The Twenty Minute VCInstinct Worth 30BN? | Microsoft Cuts 27,000 Jobs While Meta Taps Equity Markets for $95B CapEx
CHAPTERS
- 0:05 – 1:16
Anthropic’s S-1 lands: what matters and why it’s the week’s biggest story
Harry frames Anthropic’s newly filed S-1 as the defining story of the week and sets up a breakdown of what to focus on. Rory is asked to pick the metrics and key takeaways, teeing up a detailed discussion of performance and risks.
- •Anthropic’s S-1 is positioned as the central event to analyze this week
- •The group plans to separate signal from noise and focus on the questions the filing raises
- •Harry hands the discussion to Rory to choose the most meaningful S-1 metrics
- 1:16 – 2:18
Rory’s read: a top-tier filing with surprisingly few red flags
Rory argues the S-1 is strong across nearly every dimension, claiming critics are temporarily on the back foot. He highlights positive signals across revenue, margins, customer concentration, compute efficiency, and the balance sheet, while noting a few unanswered questions remain.
- •Rory says the S-1 looks great on most dimensions and should be well-received
- •He notes the filing doesn’t provide a Q3 revenue number, leaving an open question
- •He emphasizes margins, operating income (including after stock-based comp), and compute efficiency as standout positives
- •He points to strong customer concentration characteristics and a notably strong cash position
- 2:18 – 4:49
Speed to liquidity and “why bother with anything else?” venture implications
Jason and Rory focus on the unprecedented pace from founding to IPO-scale outcomes and what that does to investor psychology. They discuss how mega AI wins compress venture time horizons and make traditional long-dated paths feel less attractive by comparison.
- •Jason says Anthropic’s five-year path to IPO-scale is beyond precedent and reshapes incentives
- •He contrasts fast liquidity with long distribution timelines in traditional venture
- •Rory agrees these outcomes make founders and investors ask why they’re working on anything else
- •They discuss how early investors benefit disproportionately as these outcomes dominate fund performance
- 4:49 – 7:01
The scariest S-1 signal: six-month product shelf life and the cost to keep winning
Rory calls out an eye-catching metric: most revenue comes from models less than six months old, implying rapid obsolescence. He explains how the compute cost appears heavily weighted to training, making serving profitable—but only if the company continually spends to build the next generation.
- •Rory highlights that most revenue comes from models under six months old
- •He notes the filing doesn’t lay out gross margin in a straightforward way
- •He interprets the compute breakdown as training-heavy, with serving comparatively profitable
- •He frames the dynamic as risky if a company misses a cycle, but potentially an advantage for a leader able to fund constant reinvention
- •He emphasizes the scale of spending required just to stay in the game
- 7:01 – 9:55
Is AI debt financing cracking? Broadcom’s role and the limits of bank appetite
Jason argues the most worrying signal isn’t in the S-1 itself: difficulty placing or financing large AI-related debt, citing a situation where Broadcom had to guarantee much of a deal. Rory partially pushes back, suggesting the system is shifting from debt to equity issuance for companies with cheap capital and high market caps.
- •Jason says the concerning metric is difficulty raising certain debt even for top-tier AI names
- •He points to a chip-related financing structure where Broadcom guarantees much of the debt and rates come in higher
- •Rory says the debt digestion limit may be real, but the bigger trend is the “equitizing” of funding needs
- •Rory argues equity dilution is keeping the spending boom alive for the biggest players
- 9:55 – 11:18
Will the IPO be a monster? Indexing boosts, retail demand, and near-term expectations
The group discusses how public markets may receive the IPO and whether index inclusion matters. Rory expects strong demand, while Jason argues indexing and brand effects can amplify a retail run and says it’s hard to bet against near-term execution given growth and retention metrics discussed.
- •Rory says the company is compelling enough that it should have a strong reception
- •They discuss index inclusion mechanics and how quickly certain funds may need to own it
- •Jason argues indexing can provide a retail boost even if the company doesn’t need it
- •Jason references very high net revenue retention but cautions about how to interpret it in this context
- 11:18 – 13:04
What OpenAI should learn: self-inflicted setbacks and how Anthropic pulled ahead
Harry asks what OpenAI leadership should take from Anthropic’s S-1. Rory suggests OpenAI may regret decisions that let Anthropic outperform and says the only option is to execute and compete aggressively, while Jason criticizes OpenAI’s internal disruptions and missed opportunity in a fast-moving market.
