grow more selective, favoring startups with real differentiation or proprietary tech.
Late 2025 to 2026 โ The Crunch: This is when the โgravityโ of finite capital hits. Many startups that raised in the 2021โ22 boom face end of runway by late 2025 . Unfortunately, the funding environment now is much tougher โ interest rates are higher, and LPs (the investors in VC/Growth funds) are nervous about overexposure to AI. We can expect a sharp pullback in new funding for all but the top 5% of AI companies. The rest must either find an acquirer or drastically cut costs to survive. In mid-2026, weโll likely see a wave of down-rounds (companies raising capital at much lower valuations) and outright failures. Investor sentiment flips from FOMO to caution: as one VC noted, โthereโs far more scrutiny on unit economics and revenue tractionโ now . The mere mention of โAIโ no longer secures a premium โ in fact, hype-y startups are viewed with skepticism unless they have solid metrics.
2027 โ Mass Extinction Phase: By 2027, the global liquidity squeeze is in full effect. Earlier-stage VC funds may have the dry powder to prop up a few of their best bets, but growth equity and crossover investors (who fueled the largest rounds) largely retreat, nursing losses. Without new funding, thousands of AI startups will fold in a short period โ the โbursting of the AI bubble.โ This is analogous to the dot-com crash circa 2000โ2001, when countless internet startups went under. The survivors likely fall into two camps: (a)Infrastructure-level players (the big foundation model providers or cloud platforms โ many of whom are incumbent tech giants or heavily funded leaders like OpenAI), and (b)a handful of startups with truly defensible, domain-specific AI solutions (e.g. a company with a unique dataset or enterprise integration that gives it an edge in a niche). These survivors might consolidate the market โ mergers and acquisitions spike as the stronger firms acquire IP/talent from failed ones for pennies.
2028โ2029 โ Reset and Renewal: In the aftermath, the AI industry will likely look very different. Having shed the excess, the remaining companies can actually start to approach sustainable economics. With less crazy competition, pricing power returns for the winners โ e.g. API rates may rise to profitable levels once only a few providers dominate, and enterprise software firms that survived can charge more rational prices for clear value-add features. We may see public-market validation for a few big winners (think of how Amazon and Google emerged from the dot-com ashes). Meanwhile, many VC funds will report poor returns for their AI bubble-era cohorts, leading to a period of caution (and perhaps fewer new AI funds being raised for a while). In industry terms, this phase is healthy: it allows the real demand to catch up to the technology and for business models to mature without the distortion of easy money.
Endgame: Fewer Winners, Saner Market
When the dust settles, three forces likely hit simultaneously:
Valuation Collapse: Private and public market valuations for AI companies revert to levels based on fundamental metrics (revenue, margins) rather than hype. Multiples compress dramatically. For example, AI startups that raised at 100ร forward revenue might trade at 5โ10ร (in line with normal software firms) if they survive at all. This repricing can be swift and brutal โ weโve already seen some high-profile AI unicorns take down-rounds or markdowns in 2024โ25. The broad NASDAQ tech market fell ~33% in 2022 , but many private AI valuations could fall far more (70โ90% in some cases) before finding a floor. Investors essentially write off the bubble-era paper gains.
Mass Company Closures: As described, a huge percentage of AI startups will likely shut down within a yearโs span (the โmass extinctionโ). Weโre talking not just dozens but potentially thousands of companies disappearing. One mid-2025 report already warned that โover 90% of AI startups fail within five yearsโ. This winnowing will be painful for employees and investors in those firms, but it is the marketโs way of clearing out ventures that never found product-market fit or a path to profitability. Itโs worth noting this doesnโt mean the technology goes away โ often the IP or talent from failed startups is absorbed by larger players. But as stand-alone entities, most will be gone. Growth equity funds with heavy AI portfolios will have historically large loss ratios, as discussed.
