SpaceX Valued at $800BN & Harvey Raises $160M at an $8BN Price & Netflix Acquires Warner Brothers

Summary

• SpaceX’s $800B secondary is priced almost entirely on Elon premium: Jason’s math — $15B revenue growing ~30% means “north of 40 times run rate,” and “I’m not sure I’d be a buyer at that price… I might be looking to my seller.” The anomaly gets stark when set against Anthropic: would you rather pay 20x for 300% growth or 40x for 20-30% growth? “A lot of that is just the great leader premium.”
• 2026 could be the Facebook-2012 / Alibaba year of IPOs: SpaceX + Anthropic + Databricks going out at ~$800B/$400B/$200B is $1.4T of market cap, roughly $700B returned to VC — yet still only ~20% of the ~$2.8-2.9T total private venture FMV. Jason doubles his lost $50k bet: Anthropic and Databricks IPO in the back half of 2026, “unless the market crashes.”
• Down-round IPOs have lost their stigma: Rory’s signature line — “every time that private market valuations came into contact with public market valuations, private market valuations were found wanting” — but Jason argues the classic advice against raising too high now has “an audience of n equals zero.” Anthropic exit-price bets on record: Jason $500B (20x a plausible $26B forward), Harry $420B, Rory $350-400B.
• Netflix buying Warner Bros. Discovery for $82.7B is tech finishing off old media: Netflix at ~$470B swallows the asset for under 20% dilution while sub-$20B Paramount mounts a debt-and-Kushner-funded hostile bid. The regulatory twist: Hollywood’s objection isn’t monopoly but monopsony — Netflix as the single dominant content buyer. Digital already ate advertising and much of retail; Jason’s next candidates: fintech and banking.
• Harvey at $8B (on $160M in, <2% dilution) crystallizes the AI-app pricing debate: no legal software company has ever been worth more than ~$2B, so you must believe TAM expansion into labor plus near-total market dominance. Jason’s counter: “It is possible to lose money on a great company… the only risk you’re running is you’re paying eight for something that might be worth four” — if growth goes 10x→3x→2x you’re a year ahead; if it drops sub-100% “you’re just way wrong.”
• The unhedged model risk: a deep-reasoning step function could “Jasper” the app layer — Jason’s most heated call: today’s agents (Sierra, Decagon, Fin) are “90% right” and slow, and when deep reasoning drops from minutes to milliseconds, “the Harvey that we used in 2025 seems just quaint.” Harry’s pushback — defensibility lives in GTM and enterprise implementation — drew “you don’t even know what you’re talking about here, Harry.”
• Airwallex at $8B is either an “Asia discount” dislocation or a data-risk landmine: same ~$1B ARR as Ramp at $32B, Brex at $13-14B with worse growth. Rory would invest only conditional on relocating every China employee; Jason calls that “beyond condescending — the great CEOs will figure it out,” and reads the Rabois attack as “an objection added to a sales process,” possibly with a racist element.
• Tiger’s downsized $2.2B fund with a ~20% ($400M) GP commit is the credibility trade: “money is a great truth serum — don’t tell me what you think, tell me what you do.” Nine deals a year, concentrated, strategy-of-the-day: staying in the game after 2021’s hubris.

Deep dive

1. SpaceX at $800B: two amazing companies, one unknowable price

• Jason’s setup: it’s “an amazing rocket company and an amazing communications and Starlink company” — but $15B of revenue growing plus-or-minus 30% puts the secondary at north of 40x run rate. His honest position: “I’m not sure I’d be a buyer at that price… if I own some, I might be looking to my seller.” There’s “a huge amount of Elon magic overlay, and so far that magic has worked, but it’s definitely a lot of non-obvious math baked into the price.”
• Harry’s pushback: doesn’t a secondary at $800B just prove these companies don’t need to IPO — echoing the prior Tomasz Tunguz conversation — and might that make 2026 worse, not better, for listings? Jason concedes maybe they never IPO, “but I don’t think Databricks and Anthropic are hiding from an IPO… they’ve been very clear they’re on a path.”
• Rory’s structural worry: a private high-water mark makes IPOs “emotionally unattractive” — if public markets only say you’re worth $400B, “it doesn’t feel like a win… it’s not going to feel like the way you wanted it to be.”

