20VC: Anthropic Files S1 Ahead of $2TRN IPO | Instinct's Rumoured $30 Billion Fundraise | Microsoft Laysoff 27,000 While Meta Taps Equity Markets for $95 Billion Capex Spending Spree

Anthropic

Across the shows: 68 speakers in other episodes, 66 bullish and 55 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersBullish

  Anthropic's S-1 was top-tier across margins, cash, customer concentration and compute efficiency; skeptics now carry the burden of proof.

  He says the S-1 was great in almost every dimension: margins and operating income were better than expected, operating income held up after stock-based comp, customer concentration was good, compute efficiency was strong, and the cash balance sheet was striking. He says the questions are few and the haters have to 'go pound sand', since the burden of proof is on the naysayers.

  “In other words, the [S-1] was great in almost every dimension... It was great, better than expected on margins, op income, even op income after stock based comp. It was great on customer concentration.”at 4:34

• Jason Lemkin · SaaStrBullish

  Anthropic's five-year path to IPO and fast liquidity makes it unprecedented and better than typical seed-to-liquidity timelines.

  He says Anthropic went from seed to IPO in five years versus 24 for SpaceX, which is unprecedented. He argues that getting out via the IPO in under six years to distributions beats seed deals that take about 18 years to full distributions and other deals that take 10-12 years to IPO, so a single Anthropic outcome can return the fund faster.

  “to get here in five years from [a seed] startup, it's just beyond unprecedented... less than six years to distributions, I mean, I know Rory will will roast me for it, but you got to think, why would I do anything else?”at 5:33

• Rory O'Driscoll · Scale Venture PartnersBullish

  Anthropic's short model lifecycle is an advantage for a leader able to fund constant new model builds.

  The S-1 shows 80% of revenue comes from models less than six months old and that roughly 70-80% of compute cost goes to training, not inference. Once trained, models are very profitable to serve, but each one has a short shelf life, so the company must keep building. He argues this is scary only if you miss once, and that it's an advantage because very few players can write a $10B+ compute check, with Anthropic spending about $8B year to date and planning $16B next year.

  “80% of the revenue comes from models that are less than six months old... 70%, 80% of the compute costs are for training, not for inference.”at 7:28

• Jason Lemkin · SaaStrBullish

  Anthropic's IPO should get a good retail run in the next two quarters, given its growth metrics and Nasdaq 100 inclusion.

  He does not expect Anthropic to miss a quarter in the next two quarters. He cites 900% net revenue retention, which he admits may really be 300-400% because it includes pilots and trials. He thinks that, together with the brand and the Nasdaq 100 inclusion, it will probably have a good retail run.

  “within the next two quarters... With 900% NRR, even if in reality it's more like 300 to 400% because it includes like pilots and trials. I just think that plus the brand plus being in the Nasdaq 100, it'll probably have a good retail run.”at 12:54

• Jason Lemkin · SaaStrBullish

  Anthropic owned the coding market for about a year, and coding is the killer app, so it should keep its lead.

  He says Anthropic has owned the playbook for almost twelve months and that there is no point using another coding solution, since coding is the killer app. Open-weight models have reached rough parity, but Anthropic still leads in coding. He credits OpenAI for letting Anthropic take that position.

  “Anthropic own the playbook for the chessboard for almost twelve months. There's no point in using any other coding solution and coding is the killer app as Rory said 100 times and the math is true.”at 14:12

• Rory O'Driscoll · Scale Venture PartnersBullish

  If US open-weight alternatives fail to appear, closed labs like Anthropic gain pricing power, which is bullish for Anthropic.

  He asks whether a US company like NVIDIA can build a capable open-source model, given that roughly 75% of compute cost goes to training. If no one does, closed-source providers are left alone and their economic leverage rises. He calls the removal of Chinese open-weight models bullish for Anthropic, and says investors should buy more Anthropic in the IPO.

  “It would be interesting to see how advanced a model a US company like NVIDIA can build... If someone doesn't do it, then all you have is a closed source providers, and obviously, their economic leverage goes way up.”at 25:43

• Jason Lemkin · SaaStrBullish

  Anthropic is only 0.1% of Fortune 500 budgets; it should reach 2-3%, implying major growth ahead.