Industry Reset & Sustainable Growth: With far fewer players, the survivors can capture larger shares of customer demand. Pricing will likely increase once subsidized competition fades โ e.g. the major AI cloud providers may raise API prices to finally earn a cloud-like margin on AI services, and surviving SaaS companies will focus on customers who are willing to pay for proven ROI. Weโll also see a narrowing of features to what customers actually use and value. During the hype phase, many startups added โAI featuresโ that were more gimmick than necessity. Post-shakeout, the focus will be on core uses that deliver business value (since those are the ones customers continue paying for). In effect, a handful of โmega-winnerโ companies (likely the cloud/platform giants and a few specialized firms) will dominate, and they will have learned how to make money on AI. Margins for these survivors could improve significantly due to reduced competition and higher efficiency. For example, if OpenAI and one or two peers end up providing 90% of global AI API calls, theyโll have the market power to charge profitable rates (unlike todayโs loss-making prices). In enterprise software, a few AI-enhanced incumbents (or well-capitalized startups) will bundle AI capabilities as part of larger offerings, enjoying upsell revenue without having to support a standalone 50-company ecosystem in each niche. In summary, the bubble subsidy will be gone, but the truly useful AI applications (and companies) will remain and thrive under more rational economics.
In essence, the industry will have undergone a Darwinian culling โ leaving a leaner ecosystem where: a) far fewer companies serve the real demand, b) capital investment is aligned with actual revenue potential, and c) infrastructure providers can operate profitably (no longer incentivized to burn $1 to make 50 cents).
Counterarguments & Considerations
Itโs important to acknowledge that this hypothesis is intentionally bearish and not universally held. There are more optimistic takes on the AI marketโs trajectory:
AI Market Could Expand More Than Expected: The forecasted ~$800Bโ$1T market by 2027 might prove too conservative. AI is a general-purpose technology that could spawn entirely new industries and revenue streams by 2030. Some proponents argue we are underestimating AIโs total addressable market โ citing, for instance, that AIโs total economic impact (including productivity gains) could be $15 trillion+ to global GDP by 2030. If even a fraction of that is captured as revenue, the pie might grow enough to support more companies (though likely not 70k startups). The analogy is the Internet: early projections in the โ90s didnโt foresee trillion-dollar markets in e-commerce, cloud, online advertising, etc. Could AI likewise surprise to the upside? Itโs possible that new killer apps (e.g. AI in healthcare, finance, etc.) will unlock revenue sources that justify some of the current investment.
Extraordinary Growth of Leaders: While most AI startups struggle, the winners are growing at jaw-dropping rates. OpenAIโs revenue run-rate jumping from ~$1B to $12B in roughly a year shows that demand for top-tier AI services is real and accelerating . Anthropic similarly went from near-zero to a $5B ARR in 2025 by focusing on enterprise coding assistants . If a few companies can indeed each capture tens of billions in revenue, the overall sectorโs revenue โceilingโ moves higher. This concentration of success might mean the total AI sector revenue in 2027โ2030 ends up far larger than the average of individual forecasts โ essentially, a power-law outcome where a handful of firms achieve what dozens of smaller players had hoped to. For growth equity investors, a single big win (say, backing the next Nvidia or the next Salesforce-level AI platform) could make up for many losses. Thus, the doomsday scenario for every investor isnโt assured; it depends on whether they picked any winners.
Strategic Value and Big Tech Support: Not all AI companies live and die by immediate unit economics. Some are being kept afloat by strategic partnerships or acquisitions. For example, cloud giants (Amazon, Microsoft, Google) have strong incentive to subsidize promising AI startups via cloud credits or investments, because those drive cloud usage and ensure the tech stays out of competitorsโ hands. We are seeing collaborations like Nvidia investing in certain AI startups or Microsoftโs multibillion funding of OpenAI, which indicate that big players will absorb huge costs to remain leaders . This means some of the heavily lossmaking AI firms might not face a hard stop in funding โ they could be acquired or continually bankrolled as strategic assets. In the endgame, one could envision a scenario where Big Tech effectively โacqui-hiresโ much of the AI startup talent/IP (at lower valuations), softening the blow of the bust. The survivors might mostly be divisions of larger companies.
