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

Summary of 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

by Harry Stebbings

1h 0m•October 29, 2026
Listen to Episode

Overview of 20VC: Anthropic Files S-1 Ahead of the Next AI IPO Wave

Harry Stebbings, Jason Lemkin, and Rory O’Driscoll spend most of this episode on the biggest AI and software-market stories of the week: Anthropic’s S-1 filing, Meta’s massive AI capex raise, Microsoft’s voluntary buyout program, the shifting economics of frontier models, and a flurry of oversized private rounds. The discussion is equal parts market analysis and VC-style IC debate, with a strong throughline: AI infrastructure spending is exploding, public and private markets are increasingly financing it through equity, and the companies best positioned to win are moving faster than traditional software ever did.

Anthropic’s S-1: The Week’s Main Event

Anthropic’s S-1 filing is treated as the defining story of the week.

Why the hosts were so bullish

  • Rory argues the filing was exceptional across nearly every major metric:
    • strong revenue growth
    • better-than-expected margins
    • favorable customer concentration
    • improving compute efficiency
    • very strong balance sheet / cash position
  • The group frames Anthropic as one of the most important startup stories in modern venture history:
    • unusually fast path from founding to IPO
    • potential for a massive liquidity event
    • a company that has scaled with unprecedented speed in AI time

The one real concern

  • The biggest caution flag is the model-economics reality:
    • roughly 80% of revenue comes from models less than six months old
    • most compute spend is still on training, not inference
  • That creates a brutal dynamic:
    • the product is incredibly valuable, but its shelf life is short
    • Anthropic has to keep reinventing the next model to keep the flywheel going
  • The hosts ultimately see this as both a risk and a moat:
    • risky because a miss could be costly
    • advantageous because Anthropic is among the few companies capable of continuously funding and shipping the next frontier model

IPO reception

  • They expect strong market reception.
  • The stock is framed as something investors will want to own, especially with the brand halo, likely index inclusion, and retail enthusiasm.

AI Infrastructure: Equity Is Replacing Debt

A major theme is that the AI infrastructure boom is increasingly being financed by equity rather than debt.

Meta and the capital markets

  • Meta’s move to raise up to $95 billion is read as evidence that the largest platforms are willing to dilute equity to keep funding compute.
  • The hosts connect this to a broader trend:
    • hyperscalers and frontier AI companies are willing to absorb dilution to keep buying chips and infrastructure
    • the spend trajectory is not slowing down

Microsoft’s voluntary buyout program

  • Microsoft offering voluntary buyouts to roughly 60,000 employees is interpreted as a strong signal of the AI-era cost reset.
  • The takeaway:
    • companies are quietly deciding they can operate with fewer people
    • AI spending is forcing organizational restructuring
  • The hosts are blunt that this is a harsh but likely durable trend:
    • the “bill” for AI compute and depreciation will increasingly be paid via headcount reduction

Cloudflare as a case study

  • Cloudflare’s earlier layoffs are highlighted as an example of a company cutting hard and still growing well.
  • The broader point:
    • layoffs are becoming a normal operating lever again
    • some are explicitly to fund AI capex
    • some are simply a sign that companies no longer need as many people as they used to

OpenAI vs. Anthropic: Competitive Reflections

The episode spends a lot of time comparing Anthropic and OpenAI.

OpenAI’s missteps

  • Jason argues OpenAI “maimed itself” through internal chaos and leadership drama.
  • In his view, those delays created room for Anthropic to become the default choice for coding and frontier-model use cases.
  • The hosts suggest OpenAI let Anthropic dominate part of the market during a critical window.

OpenAI as a possible mega-round opportunity

  • Later in the episode, Jason floats a strong contrarian idea:
    • put a large share of a fund into OpenAI at a rumored $1.25 trillion round
  • His thesis:
    • OpenAI may be mispriced because the market is too focused on short-term noise
    • it remains one of the clearest paths to enormous scale
  • Rory pushes back with his usual demand for more evidence, but the discussion reinforces the episode’s core idea:
    • frontier AI companies are now in a category where “normal” valuation logic barely applies

China, Open-Weight Models, and the Security Risk

Another major topic is the weakening position of Chinese open-weight models in the U.S.