- •Rory says OpenAI can’t change the past but should reflect on decisions that allowed a competitor to do better
- •Rory suggests aggressive competition, including pricing, could be used tactically
- •Jason argues OpenAI endured major self-inflicted drama that would have sunk most startups
- •Jason claims Anthropic controlled key parts of the market for an extended period in “AI time”
- 13:04 – 16:21
Microsoft’s voluntary buyouts: the “scarlet letter” problem and morale fallout
Harry introduces Microsoft offering voluntary buyouts broadly across employees. Jason and Rory agree it’s a harsh signal that a large swath of staff are considered unnecessary, and they debate how this compares to layoffs in terms of humanity and culture impact.
- •Harry notes Microsoft offered buyouts to a large portion of employees
- •Jason argues it signals the company doesn’t want a large group of people, even if few take it
- •Jason calls it a worse “scarlet letter” than straightforward layoffs
- •Rory says winners are those who already planned to leave and can capture severance optionality
- •They discuss uncertainty in how many accept and what that implies operationally
- 16:21 – 17:43
Why AI CapEx could drive more layoffs: depreciation hits and the 2027 bill coming due
Rory connects layoffs to the financial lifecycle of massive AI infrastructure spending, arguing depreciation will start pressuring reported results. Jason adds that paying for GPUs and infrastructure creates an easy narrative that fuels backlash, and he expects headcount pressure to intensify.
- •Rory explains a timing dynamic where sellers of chips book gains while buyers haven’t yet absorbed depreciation
- •He expects depreciation pressure to rise, making it harder to maintain operating income
- •Rory points to 2027 as a period when this pressure may become more visible
- •Jason says the story becomes “to pay for infra, we get rid of people,” which can fuel anti-AI sentiment
- 17:43 – 19:27
Cloudflare’s 20% cut: a CEO playbook for operating in a no-stability market
Jason points to Cloudflare’s large layoff despite strong performance as a template that other CEOs may follow. He argues there are no longer “planes of stability” at typical SaaS scale, making aggressive cost actions more common as competition accelerates.
- •Jason highlights Cloudflare laying off 20% after an “epic year” and continued acceleration
- •He frames it as a case study for leadership decision-making under intensified competition
- •He suggests some cuts are about paying for infrastructure, while others are simply reducing drag
- •He argues stability plateaus are disappearing across company sizes, increasing stress and cost discipline
- 19:27 – 22:00
Meta taps equity markets for massive CapEx: separating chip fear from funding reality
Harry raises Meta’s equity raise tied to large CapEx plans and broader market volatility around chips and “neo clouds.” Rory separates the narratives, arguing the equity move reflects CapEx outpacing free cash flow temporarily, while the Anthropic S-1 reinforces continued demand for semiconductors.
- •Harry cites Meta raising equity alongside market volatility in chips and related companies
- •Rory says these are distinct issues: Meta turning free cash flow negative due to CapEx and choosing equity issuance
- •Rory argues semiconductors benefited from renewed confidence after the S-1 details
- •He contrasts smaller players’ turbulence with large, well-funded spenders that can keep buying chips
- 22:00 – 24:48
Chinese open-weight models in the US: bans, “dialing home,” and enterprise trust collapse
Harry asks about China-related AI funding and DeepSeek, and Jason argues Chinese open-weight models are becoming untenable in the US. He describes an incident involving Qwen behavior that triggers fears of data being sent to overseas servers, and he frames it as a career-risk for CIOs and CISOs to approve such models.
