Aug 5, 2026 · 1h 18m · news
Leo Aschenbrenner's Situational Awareness Blows Up | Moonshot AI Raises $3.5B at $35B
Nikesh Arora · 29m spoken Rory O'Driscoll · 21m spoken Jason Lemkin · 13m spoken Harry Stebbings · 6m spoken
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Host Harry Stebbings leads a venture capital panel featuring Nikesh Arora, Rory O'Driscoll, and Jason Lemkin to analyze shifting dynamics in SaaS valuations, AI model commoditization, and cybersecurity. The group explores major acquisitions, market fund collapses, cloud provider earnings, and the massive compute infrastructure driving modern tech investments.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 9.1% of the talking time here. How this is scored →
Harry as informed peer 3.1 Guest teaching 4.6 Guest disagreement 2.6 Harry pushing back 1.9
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 44:27 Nikesh dismisses ecosystem dependency on OpenAI and Anthropic
Nikesh immediately rejects Harry's framing that the tech ecosystem faces catastrophic dislocation if OpenAI and Anthropic miss targets, stating AI demand is entirely orthogonal to vendor specifics.
Hardest push from Harry ▶ 16:30 Harry presses on LP oversight in leveraged hedge fund blowupHarry refuses to accept passive explanations for the fund collapse, demanding to know how institutional LPs permitted 4x leverage on high-beta tech equities.
Biggest teaching moment ▶ 23:18 Nikesh explains enterprise vulnerability detection and response timesNikesh educates the panel on enterprise infrastructure realities, detailing the gap between 4-day detection averages and split-second model exploit capabilities.
Harry holds his own ▶ 1:41 Harry breaks down Airtable's valuation collapse with hard numbersHarry demonstrates strong financial command by citing exact ARR figures, growth percentages, and valuation multiple compression from $11B to $1.285B.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Airtable Acquired by Bending Spoons for $1.285B | 4 | 3 | 2 | 2 | Harry opens the segment with precise financial metrics regarding Airtable's acquisition by Bending Spoons ($485M ARR, 20% YoY growth, $1.285B price tag vs. $11B peak). The guests engage in an analytical discussion regarding PE appetite, founder fatigue, and Nikesh's Mercedes vs. Tesla vs. Waymo framework for AI enablement. | |
| Leo Aschenbrenner's Situational Awareness Fund Blowup | 3 | 4 | 2 | 3 | Harry details Leo Aschenbrenner's fund blowup and presses on how LP governance allowed 4x leverage on volatile assets. Rory educates on the asymmetric psychology of LPs after a 10x year and the differing risk profiles of early versus late-entering fund investors. | |
| Anthropic Model Breaches and the AI Security Landscape | 2 | 7 | 3 | 1 | Harry sets up Anthropic's model breaches across three companies, allowing Nikesh to deliver a masterclass on cybersecurity realities. Nikesh explains that model vendor demonstrations are marketing flexes and details how 55-day zero-day patch cycles are being compressed into split-second exploits. | |
| Moonshot AI's $3.5B Raise and Model Monetization | 3 | 4 | 2 | 2 | Harry prompts discussion on Moonshot AI's $3.5B raise at a $35B valuation and the dynamics of open-weight models. Nikesh delivers his thesis that average intelligence will become free while exceptional intelligence retains pricing power. | |
| Nuclear Energy Bets and the Compute Infrastructure Crunch | 3 | 3 | 2 | 1 | Harry introduces Valar Atomics' valuation jump to $6B and Sequoia's nuclear bet. The guests explore the extreme premium placed on power and land permits, with Rory grounding the conversation in regulatory timelines. | |
| Big Tech CapEx and OpenAI/Anthropic Market Dependency | 4 | 6 | 5 | 3 | Harry challenges the panel on whether the tech ecosystem has an unsustainable single-point dependency on OpenAI and Anthropic hitting 2027 forecasts. Nikesh firmly rejects this premise, arguing that underlying enterprise and consumer demand for compute is infinite regardless of who captures model leadership. | |
| Q2 Earnings Analysis: Cloud Acceleration and Enterprise Learning | 3 | 5 | 2 | 1 | Harry invites analysis of Q2 earnings across hyperscalers and Palantir's acceleration. Nikesh and Rory explain why enterprise ingestion speed and proprietary domain context will dictate long-term survival in the current CapEx supercycle. |
Statements from this episode (25)
Stebbings: Airtable acquired by Bending Spoons for $1.285B at $485M ARR
“The company was doing four hundred and eighty five million ARR, growing 20% year on year, ultimately at a 1.285 billion dollar acquisition price.”