  Anthropic holds only about 0.1% of Fortune 500 budgets today. Even the most conservative estimates, based on what Anthropic and OpenAI have achieved, point to 2-3%. Moving from 0.1% to 2-3% is significant growth ahead of the investment. He also points to Anthropic's 900% net revenue retention as the kind of evidence of durable growth he wants to see.

  “It's only 0.1% of the fortune five hundred budgets today. Clearly, it's going to be in the two to 3% range even by the most conservative estimates”at 46:52

OpenAI

Across the shows: 26 speakers in other episodes, 11 bullish and 20 bearish calls, past 30 days

• Jason Lemkin · SaaStrBearish

  OpenAI's governance turmoil and missteps let Anthropic take a lead in coding; OpenAI's IPO may be slightly worse.

  He says OpenAI, despite being a larger company, let Anthropic control part of the market for roughly five years or ten years in AI time. He points to the CEO drama and the nonprofit-to-profit fight as costly distractions. He also says OpenAI's IPO will probably be slightly worse than Anthropic's, as it is less capital efficient and slightly slower.

  “despite startup doing better than them, they basically let Anthropic control part of the market for almost a which is like five years or ten years in AI time, right?”at 14:09

• Jason Lemkin · SaaStrBullish

  OpenAI at $1.25T is mispriced; a 20%-of-fund bet offers a path to a $10T outcome.

  He sees a dislocation: the market is misreading OpenAI as struggling when the numbers show otherwise. Anthropic is on fire, SpaceX has done well, and the models are competitive, with OpenAI having closed its competitive gaps. He compares it to the Anthropic round at about $183 billion, which was a dislocation that made money. He says both OpenAI and Anthropic can become $10-20 trillion companies. A 20% position at $1.25 trillion would return the fund even at a $10 trillion outcome, and worst case he doubles his money. Unlike priced-to-perfection rounds such as Inflection or Instinct, OpenAI is not priced to perfection.

  “And I think we should put 20% of the fund into OpenAI at the $1,250,000,000,000 round. Think that it is mispriced.”at 45:50

• Rory O'Driscoll · Scale Venture PartnersBullish

  OpenAI's Codex has reached parity with Claude Code; markets are too superficial to see it.

  Every coding team his firm works with has switched from Claude Code to Codex in the last three to six months, so OpenAI has achieved parity in the product that matters. He says the quality gap has closed significantly from the prior generation, but the markets have not noticed. That is why he argues for putting 20% of the fund into OpenAI, since nothing smaller would be enough.

  “I know every coding team that we work with has switched from [Claude] Code to [Codex] in the last three to six months... They've achieved parity... The markets are too are are too superficial to see it.”at 48:06

Instinct

• Rory O'Driscoll · Scale Venture PartnersBullish

  Would unwaveringly invest in Instinct at $30B; uncapped upside as a near-universal messaging replacement.

  He points to Spark's early pre-revenue bet, now worth hundreds of times $75 million, as evidence of what outsized private outcomes look like. Unlike a competitive market with about five rivals and capped upside, Instinct could become everyone's app, a replacement for iMessage plus, worth around $200 billion. He says the three-way payoff matters (if it's amazing, good, or fails), and that every big fund should be fighting to put money in Instinct today.

  “This is why I would unwaveringly do instinct at 30. If I was coming into the IC, I would unwaveringly do it.”at 39:12

• Harry Stebbings · 20VCBullish

  Instinct at $30B is a no-brainer: downside protected by hyperscaler buyers, upside as the Western super app.

  He argues the downside is protected because the team and strategic market would attract Microsoft and other hyperscalers, who would buy this in a heartbeat. The upside is the long-promised super app for the West, covering finance, identity and security, which could expand into subscriptions, ads, and transactions across all sales and ecommerce. He calls it one of the most no-brainer deals for a fund like Andreessen Horowitz. He concedes the data is limited but argues the potential payoff justifies the bet.

