Newsletter · 2026-08-10 · Ashutosh Agarwal

SpaceX Prices the Largest IPO Ever as AI Money Circles Back on Itself - The Raise - Week of August 10, 2026

For the week of August 3 to 10, 2026, SpaceX went public in what hosts called the largest IPO in history, a fresh crop of AI infrastructure startups raised on the strength of contracts with two customers, and a home-furnishings founder walked through the term sheet that quietly capped her company.

The Raise

Week of August 10, 2026: SpaceX Prices the Largest IPO Ever as AI Money Circles Back on Itself

SpaceX pulls off the largest IPO in history, AI's data-center money keeps circling back on itself, and a home-goods founder relives the term sheet that nearly sank her company. Week of August 3 to 10, 2026.

This was a week of extremes. At the top end, SpaceX went public in what podcast hosts kept calling the largest IPO ever, a $75 billion raise at a valuation north of $1.7 trillion. Just below that, a fresh crop of AI "infrastructure" startups raised hundreds of millions of dollars each, largely on the strength of contracts with two customers (OpenAI and Anthropic), a pattern that has veteran investors muttering the words "circular financing." And at the human scale, a home-furnishings founder walked through the single worst deal she ever signed, and why she'd tell any founder today: there is good money and there is bad money.

Here's what we heard.

This Week's Rounds

• SpaceX, ~$75B raised at a ~$1.75 trillion valuation (IPO). Rocket and satellite-internet company. This was the headline event of the week and, by the hosts' account, the largest IPO in history. Rather than the usual process where banks poll big investors to find a price, "Elon Musk just came down on high and said, we're going out at $135" a share, a take-it-or-leave-it number (which, not coincidentally, is roughly the price that tipped Musk into trillionaire status on paper). Demand was about four times the shares on offer, roughly $250 billion of orders against $75 billion of stock, with Fidelity alone reportedly putting in a ~$5 billion order. NYU valuation professor Aswath Damodaran ran the numbers and got to ~$1.2 to 1.3 trillion, well short of where it priced. It was an all-primary raise (all the cash goes to SpaceX, not to insiders cashing out), the banks actually sold ~$85 billion of stock (a standard over-allotment cushion), only ~3 to 5% of the company floated (versus a typical 10 to 20%), and for the first 30 days the underwriters can't lend shares to short sellers. Translation: almost no sellers, lots of forced buyers coming as index funds add the stock, a one-way street, at least at first. (The Wall Street Skinny, "SpaceX: Revisiting the IPO and What Comes Next" (Aug 5))

• Volta Infra, $300M at a $2.4B valuation, led by a16z and Altimeter. AI cloud/data-center startup, NVIDIA-backed, founded only in January 2026 by former Brookfield Asset Management executives; it leases AI computing capacity and helps customers finance their expensive chips. The round landed days after Volta signed a $10 billion contract to supply computing power to Anthropic from a data center in Norway (133 megawatts, stocked with NVIDIA's newest "Vera Rubin" chips). One host laid out the logic bluntly: Volta raised the money "because we landed a $10 billion contract with an unnamed leading AI developer," exactly the kind of interlocking, contract-backed financing that Bloomberg and others have flagged as "circular in nature," creating "an increasingly interconnected web of dependencies… that risks magnifying losses if AI demand fails to live up to the lofty expectations." (Tech Brew Ride Home, "OpenAI Has Receipts" (Aug 4); AI Update, "Anthropic's $10 Billion Cloud Deal, Elon Musk on AI at Tesla" (Aug 4))

• Mariana Minerals, $310M Series B, led by Khosla Ventures (~$400M total raised). Vertically integrated copper and lithium mining-and-refining startup, meaning it both digs up the metals and processes them, aiming to build a domestic supply of the materials that go into batteries and electronics. A striking investor list joined Khosla: Andreessen Horowitz, Breakthrough Energy Ventures (Bill Gates's climate fund), BHP Ventures (the mining giant's venture arm), Mitsubishi Corporation, and In-Q-Tel (the CIA-linked strategic investor). The mix of top-tier tech VCs, a strategic miner, and a national-security investor is a tell for how "critical minerals" is now being treated as both a climate and a security bet. (Mining Stock Daily, "Morning Briefing: Gold and Silver Surge as Revival Gold Delivers a Major Joss Intercept" (Aug 4))

• Valor Atomics, $300M, led by Sequoia. Nuclear-energy technology startup, founded by Isaiah (described by his new backers as "a one-of-one force of nature"). This one is notable less for the amount than for who wrote the check and why: Sequoia partners Alfred Lin and Pat Grady admitted a $300 million check into a company with "multiple layers of technical risk remaining" and "a business model that is many years in the future" broke their own rules. "When the recommendation came out for a $300 million investment, some eyes popped out of some skulls," Grady said. "I said, boy, that seems like a lot of money for a company with this much risk in it." They flew out, spent a full day with the team, and "rode with the conviction" of the two partners who sourced it, Sean (a physics PhD) and Liam Corgan (a Harvard physics grad, former nuclear-industry operator, and Olympic gold medalist who'd joined Sequoia only six months earlier). (Bloomberg Tech, "Alphabet's Bond Sale Signals AI Confidence" (Aug 6))

