Cognition just closed a billion dollars at a twenty-six billion dollar valuation — and somewhere, a revenue spreadsheet is crying in a corner. Welcome to Startup Fundraising. We’ve got Cognition’s monster round, a Brazilian fintech that found where the actual money lives, and a binary analysis startup making some very bold Anthropic comparisons. Twenty-six billion. I’ve been asking, “show me the revenue” all week, and the market’s answer is apparently, stop asking. That’s gone from skepticism to something closer to structural exasperation, so let’s just say it plainly today. Into it. Ashutosh Singh, writing in The Tech Portal:
AI startup Cognition has raised more than $1 billion in a new funding round at a $26 billion valuation. The financing was co-led by Lux Capital, General Catalyst, and 8VC, with participation from Founders Fund, Ribbit Capital, Atreides Management, and several existing backers. The valuation has more than doubled from the company’s previous $10.2 billion mark reached less than a year ago, making Cognition one of the fastest-growing startups in history by valuation growth.
Twenty-six billion. Founded in 2023. More than two and a half billion raised now. Devin got out, the press was huge, and then the people running engineering teams in production got very quiet about what it actually does day to day. I want the ARR behind that twenty-six billion — not the capability story, the invoice stack. Lux, General Catalyst, and 8VC are co-leading, with Founders Fund and Ribbit in the mix. That is a serious lead syndicate, which tells you this is not a bridge dressed up as growth. But the valuation has more than doubled from ten-point-two billion in under a year, per The Tech Portal’s Ashutosh Singh. The leads are making that call on purpose. Here’s what changed for me this week: I started asking “show me the revenue” on every round, one by one. Cognition at twenty-six billion isn’t dodging the question — the market is rewarding them for not answering it. That’s a different problem. Scott Wu can say “multi-model orchestration is the future,” and sure, that’s a fine technical thesis. It’s still not a number. And for context, Socket closed at roughly a sixteen-times revenue multiple earlier this week, and we called that rich. Cognition makes Socket look cheap. With Socket, at least you could do the ARR math. Here, the denominator is basically missing. Lauren Forristal, writing in TechCrunch:
WeRoad, the Milan-based group travel startup, has raised a $58 million Series C round led by Airbnb as it prepares for its first major expansion outside Europe. The funding brings the company’s total capital raised to roughly $100 million and will finance WeRoad’s push into the U.S., beginning with Austin.
WeRoad out of Milan — fifty-eight million in a Series C, Airbnb leading, total raised now around a hundred million, and the U.S. entry point is Austin. TechCrunch had that one. Airbnb leading a round in a group travel company is a real signal. They’re not a passive check here — they want the social layer they never built themselves. What I want to know is whether WeRoad’s unit economics in Europe can actually support a cold start in Austin, or whether fifty-eight million is just the price of finding out. The “IRL economy” framing puts them in the same bucket as Timeleft and Pie, basically loneliness as TAM. That’s a real category forming. Austin as the beachhead is the part that makes it interesting — it’s a very specific bet on who the American solo traveler actually is. Every company this week has been hiding behind a narrative. Cognition at twenty-six billion is the extreme version, but even WeRoad is leaning on “loneliness epidemic” copy before we know what repeat booking rates look like in Italy or Spain. Show me cohort retention in the European markets before I buy the Austin story. From SiliconANGLE:
British software supply chain security startup RevEng.AI today said has raised $15 million in early-stage funding after developing technology similar to Anthropic PBC’s Mythos model and working out a way to put it to good use.
RevEng.AI — fifteen million, binary reverse-engineering, and they put the Mythos comparison right in the lede. That’s either the most confident technical claim of the week or the most dangerous sentence in the deck. BiNet either does what Mythos does or it doesn’t, and “similar to” is doing a lot of load-bearing work there. What matters here is where it came from — BiNet was trained alongside allied government cybersecurity units and top commercial security firms. That’s not a consumer demo environment; that’s a hardened training corpus for finding backdoors in critical infrastructure binaries. SiliconANGLE flagged the Mythos parallel, and if it holds up technically, software supply chain security is a moat that doesn’t require you to out-scale Anthropic. The use case is real — no source code access, hunting hidden backdoors in firmware and third-party executables. That’s the actual threat surface after SolarWinds, after XZ Utils. Fifteen million is a serious number for a British early-stage shop if those government training relationships turn into paying contracts. I want to know if those allied units are customers or just data partners. From Sinziana Albu at The Paypers:
Brazil-based fintech Robbin has secured USD 8 million in seed funding and structured a USD 100 million credit fund to expand supply chain finance. The seed round was co-led by Canary, Atlântico, and Caravela, with participation from AB Seed, Norte Ventures, and international investors Clocktower and Tomorrow Capital. In parallel, the company has structured a Fundo de Investimento em Direitos Creditórios (FIDC) in partnership with Augme, the asset management arm of XP.
Eight million in seed equity and a hundred million FIDC credit fund running in parallel — that’s a twelve-to-one ratio of credit capital to venture capital, and that ratio is the whole thesis. The seed is basically the equity wrapper; the FIDC structured through XP’s Augme arm is where Robbin actually makes money. And they built it on Pix, not Visa or Mastercard, so there’s no card network rake eating the spread. If they deploy that five hundred million reais by the end of 2027 like they’re projecting, you can actually model the interest income. That’s a business. That’s an invoice stack I can read. Here's ETEnterpriseAI:
Nilekani, who is backing the new platform as its anchor investor, said technology-led innovation continues to create new opportunities across sectors. Nandan Nilekani's VC firm Fundamentum Partnership on Tuesday announced the launch of a dedicated frontier-tech investment platform, F2A(Fundamentum Frontier Advisors), with an initial corpus of ₹3,000 crore to back artificial intelligence and deeptech companies in India. The platform's maiden Sebi-approved fund - Fundamentum III-AI and DeepTech - has a target corpus of ₹2,000 crore.
Out of India today — Nandan Nilekani’s Fundamentum Partnership is spinning up F2A, Fundamentum Frontier Advisors, with a ₹3,000 crore corpus. The structure splits into a SEBI-approved fund targeting ₹2,000 crore and up to ₹1,000 crore in parallel co-investments, so this is not just a fund announcement. It’s a two-track deployment vehicle. Stage-agnostic, consumer AI, enterprise AI, physical AI — that’s a wide net. Nilekani is the anchor investor himself, which tells me he’s putting his own rupees behind this, not just his name. I’d want to know what the first check looks like before I take the “frontier” branding too seriously, but the co-investment sleeve alongside a registered fund is at least a real structure, not a press release. If Startup Fundraising helps you stay sharp, take a moment to subscribe wherever you’re listening. And if you can leave a quick review, it really helps other founders and operators find the show.
You’ll find links to every story we covered today in the show notes, so if one caught your ear, you can go deeper there. Thanks for listening, and we’ll be back with more soon. That’s Startup Fundraising for today. This is a Lantern Podcast.