- •Jason predicts Chinese open-weight models are nearing the end of viability in the US
- •He points to increasing restrictions across federal agencies as a major driver
- •Jason describes concerns about models contacting overseas servers and uploading data contrary to expectations
- •He argues that even rare occurrences are unacceptable at enterprise scale and become a job-risk for security leaders
- 24:48 – 35:03
What happens to routers and inference providers if Chinese models decline? (Modal/Fireworks/OpenRouter)
Harry pressures the implications for model routing and inference companies if Chinese model options shrink. Rory and Jason discuss whether US alternatives can meet demand and how the market could reallocate toward closed models, while Jason warns OpenRouter’s surge could prove transient if model mix shifts quickly.
- •Harry asks what model contraction means for routers and inference providers that benefit from broad model choice
- •Rory argues demand for open-weight alternatives remains, but the cost to build them is immense
- •Jason and Rory agree the trend is bullish for leading closed-model providers if open-weight options shrink
- •Jason suggests OpenRouter’s growth may have been tied to a temporary gap and could reverse quickly
- •Rory argues even a smaller share of a huge market can still support meaningful businesses
- 35:03 – 37:32
“Black October” and AI-driven security exposure: old vulnerabilities found fast and exploited faster
Harry brings up AI-assisted discovery of serious software flaws and asks what it means. Jason argues the big shift is that any system can be attacked quickly now, and legacy stacks—especially older, unmaintained sites—create a large vulnerable surface that won’t be patched in time.
- •Jason says the new reality is that everyone becomes hackable quickly as AI capability rises
- •He argues legacy software and unmaintained sites amplify the risk because patches won’t be applied
- •He expects threat discovery and exploitation to become near real-time across many vectors
- •Rory agrees that falling behind current defenses creates a widening security gap
- •Jason predicts attacks will professionalize further, with criminals focusing on targets that pay
- 37:32 – 45:10
Instinct at $30B: super-app upside vs. “100 days of traction” discipline
The conversation turns to rumors of Instinct raising at a $30B valuation and whether it’s justified. Harry argues it could become a Western super-app across identity, finance, and security, while Jason vetoes the idea without more time-series evidence, insisting on clearer traction and risk control for large growth checks.
- •Harry frames Instinct as a potential Western “super app” and says he’d back it at the rumored price
- •Jason questions how long equity enthusiasm can fund capital-hungry companies and when the cycle ends
- •Jason insists on needing multiple months of traction and clearer business model signals before committing at that valuation
- •They debate downside protection, acquisition optionality, and the risks of negative gross margins in consumer products
- •Rory frames the decision as a payoff-matrix tradeoff: capped-but-clear markets vs. uncapped consumer outcomes
- 45:10 – 51:28
Why money is flowing to “proven AI executives” and Jason’s 20% OpenAI allocation idea
Rory observes a pattern of large financings driven by credibility of elite operators tackling big technical problems with limited de-risking. Jason proposes a concentrated bet—putting 20% of the fund into OpenAI at a stated valuation—arguing it’s a rare “dislocation moment,” while Rory says it’s not crazy depending on future funding dynamics.
- •Rory notes investor appetite for big rounds anchored in proven leadership and hard technical missions
- •He distinguishes operator-led binary bets from later-stage, more easily priced growth rounds
- •Jason argues for an unusually concentrated OpenAI position based on perceived mispricing and competitive convergence
- •He claims the upside case is driven by market expansion and closing competitive gaps
- •Rory says the idea can be rational risk-adjusted if it’s near the last private round and execution improves
- 51:28 – 1:00:16
WarHounds at $1.5B and the return of mega-fund behavior; Zuckerberg vs California billionaire-tax drama
They react to WarHounds’ high valuation and struggle to quantify the market, concluding it’s largely a founder-driven bet. The episode closes with commentary on mega-funds raising larger and faster, plus discussion of reported legal and board dynamics around Zuckerberg, California, and the implications of proposed billionaire taxes.
- •They discuss WarHounds’ premise and admit limited ability to model TAM, implying the bet is primarily on the founder
- •Jason links the theme to sovereignty-focused gaps that well-connected founders can sell into
- •They discuss mega-platform funds raising bigger, deploying faster, and returning to market quickly
- •Jason and Rory discuss reported tensions involving lawyers, boards, and negotiation dynamics around Zuckerberg and California
- •Rory argues billionaire taxes materially affect behavior because agency and control matter to wealthy founders