2. Down-round IPOs: nobody cares, and the advice has no audience

• Rory’s line of the episode: “every time that private market valuations came into contact with public market valuations, private market valuations were found wanting” — almost every 2025 IPO priced below its last private round. His caveat: at $800B pre, “you’re not running any other kind of meaningful risk except valuation risk, so you can’t cry like a baby when valuation risk bites you in the ass.”
• Jason’s meta-point: the classic founder advice — don’t raise at too high a price, you’ll regret it — “is as sound as it ever was, but it has an audience of n equals zero. No one cares if they come out of a hot accelerator and raise at 50 to 60 on a bunch of SAFEs with no rights and do a round at 30 later. The angst is gone around mispriced rounds.”
• Rory’s partial dissent: it survives until it doesn’t — “if you continually price something at a high price and you’re continually wrong… at some point some adult in the room will say maybe we should stop doing that and buy them when they’re cheaper.”

3. The 2026 IPO supercycle math — and Jason doubles down

• Rory ran the numbers: SpaceX ($800B) + Anthropic ($400B) + Databricks ($200B) = $1.4T of market cap; assume VCs own a little under half and that’s ~$700B returned to venture — “a pretty darn good year,” yet sobering: only ~20% of total private FMV of $2.8-2.9T. Power-law history rhymes: 2012 was a “bumper year” because Facebook ($70-80B) went public; 2018 was a “bumpy year” because Alibaba went public. “What is the Facebook or Alibaba bar of 26?”
• Jason, already out $50k on his bet that liquidity returned by end-2024 (“I thought we were going to have an IPO a day again… little did I know they would all stop growing except the AI ones”), doubles the bet to $100k: Anthropic and Databricks both IPO in the back half of 2026, market crash excepted.
• Rory agrees on Anthropic’s logic: despite the effective-altruism founding angst, they’ve been “remarkably sober-minded… pleasantly more logical than the other player in this space” — and “the conventional thing to do when you’re valued at 400 billion and you need to raise a lot of money would be to access the public markets at scale.”
• Jason’s timing caveat: since the April 15 tariff implosion this has been “the strongest bounce back from a bear market since 1982” — windows close as fast as they open, and if Anthropic has a 3x revenue year “kind of in the bag,” early 2026 is when the multiple is best.

4. Pricing the Anthropic IPO: 500 vs 420 vs 350

• Jason’s number: $500B — simple math, “20x times 26 [billion] is 520,” a multiple that “isn’t even that high for the top 10% of public companies,” and still a good IRR on a rumored $300-350B private round. Rory: $350-400B, tied to the visible growth math from the $170 million round. Harry splits it at $420B.
• Jason’s zoom-out on the anomaly: the market is simultaneously paying 20-25x for 300% growth (Anthropic, decelerating from 10x to 3x) and 40x for 20-30% growth (SpaceX). “It’s just anomalous… to be fair to SpaceX it’s an n of one, it is profitable, it doesn’t have a direct competitor — but a lot of that is just the great leader premium.”