Historical Precedent โ The Dot-Com Lesson: The dot-com bubble crash saw ~75% of internet companies fail, but those that survived (e.g. Amazon, eBay) went on to be monstrously successful, and the internet did indeed transform the economy. Analogously, even if 90% of current AI startups fail, the 10% that survive could form the backbone of the next decadeโs tech giants. From a consumer and societal perspective, the AI revolution will likely continue its momentum (AI adoption in business is still growing in 2024โ25 ). The โboom/bustโ cycle might just be a necessary phase of maturation. So a counterpoint is: yes, weโll see a painful consolidation, but no, itโs not the end of AI innovation or investment. It may actually be the beginning of a more stable growth phase, much like web 2.0 rose after the dot-com washout.
In summary, skeptics of the thesis would agree thereโs excess in the short term, but suggest the long-term opportunity of AI remains enormous. They argue the current shakeout is part of separating signal from noise. A few dominant platforms (possibly todayโs front-runners or yet-to-emerge dark horses) could justify the overall investment by eventually generating massive profits โ even if 90% of their contemporaries fail. Thereโs also the possibility that new waves of AI advancement (e.g. AGI or breakthrough applications) could reignite growth before a full bust occurs, prolonging the cycle.
Conclusion: A Probable Reckoning (with a Silver Lining)
The evidence strongly indicates that the AI sector is in a classic boom-to-bust cycle. Too many companies are chasing too little near-term revenue, propped up by an infusion of capital that cannot possibly see 10ร returns across the board. The unit economics for most AI startups are unsustainable โ many are effectively selling dollars for cents, subsidized by investor cash. And while innovators like OpenAI have achieved remarkable technological feats, even they have yet to prove a profitable business model under current pricing. All this suggests an inevitable shakeout: absent โcontinuous subsidizationโ by investors, the market will force a correction. We are likely to witness a wave of consolidations and failures in the next 1โ3 years that mirrors the dot-com collapse in scope. Growth-stage investors, in particular, are poised to absorb heavy losses as valuations normalize and weaker companies fold.
Importantly, this is not a thesis against AIโs significance โ itโs a reality check on AI as a business. The technology is revolutionary; the mistake is assuming every AI company will be. The math has the final say. As one observer neatly summarized: โToo many companies for the available spend, too much capital chasing too small a market, and too much dependency on unprofitable infrastructureโ โ something has to give. And that โsomethingโ will be the hundreds of AI startups that never had a viable path to profits.
What comes after the crunch? Likely a healthier, more mature industry. The survivors โ perhaps a few large-scale AI platforms and select specialized firms โ will benefit from reduced competition and clearer value propositions. With saner valuations, they can grow with realistic expectations and sustainable margins. For investors and founders, the coming storm will be painful, but it will also clear the way for the next phase of AI innovation grounded in real economics. In the long run, AI isnโt going anywhere; it will be as transformative as promised โ just not in the form of tens of thousands of unprofitable startups. The current thesis appears largely correct in diagnosing the excesses. The prudent move now is to adjust messaging and strategy accordingly: emphasize real use-cases and unit economics, prepare for tighter funding conditions, and focus on building or backing the few AI companies that can emerge on the other side of the capital crunch as true winners . The era of indiscriminate โAI hypeโ investment is winding down; what follows will separate the enduring players from the rest.
Love this take (I fully agree).
I basically rebuilt my CRM outside of Zoho, then put a sophisticated AI layer on top. That lets me activate the data in ways that arenโt possible inside traditional CRM walls, and the ability to customize anything and everything has been well worth it.
The platform is fully wired into our external services via API: CRM, database, LLM architecture, B2B enrichment vendors, financial suite, email, calendar, and more. I now have end-to-end visibility across every facet of the business, which was the original problem I was solving.
The front-end and import flows are still being finalized, so pardon the UI, but net-net, this gives me everything I need from one surface.
The problem with legacy vendors is that the pace is too slow. Time is everything, and I donโt want my operating speed constrained by someone elseโs product roadmap.