O'Driscoll: Airtable will reach $300M in free cash flow in two years
“I bet you two years from now it's doing 600, not 900, but I bet you it's three hundred million of free cash flow.”
O'Driscoll: VC holding periods now outlast tech platform cycles
“Now the holding period of venture is longer than the technology platform chain cycle.”
Lemkin: Airtable's low-price acquisition will trigger a wave of SaaS capitulation sales
“What I think might happen is this is the one where people capitulate both founders and investors where they say, look, folks have already had marked down since 2021, but they're not consistent. This deal in many ways was everyone capitulated. The late stage go…”
Stebbings: Citadel bought Aschenbrenner's $16B public book, making $3B profit
“Ken Griffin and Citadel bought his public book for a reported sixteen billion dollars. Ken has made out like a bandit, reportedly making about three billion dollars on the back of it in a very short amount of time.”
Arora: The average time to patch a zero-day vulnerability is 55 days
“The average time to patch a vulnerability, a zero-day vulnerability found in the wild is 55 days.”
Arora: Open-source models will automate offensive cyberattacks within months
“I think in two, three months from now, open source have distilled all these capabilities, and we'll find open source models out there, which you can find to you if you're an attacker, to actually do this on a task basis.”
Arora: Palo Alto Networks found 14,000 open-source vulnerabilities in 14 weeks
“Now we found 14,000 vulnerabilities in open source in the last 14 weeks, testing open source packets.”
Arora: Average enterprise breach detection and response time is four days
“The average time to detect and respond is four days.”
Arora: Foundation models will not disrupt Palo Alto Networks' perimeter security
“So when people tell me, oh, OpenAI is going to eat my lunch, or Mythos is going to eat my lunch, guess what? There are no perimeter security scenario, which means I still need to block the bad guy. They need to be the ingredient in my product. They're not goin…”
Arora: Only 1,200 customers have deployed Palo Alto's rapid detection system
“The only problem is I only have 1200 customers who bought and deployed it. I need to get the rest of the world to go buy it and deploy it.”
O'Driscoll: Open-weight models act as a price drag on closed AI models
“One of the really interesting things about these open weight models is the impact they're having and the ability to be a drag on price for the US, you know, closed source frontier model companies.”
Arora: Average AI intelligence will become free; exceptional models will command payment
“Average intelligence is going to be free in the long term, and the average intelligence will keep getting better. Exceptional intelligence will be paid for.”
O'Driscoll: Customer support applications will not require or use frontier AI models
“Customer support will not be using frontier models.”
Arora: Within four years, customers will pay for compute, not model intelligence
“I think in the next three, four years we won't be paying for intelligence, we'll be paying for compute through our nose.”
Stebbings: Sequoia leads funding for Valar Atomics at a $6 billion valuation
“Valar Atomics triples to six billion dollar prices Sequoia bets on nuclear for AI. It's a three-year-old small modular reactor company raised at two billion, now Sequoia leading around at six.”
O'Driscoll: Cheap Electricity Is Impossible Without Nuclear Power
“You have to be wildly supportive because there's no way to get cheap electricity without doing nuclear.”
Arora: Free consumers account for 70% of current AI compute demand
“I still posit 70% of the compute demand for AI is being consumed by consumers who are getting a free ride.”
O'Driscoll: OpenAI and Anthropic drive 70% of Google and AWS compute backlogs
“In other words, because they're 70% of the Google Compute backlog, the 70% of the Amazon backlog.”
Arora: Palo Alto Networks is transcribing all historical customer cases for AI
“I'm talking about the context needed to train the ecosystem, which means I've got every customer case that ever happened at Palo Alto getting transcribed.”
Arora: Domain context will equal model intelligence in value within five years
“So I think we're coming to a world where in the next five years domain becomes equally important with the model intelligence.”
Arora: Consumer applications will be rewritten for AI agents within a decade
“I think every consumer app will get rewritten in the next five to 10 years. You know, why would I not have my agent talk to my DoorDash app or Uber app?”
Arora: Enterprise AI Success Depends on Fast Training Data Creation
“I think the enterprise's ability to absorb this or digest this or perhaps leverage this to their advantage depends on their ability to create training data as fast as they can. And I think not enough people are focused on training data.”