DeepSeek wants seven billion dollars, and now every hard-tech founder is looking around like, oh, so the window's open. Welcome to Startup Fundraising. Today we're bouncing from orbital data centers to humanoid robots to AI security — and, yes, we're starting with the DeepSeek number. Seven-point-three billion pre-revenue clarity is a lot to unpack here. I want the cap table before anybody starts calling this a market signal. We've also got a rocket startup strapping a data center to an upper stage, which sounds either brilliant or like a very expensive way to blow up two businesses at once. That's ahead. Here's Dataconomy:
DeepSeek is in advanced talks to raise up to 50 billion yuan ($7.35 billion) in what would be the largest single funding round for a Chinese artificial intelligence company, according to reports from The Information and Reuters. The funding could value the previously self-funded startup at as much as $50 billion, a significant increase from an estimated $10 billion just weeks ago.
DeepSeek is in advanced talks for a $7.3 billion round that would put the company at fifty billion, up from about ten billion just weeks ago. And the lead isn't a VC firm — it's China's state-backed National AI Industry Investment Fund, with Big Fund III also in the mix. A five-times jump in weeks? That's geopolitical procurement, not price discovery. Beijing decided DeepSeek is strategic, and the cap table followed the flag. Worth flagging: Tencent reportedly wants a twenty percent stake, but the founder controls nearly ninety percent of the company and is personally putting in almost three billion yuan. So he's giving up as little as possible while still taking the state's money. So what has to be true for this to make sense at fifty billion? They built efficient models on restricted chips, sure. But once you're taking state capital and Big Fund III — whose whole mandate is semiconductor independence — you're not really a startup anymore. You're a national program with branding. From Jason Rainbow at SpaceNews:
TAMPA, Fla. — Cowboy Space, founded less than two years ago as Aetherflux to develop space-based solar power, has raised $275 million at a $2 billion valuation to build rockets with upper stages that would serve as data centers once in low Earth orbit (LEO).
Cowboy Space — formerly Aetherflux, formerly a space-based solar power play — just closed a $275 million Series B at a $2 billion valuation. Index Ventures led again, same as the Series A, so at least there's continuity on the cap table. Two years old, one name change, and now it's rockets, orbital data centers, and space-based power? Pick a lane. That's not a pivot, that's a mood board with a valuation attached. To be fair, Index coming back to lead the B after leading the A is real conviction, not a tourist check. But $2 billion on roughly $365 million raised total, and no rocket flying yet, is a pretty steep ask on the trust-me curve. What has to be true in two years for this to work? They need a rocket, a working orbital data center, and a paying customer before Starcloud, SpaceX, and Blue Origin eat the category. Baiju Bhatt built Robinhood, fine, but Robinhood is software. Orbital infrastructure is not software. From Greg Bock at AI Insider:
Chinese embodied AI startup Vbot has raised about USD $73 million, in a Pre-A funding round to expand robot production and develop full-size humanoid robots, according to PanDaily. The round was co-led by Oriental Fortune Capital, Huatai Zijin Investment and Fosun RZ Capital, with participation from investors including SAIC Motor’s Shangqi Capital.
Chinese embodied AI startup Vbot just raised seventy-three million dollars in a Pre-A — yes, Pre-A, not Series A — co-led by Oriental Fortune Capital, Huatai Zijin, and Fosun RZ Capital, with SAIC Motor's venture arm also in the mix. Hat tip to Greg Bock at AI Insider, who pulled this from PanDaily. Seventy-three million at Pre-A for a company founded in late 2024 is a big number for something that's still mostly in blueprint mode. The stated use of funds is mass production, retail expansion, and humanoid development — so that's three different bets inside one round. The concrete anchor here is the robotic dog platform. They're scaling that from five hundred to twenty-five hundred units a month. That's the real business right now. The humanoid and world model stuff is the pitch deck. And I need to know what 'offline retail expansion' means for a robotics company. Are they opening stores? For robot dogs? What does this look like in two years if the humanoid pivot stalls — a glorified pet toy manufacturer with a very expensive cap table? Ynet News writes:
Frame Security announced its public launch today alongside a $50 million funding round led by Index Ventures, Team8 and Picture Capital, with participation from Wiz CEO Assaf Rappaport and technology investor Elad Gil, who first backed the company as an angel investor and has since doubled down through his fund, Gil Capital.
Frame Security out of Israel just went public with a fifty million dollar round — Index Ventures and Team8 co-leading, Elad Gil going from angel to fund check, and Wiz CEO Assaf Rappaport on the cap table as a strategic name. That's a credible syndicate for a stealth-to-launch. The pitch is basically: security awareness training doesn't work — ninety percent of breaches still involve a human — so here's an AI platform to fix the human. I've heard that framing a hundred times. What I want to know is whether this is a product or just a deck with a nice TAM slide about a thirteen-billion-dollar market. To be fair, having Rappaport involved isn't just optics. Wiz scaled by selling to security buyers who actually sign the checks, so if Frame is in his orbit, they probably understand enterprise procurement. Sure, but fifty million to launch means this thing needs to land real enterprise contracts fast, not just pilots and 'we're redefining human risk.' Tell me what the product does on day one when a deepfake CFO calls an employee and asks for a wire transfer. This one's from IT Digest:
Basata, the AI company rebuilding the operational layer of US healthcare, announced a $21 million Series A led by Basis Set Ventures, with participation from Cowboy Ventures, PHX Ventures, Zenda Capital, and Victoria Treyger. Basata’s AI agents handle the administrative work still running on fax machines and phone calls, like referrals, intake, patient scheduling, and follow-up end to end.
Basata just closed a $21 million Series A led by Basis Set Ventures, with Cowboy, PHX, Zenda, and Victoria Treyger in the round. Total raised is $24.5 million, so the seed was very thin before this. Okay, this is actually one of those unsexy-problem, real-product stories I want more of. Fax machines. Referral queues. Appointment scheduling taking weeks. That's not a narrative — that's a broken workflow with a body count. The numbers they're giving are specific enough to take seriously: 500K patients served, 100K in the last month alone, and 70% of new sales from customer referrals. That last one is the tell. A referral rate that high in a sector this sticky means the product actually works inside the chaos of a real specialty practice. That's not a demo environment. I want the unit economics at the provider level — are they charging per referral, per seat, what? If Startup Fundraising is part of your routine, take a moment to subscribe or leave a review wherever you're listening. It really helps other founders, operators, and investors discover the show.
If you want to dig deeper, we've put links to every story from today's episode in the show notes. Take a look there for the pieces that caught your attention.
That's Startup Fundraising for today. This is a Lantern Podcast.