5. Netflix–WBD: the outmatched hostile bid and the monopsony problem

• Jason’s zoom-out: “Netflix won. They ate the media industry.” A ~$470B company putting ~$100B on the table (sub-20% dilution) against sub-$20B Paramount bridging with debt plus PE money including “the Kushner fund.” The WBD board’s calculus is certainty of close: “the person who has $500 billion is more likely to close than the person who has $20 billion. Hmm, I’ll go with him.” David Ellison noted the board hasn’t even responded to Paramount’s offer.
• Regulatory asymmetry per Jason: Netflix is the FTC’s monopoly concern (streaming share over 30% with HBO), Paramount is the FCC’s broadcast-license concern — plus political overtones, with the Ellisons “pretty tight with the White House” and Trump already commenting, “he said gently.”
• The fun part — why Hollywood hates it: “It’s not monopoly power screwing the consumers. It’s monopsony power. If Netflix becomes the biggest single buyer of content… the media buyer for Netflix becomes the most important person in your life.” Star translation: “grinding out rom-coms on a low budget somewhere in Romania.”
• Harry on Zaslav: management “just doesn’t want to do this Paramount deal” — and Jason on the revenue arbitrage: Netflix trades at a far higher multiple, so “if they can inflate all the assets to 10x revenue, this is an incredibly accretive deal.” Jason’s elegy: “poor HBO… they’ve taken the best high-end TV product in history and just thrashed it through five different corporate owners. Whoever gets this thing, please be nice to HBO.”

6. What digital eats next: fintech, banking, maybe auto

• Jason’s 25-year arc: digitization ate advertising first (“Facebook and Google have 70% of not just all digital advertising but all advertising — thanks for playing, newspapers”), then a good slug of retail (Amazon/Whole Foods), and now entertainment: “little tech weenies in Silicon Valley — we just did it to Hollywood.”
• The next one, both agree, is fintech: “At what point does someone like Revolut say, I woke up and realized I’m the largest market-cap bank company in Europe — maybe I’ll buy one of you branch-baby thingies?” Banking, “and at some point maybe auto.”

7. Tiger’s $2.2B fund: love of the game, proven by a $400M GP commit

• The facts: Tiger downsized to $2.2B from 2021’s lofty heights, did nine deals this year versus some-odd deals in 2021, with a ~20% GP commit. Jason’s read on both eras: “the Tiger strategy in 2021, when we had an IPO a day, was the strategy of the day… now they’ve tilted to the strategy of the day” — concentration in the unlimited-growth winners.
• Rory gives Chase Coleman full credit for humility: “the guy doesn’t need the grief… I’m sure a team that smart looks back on 2021 and says that was hubris and I was wrong” — instead of retreating to “my house in Palm Beach,” they raise smaller, take the knocks, and stay in the game.
• On the commit itself, Jason’s aphorism: “money is a great truth serum. Don’t tell me what you think, tell me what you do… Do you love the game is one question. Do you love it at 400 million is a very different damn question.” Rory: LPs under-weight this signal — “when you’re trying to, God forbid, get your capital back versus play to win, that 400 million will keep these guys focused.” Jason adds the average growth fund GP commit is ~1%, often manipulated via loans or a couple of rich partners.

8. Naveen’s $500M seed at $5B — and the stair-step round game

• Rory’s investment memo, written honestly: “once you’re lucky, twice you’re good — this guy’s done it twice. Back him. Which is different than saying I see a $25 billion outcome here. I just don’t know it.”
• Harry’s insider observation on these mega-seeds (Thinking Machines et al.): the $5B sticker often hides several earlier, considerably cheaper tranches by Sequoia and a16z that “no one’s picked up on.” The mechanic is that one investor commits at 300, 500, and a billion simultaneously, blends to a 5-600M cost, and the founder announces “a headline billion-dollar raise… it’s widely up there on the list of weird stuff you see at tops of markets.”
• Jason’s defense of the practice: stair-stepping “even in a matter of hours” gives hot-deal investors a clean opt-in/opt-out — “it is a little douchy, but it actually leads to a very efficient process.” Precedent: YC has done pre/post-demo-day step-ups (6 pre → 9 post) “since I started investing in YC companies in 2014 — it just didn’t used to be up to 60 or 100.”

