A $20 billion valuation to cap a week of enormous checks—so who's actually growing fast enough to earn it? New here? Here's where things stand. AI startup funding keeps clustering around foundation-model talent, infrastructure, applied tools, agent security, and sovereign AI. Before today, the clearest marker was Lumilens: it launched with $900 million in funding, including a $700 million Series C at a $5.51 billion valuation, to build optical-networking chips for AI clusters. This is Startup Fundraising. Live shopping, legal AI, satellites, grid procurement—and a week of valuations moving faster than the proof. Here's Ian Thomas at CNBC:
Building on the growing popularity of livestreamed shopping, startup market Whatnot said on Friday that it has closed its latest round of funding at a $20 billion valuation, nearly doubling its valuation from less than a year ago. With the funding from this latest round, a $545 million Series G led by Iconiq, Lightspeed and Avra, the company plans to invest in ways for sellers to grow on the platform, according to Grant LaFontaine, co-founder and CEO of Whatnot.
Whatnot says it has 60% of a $22 billion live-shopping market. That's roughly $13 billion in activity against a $20 billion valuation. For this Series G to work, the whole market has to get a lot bigger, fast. And it may be. CNBC says more than 650,000 people are joining Whatnot every week, buyers have more than doubled, and it's already passed last year's $8 billion GMV. Those are real operating numbers. Receipts, yes—but GMV is merchandise changing hands, not Whatnot's revenue. LaFontaine says the $545 million is going to seller tools, AI, and new markets. Fine. Now show me those tools can turn a collectibles livestream into durable take-rate dollars. Iconiq, Lightspeed, and Avra co-led. The climb is steep: nearly $5 billion in January 2025, $11.5 billion last October, now $20 billion. That growth makes the price plausible. The market-share math still has to work. From Daniel Levi at Tech Startups:
Legal AI startup Harvey is in advanced talks to raise at least $500 million at a $15.5 billion valuation, just five months after investors valued the four-year-old company at $11 billion, according to an exclusive report from The Information. The proposed deal would mark a 40% valuation jump in less than half a year. There is something behind the eye-popping number: Harvey’s revenue has been climbing at nearly the same breakneck pace.
Harvey's in advanced talks, not closed, for at least $500 million at $15.5 billion. Lightspeed has expressed interest in leading, but there's no signed term sheet yet, and final terms can still move. But the operating numbers are finally substantial: roughly $190 million ARR at year-end 2025, about $300 million now. If that pace holds, there's a case for $15.5 billion. If it slows, investors are paying up for a very expensive legal brief. Here's another sign of how concentrated AI venture money has become: Harvey is chasing $500 million at a $15.5 billion valuation. A 40% reprice in five months is fast, though Tech Startups says ARR has climbed even faster. Four years old, already more than $1.2 billion raised, and this would push total funding past $1.7 billion. It needs to be core infrastructure at giant firms. A polished tool associates open during panic season won't justify that kind of funding. Lucie Pazderkova, writing in EU Perspectives:
On 5 August, that company became the first recipient of a brand-new EU-backed investment vehicle called the Scaleup Europe Fund (SEF). Together with the US investor General Atlantic, SEF co-led a funding round that raised €1bn for Iceye, pushing the company’s valuation past €10bn. SEF itself put in €300m.
SEF's first check is easy to read: €300 million of Iceye's €1 billion round, alongside General Atlantic, at a valuation above €10 billion. Finally, a strategic check where you can see the denominator. Iceye also has an asset you can point at: the world's largest commercial radar constellation. It sees through clouds and darkness. That's sovereignty with hardware, customers, and a very expensive reason not to sell early. The Scaleup Europe Fund is targeting €5 billion. It was built to keep European companies from drifting to Silicon Valley. Starting with a Finnish satellite operator alongside General Atlantic tells you Brussels is chasing scale-up exposure, not ceremonial innovation grants. Good. But at €10 billion, Iceye has to turn orbital imagery into durable contracts across defense, disaster response, and infrastructure. Stunning cloud pictures won't carry that price. This one's from FinSMEs:
Fluxco, an Austin, Texas-based developer of an AI-powered procurement platform for electrical grid infrastructure, raised $26m in seed funding. The round was co-led by 8VC and Congruent Ventures, with participation from Trust Ventures, Overture Ventures, and a syndicate of industrial technology investors. The company intends to use the capital to accelerate the development of its AI-native engineering specification parsing models, expand its global supplier manufacturer network, and scale its end-to-end supply chain execution operations.
Finally, $26 million for a problem with actual invoices attached. Fluxco is parsing transformer specs, matching bids across 150-plus OEMs, and tracking manufacturing timelines—because a utility waiting on equipment does not care how cinematic your AI demo is. The setup here is refreshingly clear: 8VC and Congruent co-led, Trust Ventures incubated Fluxco and stayed in, and the company spelled out the spend—models, supplier network, execution. Two leads on a seed round. We've survived worse. But $26 million is a big seed, so Brian Tochman has to make “within days” measurable: better bids, fewer specification mistakes, and transformers delivered faster than the old procurement circus. If it's just a clever parser, that money evaporates fast. Fluxco's claim is end-to-end accountability. Parsing an engineering document is software; getting a global manufacturer network to honor a timeline is the business. This money funds both. This one's from PR Newswire:
Malachyte, a behavior intelligence company, today announced the close of a $10 million seed funding round co-led by Bessemer Venture Partners and Gradient Ventures with participation from Harpoon Ventures. Malachyte was co-founded by Sidd Motwani (CEO), Ian Anderson (CTO), and Shivaditya Sinha (COO), the team behind the behavioral intelligence infrastructure that powers 90%+ of Spotify's recommendations across 800M+ users and 1B+ items.
A $10 million seed to fix e-commerce conversion? Good. Brands are apparently losing $29 on every new customer once ads and returns are counted. If Malachyte can stop retailers paying Meta to send people to a dead-end storefront, that's a very real product. The round is clean: Bessemer and Gradient are co-leading, with Harpoon participating. And the founders' Spotify résumé is unusually concrete—they say their technology powered more than 90% of recommendations across 800 million-plus users and a billion items. Spotify is one giant closed ecosystem. E-commerce is a thousand messy retailers: returns, discount codes, bad inventory data—the whole circus. “Reads behavior before the first click” sounds slick; show me it lifts conversion without becoming a fresh privacy headache. Bessemer's making another early bet this week, this time on consumer behavior rather than clinical AI. After Whatnot's huge live-shopping share claim, Malachyte is tackling the less-glamorous half of commerce: getting visitors to buy. Have feedback, a story idea, or a correction? Email us at startupfundraising at lantern podcasts dot com. Your notes help make Startup Fundraising better.
Every story is linked in the show notes if you want a closer look. Thanks for spending part of your Sunday with us. That's Startup Fundraising for today. This is a Lantern Podcast.