  “Microsoft would buy this in a heartbeat... One, so your downside, I do think, is protected. And I think your upside is really, as we said, the super app of the West”at 40:36

• Jason Lemkin · SaaStrBearish

  Passes at $30B Instinct valuation: no revenue, unclear travel-based model, negative gross margins.

  Instinct has no revenue and an unclear business model based on travel, and its growth is in a fungible product with negative gross margins. He wants about 100 days of durable traction before paying a growth-round price. He has seen only about a month of traction, so at $30B he would pass. He is not questioning the team or the strategic market, and he says he is not convinced by the liquidation preference either.

  “I just think with no revenue and an unclear business model based on travel, I just think we should unfortunately pass at $30,000,000,000 I just want to see one hundred days of something besides negative gross margin growth of a fungible product.”at 41:57

Meta Platforms

Across the shows: 59 speakers in other episodes, 47 bullish and 33 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersNeutral

  Meta's $95B equity raise reflects its first free-cash-flow-negative quarter, not a broader distress signal.

  He argues the Meta raise is unrelated to the chip selloff. Meta's core business is one of the best on the planet, generating about $33B of cash flow a quarter. This quarter capex went above free cash flow for the first time, so Meta issued equity to cover it. The move reflects funding needs, not a weak franchise.

  “their core business is one of the best business on the planet. It spits out money... This is the first quarter they went negative, so they just did some equity. So that's that, and it's totally unrelated.”at 21:28

Alibaba Group Holding

Across the shows: 2 speakers in other episodes, 2 bullish and 0 bearish calls, past 30 days

• Jason Lemkin · SaaStrBearish

  Alibaba's Qwen model secretly uploading data to its servers makes enterprise trust in it very hard to win back.

  Qwen, Alibaba's model, was found calling Alibaba servers in China and Singapore and uploading data despite saying it did not. He thinks it is probably a training error rather than a deliberate backdoor, but says that doesn't matter to enterprise security teams. He adds that the government could exploit or take over such a channel, and that Alibaba said nothing publicly.

  “[Qwen's] existing model was calling to Alibaba servers, both in China and in Singapore, pinging them and uploading data to them when it said it wasn't.”at 23:42

Warhounds

• Jason Lemkin · SaaStrUncertain

  Warhounds' military dog market may be small, but the sovereignty angle and founder channels make it real.

  He is unsure whether the market is worth the valuation and does not know how big the models are. He notes that key national security components, such as rugged laptops and the dogs themselves, come from outside the US, which is a sovereignty gap. The founder Palmer Luckey sits at the core of the defense industry and has the channels to sell, so the markets are real and visible to him. Maybe the TAMs are small, but the models make sense.

  “How big these models are, I don't know... And maybe the TAMs are small, but the models make sense.”at 51:31

NVIDIA

Across the shows: 45 speakers in other episodes, 38 bullish and 15 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersBullish

  Nvidia looks cheap on a forward P/E basis, so tech valuations are not a bubble.

  He says stocks like Nvidia look cheap on a forward P/E basis. He notes the aggregate market's earnings have been inflated by one-off gains from private investments, so trailing multiples look high. Forward estimates give a cleaner view, and on that basis the leading AI names are not expensive. He acknowledges a long cyclical debate but argues the valuation case is favourable.

  “On a current PE basis, stocks like Nvidia on a forward PE basis look cheap.”at 53:34

Venture capital & startups

• Jason Lemkin · SaaStr

  Risk of backing Chinese-linked AI startups has risen after the DeepSeek ban and the Qwen data issue.

  He says VCs would not have taken direct-investment risk in Asian open-weight startups two years ago, and that the Manus unwinding made such deals unlikely to be repeated. Recent events, including the US blocking DeepSeek and the Qwen data issue, are negatives that have raised the risk level a lot. He still backs the IC decisions, because growth is strong enough to justify the risk.

  “the risk level didn't just go up in the last week. A lot. And these deep seek being blocked by the US government and this whole Quinn thing and dialing home... These are negatives.”at 32:27

• Jason Lemkin · SaaStr

  Mega-rounds for AI labs depend on capital pools that may not last; late-growth funding has a structural issue.