9. Harvey at $8B: “you just do the deal” vs. paying three years ahead

• The round: $160M led by a16z at $8B — under 2% dilution — on stated metrics of $15M ARR growing 300%, 98% GDR, 168% NDR (Jason later cites 50→150 this year, 10→150 over two). Rory’s reality check: no legal software company has ever been worth more than ~$2B (even Westlaw-type data companies sit in the fives and tens of billions), “so this implies huge TAM expansion into labor” and beyond legal into professional services — plus, per Harry, winner-take-all dominance, not an oligopoly, and no vertical unbundling by the GC-AI-style niches “we are both invested in.” Complication: Legora at $40M growing 10x and winning Europe.
• Jason’s simple-man case: “If it really has 170% NDR and 98% logo retention and it’s at 150 accelerating, this is just the bet you do. You don’t pull your hair out or have lengthy dinners about supplanting labor with AI… the guys that got in at 3 billion at the beginning of the year got an insane deal.” When Harry protests you can’t extrapolate infinitely: “Great. Then step out of venture and return your last fund to your LPs.”
• Jason’s framework: in SaaS, “the growth rate for any year was roughly 85% of the prior year’s growth.” Here the decay path decides everything: 10x→6x→3x means “any price can be paid”; 10x→3x→2x means good prices, slightly ahead of yourself; sub-100% growth means “you’re just way wrong.”
• Harry’s structural worry, endorsed by Rory: at 8B you’re paying ~3 years ahead for a public-company multiple — exactly where 2021 peaked (“the classic 100x ARRs in 2021 were two to three years ahead”). Rory’s crystallization: “It is possible to lose money on a great company. There’s only one risk you’re running… you’re paying eight for something that might be worth four. So far, no one has been wrong on price in AI. Right now, that’s the right bet to take.”

10. The Jasper risk: could deep reasoning obsolete the app layer?

• Jason’s warning, delivered hot: “We might be underestimating a GPT-6/7, Anthropic-5 risk… all of these apps could become obsolete in an era of infinite deep reasoning.” His evidence: Replit existed 8 years before Claude 4 made it work; Gamma 4 years; Replit literally today removed model selection entirely. Today’s agents — Sierra, Fin, Decagon — “are 10 times better than what we had two years ago… on the other hand they are slow, complicated to set up, and hallucinations are not a non-issue.” When reasoning goes from 5 minutes to a second, “the Harvey that we used in 2025 seems just quaint… we haven’t even begun to imagine what B2B will do with AI.”
• Harry’s pushback: defensibility lives in GTM and implementation — Wilson Sonsini and Cooley have spent a year building pilots and workflows, and a 3-5% better model doesn’t dislodge that; apps sit on top and benefit. Jason’s retort, verbatim and unsmoothed: “You have a hubristic VC argument… you don’t even know what you’re talking about here, Harry.”
• Rory’s synthesis: domain workflows give some lock-in “if the models move slowly or you stay on top of them,” but a non-negotiable discipline before any AI-app investment is trying to rebuild it on the core models — and “product-market fit is a rolling feast” at the moment… when the new model comes out, it’s pizzas late at night. Until this pace normalizes, you can be displaced." His hedge on the step function: “which I’m not holding my breath for, by the way.”

11. Benioff’s “LLMs are a commodity” — half right, and the ChatGPT-goes-away test

• Rory grants the swappability (unlike cloud, where migration is “two years, millions of dollars, a pain in the ass” — Sierra runs a constellation of models in real time) but rejects the implied conclusion that models are bad businesses. The hard-disk analogy breaks down because “capitalism works: 27 of 30 go bust and the two or three survivors extract enough profits.” Noah Smith’s airlines analogy cuts the same way: “if the antitrust people would let them, we’d only have United in San Francisco charging through the nose.” The second defense: grab the apps — Anthropic owns coding, OpenAI owns the ChatGPT consumer surface — “I can’t be sitting in the back getting swapped out by some round-robin.”
• Jason’s provocation on why Sam Altman called the code red: “if ChatGPT went away tomorrow, we’d be fine in a couple of days. Google AI Mode’s pretty good… it’d be like a TV show — we would shrug and switch.” Harry counters memory is the sticky moat; Jason: “it could be just as sticky as Yahoo Mail.”
• Harry’s rebuttal is the episode’s best consumer heuristic: “If you lost it today, would you go out and buy another one tomorrow morning?” — a phone yes, a coffee maker within the hour; and for ChatGPT, “I’d be like, I gotta go find something. I don’t think Gemini is quite there.” Yahoo Mail took “a sustained 5 years of under-execution” to lose. His market-structure gut: like the Big Bang, “how you end up ranking by the end of 26-27 will determine the trajectory for 10 years.”
• The Chinese-model undercurrent: Jason claims “so many of the hot companies use their own LLMs and they’re Chinese models — Cursor’s own model, Windsurf” (Cursor’s own model 30-40% of usage), while Rory corrects the viral a16z stat via Martin Casado: it’s ~20% of companies using open source, and 80% of that open source is Chinese — “there is a latent demand for cheap open-source models which the Chinese tech industry appears willing to fill” now that Zuckerberg stopped “churning out excellent Llamas.”