• Bland, $100M+ raised (through Series C). Voice-AI startup building phone agents for regulated industries (think automated calls that sound human). The backstory is the interesting part: the co-founders started the company at 19 and 24, went through Y Combinator and "flunked out… no revenue, rejected by 180 investors," and then raised more than $100 million over the following three years. Growth came from stunts, not a big-name CMO, most memorably a billboard campaign in New York and San Francisco reading "Still hiring humans?" with a phone number that connected callers directly to the AI (the ad itself was the demo). That campaign did "over a billion impressions" and booked out the sales team "three to four months in advance." (The Dave Gerhardt Show, "The Viral Marketing Bets Behind High-Growth Voice AI Company Bland.ai, with Ethan Clouser" (Aug 3))

• Boardy, $8M seed. AI "super-connector" that makes warm introductions over phone calls, WhatsApp, and email; it now has more than 160,000 users. The fun detail: the round was raised "through autonomous pitching," the product itself did some of the fundraising legwork. Founder Andrew D'Souza is a repeat entrepreneur (he previously co-founded fintech unicorn Clearco, which deployed over $3 billion to more than 10,000 e-commerce businesses). On the brutal early days of getting a network off the ground: "I begged everybody I'd ever met to try and talk to Bordy… the first thousand people, first ten thousand people, was really, really tough." He's also spun up a small investment vehicle, Boardy Ventures. (Spark of Ages, "The AI With 160,000 Friends That Networks For You / Andrew D'Souza, Boardy.ai" (Aug 7))

• A "predict-the-future" simulation startup, $80M Series A at a valuation "well over a billion dollars." Founded by two people who started building as teenagers, the company simulates entire populations to forecast behavior, and claims to be the most accurate in the world at predicting elections, consumer purchase intent, and social-media shareability. Asked what the money is for, the founder's answer was blunt: "People… getting the best people, the smartest people, the most driven people." Not everyone's a fan, the founder cited a New York Times op-ed headlined "this is what will ruin public opinion polling for good." (The company's name wasn't stated in the episode.) (TechStuff, "They Founded a Company as Teens. Now They're Predicting the Future" (Aug 5))

Founder Story of the Week

Lily Kanter & Serena Dugan, Serena & Lily. (How I Built This with Guy Raz, "Serena & Lily: Serena Dugan and Lily Kanter. They Built a $20M Brand, Then One Investor Almost Destroyed It" (Aug 3))

If you want a single episode that explains why "The Raise" exists, this is it. Kanter and Dugan built Serena & Lily, the home and baby-bedding brand, from a cash-hungry wholesale business (600 to 800 boutiques, no online store, "not even a shopping cart on our website") into a company doing $20 million in revenue by 2010. The product was working. The problem, twice over, was the money.

The term sheet they walked away from. Early on, needing working capital, they set out to raise a $1.5 million friends-and-family round. A friend sent them to a private-equity firm just to get an opinion on valuation. Ten minutes in, the CFO fetched a general partner who, as Kanter tells it, "literally patted us on the heads like we were dogs or puppies," said "you girls sure have been busy," and offered to write the entire $1.5 million as a single check, so they wouldn't have to "pass the hat around." It sounded amazing. Then the term sheet arrived. Kanter sent it to a lawyer she knew, who called back and said: "I forbid you to do this deal." The catch buried inside: the firm was "asking for controlling interest in your company for 17% of your equity." When the partner also balked at the founders paying themselves $150,000 salaries, after they'd gone three and a half years without a paycheck, Kanter's husband overheard on speakerphone and said, "honey, we're going to take a second on the house. You're not doing that deal." They walked, and instead raised the full $1.5 million from friends and family in a 17-day sprint, "every one of our family members, and a slew of friends, and Lily's rabbi."

The deal that actually did the damage. The near-fatal mistake came later. A prickly investor turned litigious, suing for "irreparable harm" to his investment, and the board decided he had to be bought out. To fund the buyout, their major Sand Hill Road investor put fresh money in "at a very heavy-duty preference of a 2X preferred participating, very gnarly security." In plain English: that investor would get double their money back first, ahead of the founders, employees, and everyone else, before anyone saw a dime. It solved the immediate fight, and quietly capped the company. As Kanter told her board chair at the time: "we will never be able to raise another dollar for this company," because any new investor "would want the same terms." When two acquisition offers later came in, they were "nearly impossible to take" because so much of the value sat in an earn-out that the preference would have swallowed, and one acquirer even wanted the founders' "name, image, and likeness in perpetuity."