DeepSeek and the Chinese model stack

  • DeepSeek’s rumored large raise is noted, but the mood is skeptical.
  • The hosts believe Chinese open-weight models may be losing traction in the U.S.
  • They cite a broader shift:
    • U.S. agencies are increasingly wary of these models
    • enterprise buyers are more concerned about data handling and sovereignty

The Cohere / Qwen incident

  • A major concern discussed is a reported issue where a Chinese model appeared to call servers in China and Singapore despite claiming it would not.
  • The key takeaway:
    • even if the behavior was a training error rather than malicious backdooring, it is still disqualifying for many enterprise and government buyers
    • CIOs and CISOs will be far less willing to approve models with this kind of uncertainty

Impact on the AI ecosystem

  • If Chinese open-source/open-weight models shrink in relevance, several layers of the AI stack could be affected:
    • inference providers
    • routers / model gateways
    • open-model orchestration tools
  • The hosts debate the effect, but converge on a mixed view:
    • it is bad for some infrastructure providers that depend on broad model diversity
    • it is bullish for closed frontier providers like Anthropic and OpenAI
    • the inference market remains huge even if one source of demand weakens

Cybersecurity: “Black October” and the New Vulnerability Era

The episode also frames AI as an accelerator of cyber risk.

PHP vulnerabilities and broader exposure

  • Jason says AI is making it easier to discover vulnerabilities in old systems, especially legacy PHP infrastructure.
  • The message is stark:
    • many older systems are effectively unpatched
    • AI tools can uncover hidden flaws faster than humans ever could
    • cybersecurity is entering a far more aggressive, automated phase

The “everyone gets hacked” thesis

  • The hosts argue that:
    • every exposed system will eventually be probed
    • ransomware and exploitation will become more professionalized
    • the only limit may be the diminishing pool of targets that can actually pay or be meaningfully disrupted

Private Market Frenzy: Massive Rounds Everywhere

The episode covers a wide set of large private financings and rumored raises.

Instinct’s rumored $30B round

  • This becomes a major side debate.
  • Jason argues the market for “super apps” and AI-native consumer platforms is so large that Instinct may justify the valuation.
  • Rory’s counter is classic:
    • he wants more operating history
    • he wants 100 days of evidence, not just hype
  • The debate reflects a broader tension:
    • there is enormous appetite for growth
    • but the data requirements for underwriting still matter

Other notable deals

  • Lagora: large funding round discussed as a known-good, late-stage company with strong momentum.
  • Arda: Bob McGrew’s new company raises significant capital.
  • Warp Science: Kevin Weil’s move is framed as part of a wave of talent-backed, pre-product bets.
  • War Dogs / Palmer Luckey’s defense-adjacent startup: discussed as a niche, high-conviction, sovereign-capability bet.
  • Flapping airplanes: another example of investors funding technically ambitious, founder-led projects.

What this says about the market

  • There are really two classes of deal right now:
    1. Late-stage, proven winners with clear traction
    2. Founder-led frontier bets with huge upside if they work
  • Capital is flowing to both, and the episode suggests this may continue as long as growth is extraordinary.

Key Takeaways

  • Anthropic’s S-1 is the clearest proof yet that frontier AI can reach IPO scale extremely fast.
  • AI capex is increasingly being financed through equity, not just debt.
  • Layoffs are becoming part of the AI cost structure, not just a downturn response.
  • Chinese open-weight models face mounting trust and regulatory problems in the U.S.
  • Cybersecurity risk is accelerating as AI makes vulnerability discovery cheaper and faster.
  • The venture market is extremely risk-on for AI, especially for companies with credible distribution or elite founders.

Notable Line of Thinking

  • Anthropic is not just “a hot startup” — it is being treated as a benchmark case for the entire AI era.
  • The hosts repeatedly suggest that the economics of AI are forcing a new capital regime:
    • huge checks
    • rapid model turnover
    • short product half-lives
    • much higher tolerance for dilution
  • In their view, the winners will be the companies that can keep funding the next wave of compute and model innovation without breaking.