12. Airwallex, the Asia discount, and the board-condescension fight

• The setup, with Harry’s disclosure: he has a $7M Airwallex position and Keith Rabois is a friend and LP — “I am screwed both ways.” Airwallex raised $330M at $8B led by Lee Fixel at Addition; Rabois publicly alleged China data-flow risk. The valuation puzzle: roughly Ramp’s ARR at a quarter of Ramp’s $32B — and Brex sits at $13-14B “with a much worse growth rate and $400M less in ARR.” Jason and Rory decompose it: fintech multiples are bounded versus open-ended AI multiples, plus “a significant Asia discount.”
• Jason on the substance: Airwallex is “absolutely not a Chinese company” — Australian-founded, internationally headquartered — but critical mass of engineers or data centers in China means Chinese rights of inspection under standard law, just as US and EU assert theirs. “We’re dealing with a gradual unwinding of globalization, and this is what it looks like” — expect all-American competitors to weaponize it: “these guys are a bunch of commies; our stuff is in Texas; we win.”
• The row worth keeping: Rory would do the deal at 8 only with a closing condition to relocate the entire China-based services organization (“strategic fatal-error risk is exactly what a board has to do — the ability to sell the company is something a board has to work on”). Jason: “It is beyond condescending. The great CEOs will figure it out — take a mulligan and delete this section.” Rory: “No, I won’t.”
• Harry’s sales-lens read of the Rabois attack: “an objection has been added to a sales process — maybe to derail a financing, maybe to limit the round… is there maybe a racist element to it? It feels like it. I don’t like it.” Jason distinguishes Keith from a possible racist element and cites the earlier treatment of Eric Yuan at Zoom, which has more China engineers. Rory disagrees on racism — it’s “large country blocs increasingly uncomfortable with how the other’s systems are governed.” Both would own Airwallex at 8 over Ramp at 32: “you could do worse than invest in a dislocation in the market.”

13. Kalshi at $11B, Polymarket at $13B: pick whichever is better for insider trading

• Jason, fully facetious and fully serious: “You can have Kalshi at 11 or Polymarket at 13 — I want whichever one is better for insider trading… you don’t have to be Nancy Pelosi anymore. You could just be a senior engineer at Meta or Google and make millions on the side. It’s a great time to be alive.” Context: an anonymous trader made millions correctly betting Google’s top query term days in a row.
• Rory’s two-level taxonomy: pure insider betting actually “contributes to the information in the system” (why insider trading was once legal), but micro-bets you can manipulate — the third-quarter athlete prop — are “where you go from merely insider trading to something worse… the point at which the thing stops working.”
• The prediction on record: “there’s a cesspit of issues coming here… in about three or four years, just like the quiz games in the ’50s, there’s going to be congressional hearings where someone asks the Polymarket CEO, ‘How did this guy get that right 27 days in a row? Do you know your customer?’ And he’s going to go, ‘Uh, it’s all crypto-based, I don’t know.’ And they’re going to say, maybe that’s not how we’re going to roll anymore.”