  AI companies need enormous capital to pay for inference and compete, so even large rounds from sovereign funds like GIC are not a vanity raise. But the total pool of money for these companies is finite, and he does not know how big it is or when the merry-go-round will end. He calls this a structural issue in the late growth phase, although money currently feels infinite.

  “I just don't know how much it is when that merry-go-round ends. We just have a structural issue in this late growth phase.”at 38:31

• Rory O'Driscoll · Scale Venture Partners

  Market shows wild appetite for proven executives with big ideas at high prices and little derisking.

  A plethora of this week's deals involve proven executives with big opportunities and hairy technical challenges, such as Arda and Flapping Airplanes. Investors pay fairly high prices for them with relatively little derisking, because the executive is talented and the opportunity is huge if it works. In contrast, late-stage known companies are easier to price, since the outcome depends less on one person.

  “There's just clearly a wild appetite for that at fairly high prices with relatively little derisking.”at 44:50

AI infrastructure & data centers

• Rory O'Driscoll · Scale Venture Partners

  AI capex is increasingly funded by equity issuance from companies with cheap capital, and the trend will continue.

  He says the funding shift is from debt to equity: Anthropic is about to issue around $100B of stock, Meta issued $70B and Google about $80B earlier this year, and SpaceX has done an IPO. He argues these companies have cheap cost of capital and are willing to accept dilution to fund compute. He concludes that this is a macro trend that will keep happening, so the spend will be there.

  “we're equitizing the issue... Anthropic is about to issue plus or minus a $100,000,000,000 of stock. Meta issued $70,000,000,000 of stock. Earlier this year, Google issued about $80,000,000,000 of stock... This trend is not stopping.”at 10:40

• Rory O'Driscoll · Scale Venture Partners

  AI capex depreciation will hit hyperscaler earnings in 2027, forcing headcount cuts to protect operating income.

  In 2026, the chipmakers are booking gains while the companies buying the capex are not yet booking the depreciation. As they look to 2027, depreciation will start hitting Microsoft, Google, Meta and similar companies. To keep operating income up, these companies will have to push hard on headcount. He calls this the moment the AI bill comes due, and says it may feed the anti-AI backlash.

  “as you start to look to '27, that depreciation is going to start hitting in Microsoft and Google and Meta and places like that. And to keep the operating income up, you're going to be really pushing on headcount.”at 18:16

Open-source AI models

• Jason Lemkin · SaaStr

  Chinese open-weight models are finished in the US after DeepSeek's federal ban and the Qwen data-upload issue.

  DeepSeek is being banned across federal agencies. The Cohere-Qwen episode showed that Alibaba's Qwen model called servers in China and Singapore and uploaded data while told not to. He argues no CIO or CISO will approve such a model, and that enterprises face job-loss risk if they keep using these models. He calls it the beginning of the end.

  “I think that the Chinese open weight models are pretty much over in The US. I think it's the beginning of the end.”at 23:12

• Rory O'Driscoll · Scale Venture Partners

  Open-weight inference demand will survive: even 10% of a huge inference market is still big business.

  He added up the inference competitors' revenue, around $5B, and compared it with the roughly $100B already going to the two closed frontier models. He argues that if 10% of spend will accept lower-quality open-source output to save real money, there will be real demand. He says he does not know whether it reaches $30B, but a 10% share of a huge market is still big.

  “The inference market is still so big... If 10% of the spend is willing to kind of do... Put up with the brain dead of open source to... save real money, then there's gonna be demand for that.”at 34:08

Credit markets

• Jason Lemkin · SaaStr

  Debt financing for AI labs like Anthropic and OpenAI is getting harder and bank appetite for neolab debt is limited.

  Broadcom had to guarantee most of Anthropic's debt for its chip purchases, and the debt came at higher interest rates than expected. He reads this as a sign that fundraising is getting harder even for the top labs, and says banks are finding it harder to sell down neolab debt. He calls this the one real thing to be nervous about. Rory counters that banks have hit their limit, but that the market is shifting to equity issuance instead.

