A week of orbital compute and fusion billions, and today the money goes shopping for plumbing. Good. If you're just joining us, AI venture money's been piling into startups that can claim a foundational spot: model platforms, infrastructure, applied tooling, physical AI, sovereign AI. Recent beats gave us Moonshot AI's three-and-a-half billion, Generalist AI reportedly in talks near three billion, and Onyx Security's hundred-and-thirteen-million raise at roughly a six-forty valuation, to lock down autonomous agents inside the enterprise. And today? We've got real receipts. A data-pipeline raise with actual metrics, a couple of seeds that know what a seed is — and one five-billion-dollar rumor that hasn't closed a thing. Let's dig in. We'll keep tracking this story — AI venture market concentration. Follow the show so the next update finds you. This one's from The Next Web:
The glamorous part of AI is the models. The expensive, unglamorous part is getting them the right data. A startup just raised $40 million betting that job is the next big enterprise layer.
Finally. 400% revenue growth, 180% net revenue retention, zero churn. Actual receipts. This one passes my first filter. 180% NRR at Series B is the number that stopped me. You'd expect a figure like that deep into a late-stage SaaS story. This company has raised just fifty-nine million total since 2023. Insight Partners has seen enough enterprise software to know exactly what that means. Right, and Insight leading tells you they see a real multiple here. Harder question: can a data-pipeline shop with fifty-nine million raised hold the line when AWS decides routing telemetry is a feature it wants to own? That's the moat question. But a neutral control plane governing what an agent's even allowed to see is harder for a hyperscaler to bolt on without owning the whole stack. It's also the second multi-stage firm this week planting a flag on the data-access layer for agents. Forbes, with Rashi Shrivastava, Iain Martin and Richard Nieva:
Flapping Airplanes, a research-focused startup that’s trying to find ways to use less data to train AI models, is in talks to raise capital at a $5 billion valuation, multiple sources familiar with the deal told Forbes. VC firms Index Ventures and Lux Capital are leading the round and Kleiner Perkins is participating, the sources said.
Careful with this one: Flapping Airplanes is in talks at a $5 billion valuation. It hasn't closed. Forbes cites multiple sources, with Index and Lux leading and Kleiner participating. Until the wire hits, treat it as market color. Five billion for a research org trying to train models on less data. That's the whole pitch: use less data. We just aired DataBahn, whose entire thesis is that enterprises need more data infrastructure. Both can't be the future. And here's the concentration problem again: hundreds of millions chasing a five-billion mark on something still at the research stage. Late-stage AI capital is still shopping for new entries. So what has to be true two years from now for five billion to pencil? A data-efficiency breakthrough that actually works and ships. Right now, there's only a hypothesis holding up that number; the product hasn't arrived. Index closed a two-billion-dollar fund and, almost immediately, is co-leading this. Fast conviction or a fast clock — I'd want to know which. From VentureBeat:
Intropy, the AI-native platform automating inventory, pricing and other critical decisions for spare parts businesses, today announced that it has raised $11 million in seed financing. The seed round was led by Felix Capital, with participation from Quiet Capital and early investors General Catalyst and firstminute capital
Spare parts. Finally. Eleven million to fix what keeps trucks on the road — four billion dollars' worth of automotive parts move every day, with the business still running on spreadsheets and software older than the founders. Felix Capital leads, with Quiet Capital in the round. Earlier investors General Catalyst and firstminute are back too. The insiders re-upping this early tells you there's real conviction. Compared with the DataBahn number we just hit, this is the other end of the barbell. There's no 180% NRR to wave around, but you can count the SKUs and point to a customer who bleeds when a part's mispriced. Kirschner and Teh started this in 2024. Two years later, they've raised eleven million to serve a real vertical. That's a seed that looks like a seed. Refreshing, given the week we've had. TechCrunch, with Dominic-Madori Davis:
Ellis AI announced Thursday its emergence from stealth with $10 million in seed funding from investors including First Round Capital, 645 Ventures, Harlem Capital, Khosla Ventures, Thrive Capital, Slow Capital, Kearny Jackson, and Ariel Alternatives CEO Mellody Hobson.
Ellis AI is out of stealth with a $10 million seed. And look who's in: First Round, Khosla, Thrive, Harlem Capital, plus Mellody Hobson with a personal check. That syndicate looks like a bet on Ryan Williams as much as on the product. He co-built Cadre back in 2014. I actually like how specific the pain is: private credit managers drowning in Excel, downloading files, reconciling balances by hand. It's boring, it's real, and you know exactly who the customer is. After DataBahn, this is the second round today where I can picture exactly what the money buys. Nice change of pace. Cadre's arc matters here, though. It raised over $160 million, peaked at $800 million, then sold to Yieldstreet in 2024 for an undisclosed number. Investors are paying a repeat-founder premium here, even though his last company didn't fully land. Right. He saw the constraint from inside Cadre. Now he has to prove closing a fund's books can support a venture-scale company before existing back-office software absorbs it as a feature. This one's from GeekWire:
Less than a year after emerging from stealth operations with $12.7 million and partnerships with pharmaceutical giants, Seattle-based biotech startup Accipiter Biosciences has added $10.5 million to its funding total. The new cash will let the company move faster on promising drug candidates.
Accipiter tops up its seed: ten and a half million, no new names on the cap table, existing investors leaning in. And the CEO's line says it all: 'we could've stuck with the original plan and been just fine.' So runway isn't driving this. Right — it's the quiet opposite of a crossover round. Morgan Stanley isn't sketching an exit slide, and no fresh crossover fund is parachuting in. The people already on the paper are writing another check because the science outran the timeline. And I buy the acceleration here. The Pfizer deal alone is worth up to three-thirty in milestones, plus the Kite agreement with Gilead. The case is right there on the partnership sheet. They can afford to go from one or two clinical programs to three or four. The team came out of David Baker's Institute for Protein Design at UW — the Nobel lab. That matters. They've raised twenty-three million total in under a year, with a moat built on people who know how to fold the protein. It's a twenty-two-person team making proteins that bind two targets at once, with real pharma money behind it. After a week of nine-figure decks and soft math, this round actually adds up. If you follow startup capital, try Infrastructure Secondaries Daily. It covers LP stake sales, GP-led continuation vehicles, and discount-to-NAV pricing: who's selling, who's buying, and what the spread means. Find it wherever you listen to podcasts.
You'll find links to every story in today's show notes if you want to take a closer look at anything that caught your attention. That's Startup Fundraising for today. This is a Lantern Podcast.