Megafunds are chasing defense factories and AI pipes today. The question is who, exactly, got to set the price. Here's how we got here. AI venture money has been clustering around foundational-model talent, infrastructure, applied tooling, agent security, and sovereign AI. The last big marker before today's show was Yann LeCun and former Gemini co-technical lead Oriol Vinyals joining 224 Ventures, a new early-stage AI fund with more than $100 million in assets, after a run of big rounds in AI security, chips, and enterprise agents. This is Startup Fundraising. We've got a defense manufacturer, a stealth chip startup debuting after a Series C, and Google reportedly shopping for an acqui-hire. Hadrian first. We'll keep tracking this story — AI venture market concentration. Follow the show so the next update finds you. Julie Bort, writing in TechCrunch:
Defense tech Hadrian announced on Thursday that it had raised a fresh $1.37 billion round at a valuation of $7.87 billion from a mile-long list of well-known investors. The lead investors in the round include WCM Investment Management, Washington Harbour Partners, Valor Equity Partners, 137 Ventures, and Baillie Gifford.
Hadrian raised $1.37 billion at a $7.87 billion valuation, with five listed leads. WCM, Washington Harbour, Valor, 137, Baillie Gifford—at that point, I want to know who actually set the price and terms, versus who agreed not to ruin the group photo. At least Hadrian makes physical parts for submarines. It opened its fourth facility in Alabama in March, and that public-private project was pegged at $2.4 billion. There's a real factory plan under this number. But the capital stack has changed fast. Founders Fund and Lux led a $260 million Series C about a year ago; now PitchBook estimates roughly $2 billion raised. The VC funds, asset managers, and private-capital crowd all need an exit eventually—preferably on a timetable defense manufacturing can actually meet. Exactly. An $8 billion mark asks a machine shop serving defense primes to meet growth-fund expectations too. Hadrian needs automated plants cranking out parts, contracts converting, and margins holding up—not another patriotic PowerPoint. From Maria Deutscher at SiliconANGLE:
Lumilens Inc., a provider of optical networking chips for artificial intelligence clusters, launched today with $900 million in funding. The company raised the bulk of the capital, $700 million, through a Series C round that valued it at $5.51 billion.
The money keeps concentrating around AI. Lumilens launches with $900 million, and $700 million of that came through a Series C at a $5.51 billion valuation. A company can launch as a Series C now? Fine—then account for the other $200 million, and name the data-center operator actually taking these optical chips at scale. “Billions in orders” sounds great; orders aren't shipped silicon. Atreides, Bain Capital Ventures, Meritech, Seligman, and Spark jointly led the round. Five lead investors can mean deep conviction. It can also mean pricing by committee, with everyone agreeing the spreadsheet is somebody else's problem. Ankur Singla sold Contrail and Volterra for a combined $676 million, so investors are buying a proven networking founder. But Lumilens is valued at eight times that exit history before we've seen who turns those chip orders into deployed racks. From Dealroom:
Mechanize, a San Francisco startup that trains AI agents to write software, raised a $9.1 million seed round in April 2026 at a $500 million valuation. Roughly 103 days later, Google is in talks to pay more than $1.5 billion for its technology and team, according to Business Insider.
Mechanize raised $9.1 million at a $500 million valuation in April; 103 days later, Google is reportedly discussing more than $1.5 billion. Seed pricing is becoming a very aggressive trailer for the exit. For a 35-person company, Google may be paying $1.5 billion to patch a coding-agent problem after Gemini kept slipping against Codex and Claude Code. What did Mechanize build in 103 days that Google couldn't recruit around? And the structure matters: sources say it's hires plus a non-exclusive technology licence, rather than a clean company purchase. The seed backers included Nat Friedman, Patrick Collison, and Dwarkesh Patel—but a headline valuation and a non-exclusive licence don't distribute value the same way. Google ran a version of this with Windsurf for about $2.4 billion last July. Call it talent acquisition if you want. At these prices, every AI lab is turning recruiting into an M&A event. Here's Inc42:
InRisk Labs has raised $27 Mn in a Series A funding round, while its subsidiary EarthRe has secured an IFSCA reinsurance licence. The funding and licence position InRisk Labs to scale AI-led reinsurance offerings as demand for climate-risk and catastrophe coverage grows in India.
Finally, a $27 million round where the money has a clear job. It goes toward underwriting and actuarial work, catastrophe models, and regulatory capital. EarthRe's IFSCA license means InRisk can actually take reinsurance risk instead of just selling AI slides about it. Bessemer and Northpoint co-led the round, but the license is the more consequential line item here. EarthRe is now the first incorporated reinsurer licensed from GIFT City, giving the group a regulated operating arm alongside InRisk's technology stack. India's reinsurance market is pegged at $10 billion, and climate risk creates actual demand—not a made-up TAM. Now EarthRe has to prove its catastrophe models can price floods, crop losses, marine claims, and everything else before one ugly season reprices them. Parametric coverage can pay automatically when a predefined weather event hits. That can be useful, but the contract design matters enormously. The $27 million is modest beside the mega-rounds we just covered—and much more tethered to a regulated balance-sheet build. If you follow startup fundraising, you may also like Infrastructure Secondaries Daily. It covers LP stake sales, GP-led continuation vehicles, and discount-to-NAV pricing—a daily look at who's selling, who's buying, and what the spread means, wherever you listen to podcasts.
Links to every story are in the show notes if you'd like to dig into any of them. Thanks for listening, and have a good Friday. That's Startup Fundraising for today. This is a Lantern Podcast.