  “going forward fundraising even for Anthropic and OPI and even if it's somewhat indirect is it's getting harder... it's getting harder for the banks to lay off and sell the Neolab debt, I think that's the real only one to be nervous about.”at 10:06

AI & jobs

• Jason Lemkin · SaaStr

  Layoffs will accelerate as companies fund AI infrastructure and cope with endless competition, with no stability.

  He cites Cloudflare cutting 20% of staff while its growth accelerated, with no visible business downside. He says SaaS companies used to reach a stable plateau, but now there is none, and competition from new startups is constant. He sees layoffs 2.0 and 3.0 coming, both to pay for AI infrastructure and to escape the performative AI environment.

  “There's no planes of stability at all... So I just think layoffs are just gonna accelerate because we have no choice.”at 20:21

AI chips & semiconductors

• Rory O'Driscoll · Scale Venture Partners

  Chip stocks have been rising since the Anthropic S-1 and memory chips are strong, though CoreWeave weighed earlier.

  Chips had fallen about two weeks earlier on CoreWeave concerns, but since the Anthropic S-1 came out, chip stocks have bounced because the consensus is that the filing looks good. He points to DRAM being sharply higher and Korea's SK Hynix up 10%. He says the semiconductor capex business has had a good back half.

  “since the Anthropic S1 hit, chips have been bouncing up because the consensus is that looks good. One thing that the S1 was good for was chips. So those stocks are actually up.”at 21:08

Cybersecurity

• Jason Lemkin · SaaStr

  AI makes hacking near-instant; unpatched old sites and laid-off IT teams mean attacks worsen for a year or two.

  Since the models got good, anyone can be hacked in seconds and attempts are accelerating. The PHP vulnerability is worse than it looks: PHP is still roughly 40% of the internet, much of it old sites that will never be patched. Government and nonprofit sites took their sites down, and IT teams were laid off years ago. He expects a year or two of worsening exposure as every threat vector is exposed almost in real time, with ransomware professionalizing. The only limit he sees is that few can pay large ransoms and few systems control dams or other natural resources, so hacking may decline for a while.

  “Now anyone can be hacked in seconds. People attempt to hack faster every day. So everyone's going to get hacked.”at 35:11

US stock market outlook

• Rory O'Driscoll · Scale Venture Partners

  Market is expensive on trailing P/E but cheap forward; only 40-50 AI and tech stocks are working.

  On a ten-year P/E basis the market is at an all-time high. On a forward P/E basis it is actually quite cheap, because trailing earnings have been boosted by one-off gains from privately held investments in public companies. The aggregate market is not hurting, but about 450 stocks are struggling. Only 40 or 50 S&P stocks, mainly tech and AI, are working, and everything else is quietly subpar. He calls it a really weird market.

  “On a ten year PE, they're at an all time high... Interestingly, on a forward PE basis, they're actually quite cheap because... earnings have been so amazing in part driven by one off gains from privately held investments in the public companies.”at 53:34

IPOs, M&A & deal activity

• Rory O'Driscoll · Scale Venture Partners

  IPO activity is weak as everyone waits for a trillion-dollar listing, so platforms should fund privates now.

  Despite a market where only a few dozen tech and AI stocks are working, there have not been many IPOs. Everyone is waiting for a trillion-dollar listing, so the whole world is in suspended animation. He concludes it is a perfect time for big platforms to put large sums into private companies, because demand for capital there is strong and the public exit path is blocked.

  “And then despite that, there haven't been a ton of IPOs because everyone's been waiting for this trillion dollar thing, and the whole world is in suspended animation.”at 54:21

Wealth, inequality & billionaires

• Rory O'Driscoll · Scale Venture Partners

  The billionaire tax will hit income tax as wealthy people leave; billionaires will not accept losing 5%.

  Anyone who thinks the billionaire tax has no impact has never tried to take 5% of a billionaire's money. Billionaires react badly, especially given voting-control criteria and the prospect that the legislation can be repeated soon. He says the tax has already driven out enough billionaires that the state will see an ordinary income tax hit, whatever the vote outcome.

  “The billionaire tax is up and we'll know in a week or two... people who think there's no impact from that have never met a billionaire when you've tried to take 5% of their money away.”at 55:55