The lesson, in Dugan's words: "There's good money and there's bad money… there is definitely a correlation between how badly you need the money and, when you are raising money, what the terms are. So egregious terms are more likely if you need the money and they can read that." Kanter's own hindsight: "I probably would have pounded my fist on the table a little bit harder and said, no, we're not taking that deal."

Also Heard

• The week's other trillion-dollar story was a fund blowing up, not a startup. Leopold Aschenbrenner, the 25-year-old "wonder kid" whose "Situational Awareness" memo turned into a roughly $225 million investment vehicle that at one point held ~$45 billion of assets, got wiped out. The 20VC panel's verdict: "absolutely right on trend, absolutely wrong on portfolio construction." He ran the book with 4x leverage on volatile AI stocks; when it turned, Ken Griffin's Citadel bought his public positions for a reported ~$16 billion and reportedly cleared ~$3 billion in a matter of days. Sequoia separately confirmed it had been approached to take his private Anthropic stake but "Ken Griffin showed up with a better solution." The consolation for LPs who got in early: up ~440% and then back down still leaves you roughly where you started. The ones who bought in around April to June "have been wiped." (The Twenty Minute VC, "20VC: Airtable Sold for $1.285BN | Moonshot AI Raises $3.5B at $35B…" (Aug 6); Bloomberg Tech, "Alphabet's Bond Sale Signals AI Confidence" (Aug 6))

• Airtable sold to Bending Spoons, and VCs called it "capitulation." The no-code software company, valued at $11.7 billion in 2021 (after a $735 million raise), was acquired by Italy's Bending Spoons for an enterprise value of ~$1.285 billion (about $2.25 billion in implied equity, mostly cash on Airtable's own balance sheet). The business is healthy on the surface, roughly $480 million in annual recurring revenue, still growing 20%+, used by 80% of the Fortune 100, but ChatGPT- and Claude-style tools have eaten into its edge. On 20VC, the read was that late-stage investors got their money back (1x), the founders made about $150 million, "a lot less than they thought, but certainly enough," and the deal may mark the moment more 2021-vintage companies quietly accept that "it's time to capitulate." (Tech Brew Ride Home, "OpenAI Has Receipts" (Aug 4); The Twenty Minute VC, "20VC: Airtable Sold for $1.285BN…" (Aug 6))

• How Sequoia actually sized its Anthropic check. In an unusually candid interview, Alfred Lin and Pat Grady walked through the mechanics of their latest Anthropic investment. The original recommendation was $1 billion. Then, in the room, "the number started at five billion," described as "a theoretical number mentioned to be provocative." They landed at $2.5 billion, which Grady said was simply "the most we can do out of the core funds' committed capital," pointedly adding that Sequoia doesn't do SPVs (special vehicles that let a firm "speak for" more money than it actually has committed). A rare peek at how a top firm decides how much of the franchise to risk on one bet. (Bloomberg Tech, "Alphabet's Bond Sale Signals AI Confidence" (Aug 6))

• Moonshot AI's raise got a hard number: $3.5B at a $35B valuation. After weeks of reporting on the Chinese AI lab's fundraising in the $25 to 30 billion range, 20VC put a firmer figure on it. The host's one-line summary of the winners and losers as compute demand keeps compounding: "Moonshot is happy, NVIDIA is happy, enterprise is happy, OpenAI very, very sad." (The Twenty Minute VC, "20VC: Airtable Sold for $1.285BN | Moonshot AI Raises $3.5B at $35B…" (Aug 6))

• "Fake it until it actually works." Akemi Tsunagawa, founder of Bespoke, described starting in 2015 with BeBot, a multilingual chatbot for tourists that she ran with humans pretending to be the bot until she'd proven people actually wanted it, then pivoting twice more (to AI-powered worker training with smart glasses, and finally to robotics for shipyard inspections), each turn learned from marathon customer interviews at train stations. A useful counterweight to this week's billion-dollar headlines: sometimes the "AI company" starts as people behind a curtain. (This Week in Startups, "How Bespoke faked AI until it actually worked (w/ Akemi Tsunagawa)" (Aug 3))

• The soda exits keep coming. Two prebiotic-soda founders told their build stories this week. Stephen Ellsworth walked through reformulating Poppi's recipe one week before filming Shark Tank, moving it from an "apple cider vinegar beverage" that "tasted terrible" to a colorful "prebiotic soda" sold on taste, a repositioning that helped carry the brand from a Dallas farmers' market to a ~$2 billion sale to PepsiCo. And OLIPOP's Ben Goodwin recounted dropping out of college at 20 to work at a kombucha company before eventually building his own better-for-you soda brand around prebiotic fiber. (Faith Driven Entrepreneur, "Episode 386 - Scaling Poppi: Farmers Market to $2B Pepsi Exit | Stephen Ellsworth" (Aug 4); How Leaders Lead with David Novak, "#303: Ben Goodwin, Cofounder, CEO & Formulator, OLIPOP" (Aug 6))