Metagame

Texas closed 106 deals at $6.24B in Q1, holding the #3 state position behind New York and California—Dallas led with 36, Austin 26, Houston 16. The Texas tech narrative isn't riding AI megarounds or chasing San Francisco's playbook; it's building a dependable third lane that doesn't need hype cycles to post volume. If your portfolio thesis bakes in concentration risk or narrative momentum, is "dependable and durable" the position you're underweighting?

Horizontal SaaS is down 35% over the past 12 months while vertical SaaS holds essentially flat at +3%, per Redpoint's 2026 Market Update — the application layer is rotating away from generic productivity and toward industry-specific AI-native tools. The megarounds are still chasing scale, but the sharpest Series A and B capital is now chasing depth in one industry. Which deck are you holding?

Enterprise AI budgets just hit an average of 207 million dollars per company while 42 percent have already ditched at least one AI project and 79 percent are actively struggling to make it work. The market isn't retreating though - it's splitting hard between a tiny group of companies that can clear the new security and governance gates and everyone else bleeding budget on pilots that never scale. Wildest shift: 63 percent of buyers now require human validation of AI outputs, up from just 22 percent a year ago, which means the fully autonomous agent pitch that dominated early 2025 is now basically a dealbreaker.

KKR just filed 14.7 billion dollars across two vehicles in Q1 and then dropped over 10 billion to build Helix — a platform designed to own the actual physical infrastructure of AI before most VCs even realized the game changed. While traditional venture was still writing Series B checks into software companies, private equity locked down data centers, power grids, and is now negotiating with Google to become the direct distribution layer for AI models into their portfolio companies. This isn't PE competing with VC anymore — it's PE rewriting what the capital stack even looks like, and if your fund strategy was built for a world where 500 million was considered large, you're not just outgunned, you're playing a sport that got reclassified.

While everyone obsessed over OpenAI's $122 billion raise, $4.3 billion quietly flowed through SEC filings into eight companies nobody talked about—inference chips, data infrastructure, AI hiring platforms—the actual plumbing that makes frontier AI work. Marvell raised $2 billion for custom silicon, MatX got $530 million for inference optimization, and Mercor landed $330 million targeting the skills gap that's killing 95% of enterprise AI pilots. The mega-rounds got the headlines but the SEC Form D tape just showed you what smart money is actually buying: the companies OpenAI can't operate without.

Pharma deal flow didn't collapse 62.5 percent last quarter — it just stopped filing as pharma. AI-native drug discovery companies like Recursion and Insitro are filing under "Other Technology" because their value lives in computational platforms, not clinical pipelines, which means healthcare funds tracking SEC codes are literally missing deals before they hit the committee. Total healthcare-adjacent deal count stayed completely flat at 74 deals, but if your allocation strategy runs on sector labels instead of manual tagging, you're benchmarking against a category that no longer describes where the actual capital is flowing.

NYC just closed 119 venture deals worth $70.4 billion in Q1 while San Francisco proper did 39 deals at $4.7 billion, and somehow the entire VC world is still pretending you have to be in the Bay to access deals. The gap isn't even close—New York has a 65% deal-count advantage, and here's the kicker: Atlanta matched San Francisco at 39 deals each. Everyone's optimizing for geographic proximity to AI labs while SEC Form D data shows the actual repeatable deal velocity is happening 3,000 miles east, but nobody's updated the conference talking points yet.

That $300 billion Q1 venture number everyone freaked out about? Strip out four mega-deals – OpenAI, Anthropic, xAI, Waymo – and the real market was 586 companies raising a median of $22.9 million, running 13 deals per business day while you were staring at the wrong scoreboard. Horizontal SaaS just got crushed with 35% down multiples while vertical software held strong, and now OpenAI and Anthropic are literally trying to buy implementation firms because the bottleneck moved from AI models to actually deploying them inside companies.

One company just raised $16.6 billion and accidentally became 57% of an entire tech sector category tracked by the SEC. Strip out xAI's single filing and "Other Technology" doesn't just shrink — it collapses from $29 billion to $12.5 billion, and the average deal size drops from $164 million to $71 million. If you're building investment models or AI benchmarks off those aggregate numbers, you're not tracking a market — you're tracking one Palo Alto address with a spreadsheet wrapped around it.

VCs just deployed $300 billion into AI this quarter while 95% of enterprise AI pilots delivered literally zero P&L impact and 42% of companies abandoned most of their projects. The gap between what enterprises are spending and what they're actually able to operationalize has never been wider in software history, but venture funds are still marking billion-dollar valuations like the 95% failure rate doesn't exist. Half the market is writing massive checks into infrastructure that's printing money while the application layer starves—and most LPs have no idea which side of that bifurcation their portfolio is actually on.