Google’s AI Brain Drain, SpaceX's Huge Quarter, Airtable’s 90% Collapse, US Data Fuels China AI
Episode Details
The recent episode features Brad Gerstner, David Friedberg, David Sacks, and Jason Calacanis discussing major tech industry shifts, while also promoting the upcoming All-In Summit at Universal Studios, featuring prominent guests like Bill Gurley and Jensen Huang (CEO of Nvidia). A primary focus is the ongoing restructuring at Google. Key leadership changes include Demis Hassabis taking a broader role across Google and DeepMind, while legendary engineer Jeff Dean left to establish a new startup, Discovery Loop. The panel suggests this brain drain stems from the massive Capex Boom, as Google prioritizes building massive AI Data Centers over exclusive internal model development, aiming for a reliable Return on Invested Capital (ROIC). This creates an industry bifurcation. On one side are the closed, premium Frontier models dominated by a duopoly of OpenAI (creator of ChatGPT) and Anthropic (creator of Claude), both supported heavily by infrastructure from Microsoft and others. On the other side is the rapidly advancing Open source AI ecosystem, championed by companies like Meta. While Google's own frontier model, Gemini, faces internal channel conflicts, Google retains a massive distribution advantage through its Android platform—similar to how Apple dominates premium mobile monetization. Google also retains specialized AI bets through subsidiaries like Isomorphic Labs. Meanwhile, Elon Musk is leveraging SpaceX to build an AI infrastructure juggernaut. Despite a slight stock dip, SpaceX's Q2 earnings were spectacular, fueled by the massive cash flow from Starlink, which stands to grow exponentially with the success of the new Starship rocket. SpaceX is renting compute to Anthropic while simultaneously supporting internal AI projects like Grok and the coding assistant Cursor. The massive AI compute buildout has broader market impacts, affecting the valuation of specialized GPU cloud providers like Coreweave. The hosts also analyze the current SaaS Industry Slowdown, using AirTable's heavily discounted acquisition by Bending Spoons as a prime example. They note AirTable faltered when it strayed from Product-led growth to force a sales-led motion, culminating in a sale where a standard Liquidation preference ensured late-stage investors only got their money back. However, the panel notes this slowdown isn't universal; exceptional software companies with passionate users, like Figma, and robust data platforms like Snowflake, continue to perform exceptionally well. Finally, the podcast explores the geopolitical AI race between the United States and China. Reports indicate that top Chinese AI labs—including ByteDance, Alibaba, Tencent, and Moonshot AI—are catching up by purchasing high-quality AI Training Data from US-based labeling startups. The panel debates whether the US should implement strict export controls or simply rely on its continuous pace of innovation to maintain its technological lead.
The episode discusses significant shifts in the AI landscape, including talent migration from Google to startups, the bifurcation of AI models into frontier and open-source, and the infrastructure buildout by companies like SpaceX. It also touches on the SaaS industry slowdown, exemplified by AirTable's acquisition, and the geopolitical implications of China accessing US AI training data.
Portfolio lens: A basket of companies and themes at the forefront of the AI infrastructure and model development race, with a focus on both established tech giants and emerging players leveraging unique advantages.
Generated with gemini-2.5-flash-lite on 8/9/2026, 5:17:09 AM. For research only. Not financial advice.Google's AI Talent Migration and Infrastructure Focus
Google's strategic shift towards AI infrastructure and data centers, driven by high ROIC and tax advantages, may lead to a talent drain of top AI researchers focused on model development, potentially impacting their frontier model capabilities while strengthening their infrastructure business.
The departure of key AI researchers like Jeff Dean and Demis Hassabis's broader role suggests a company-wide prioritization of capital expenditure on AI data centers over exclusive internal model development. This is supported by the discussion of Google's $200 billion capex commitment and the attractive ROIC of compute infrastructure.
- Demis Hassabis has moved to a broader role across Google and DeepMind, framed as a promotion but seen by some as being 'kicked upstairs'.
- Legendary engineer Jeff Dean left to establish a new startup, Discovery Loop.
- Google's share price was down 4% on the news of Dean leaving, correlating with a $200 billion loss in market cap.
- Google has made a commitment to deploy $200 billion in capex this year in AI infrastructure data center buildout.
- The capex and accelerated depreciation make AI compute investment in the US hugely tax-advantaged.
- Scientists transitioning out are those at the core of model development, who may see better personal paths in startups given the capital allocation towards infrastructure.
- Google Cloud posted 82% year-over-year revenue growth.
- Further high-profile departures of AI researchers from Google.
- Increased competition for AI talent from well-funded startups.
- Continued strong revenue growth from Google Cloud's infrastructure offerings.
- Google may still retain significant AI model development capabilities despite talent departures.
- The company's distribution advantage through Android and other platforms could mitigate the impact of losing top model developers.
- Google's specialized AI bets through subsidiaries like Isomorphic Labs could continue to yield results.
- Monitor Google's AI research output and publications for signs of innovation.
- Track talent movement within Google's AI divisions.
- Analyze Google Cloud's revenue growth and profitability trends.
- Assess the competitive positioning of Google's Gemini models against OpenAI and Anthropic.
SpaceX's AI Infrastructure and Starlink Synergies
SpaceX is rapidly building an AI infrastructure and compute powerhouse, leveraging its Starlink cash flow and Starship capabilities, which, combined with its AI model development (Grok, Cursor), positions it as a significant player beyond its traditional space business.
SpaceX reported spectacular Q2 earnings driven by massive cash flow from Starlink, which is poised for exponential growth with Starship. The company is renting compute to AI firms like Anthropic while developing its own AI projects, indicating a dual strategy of infrastructure provision and model development.
- SpaceX's Q2 earnings were spectacular, with $7.8 billion in revenue, up 92% year-over-year.
- AI revenue (Elon Web Services) more than tripled quarter-over-quarter to $2.6 billion.
- SpaceX is renting compute to Anthropic and Google from its server collection.
- Starlink generated $2.6 billion in adjusted EBIT in Q2, with 12 million subscribers, doubling year-over-year.
- Starship test flight was successful, enabling more flights and paving the way for V3 satellites with much higher bandwidth for Starlink.
- Grok tripled tokens in July, and Cursor is on track to reach $10 billion by year-end.
- SpaceX's data center buildout is projected to increase significantly, with a focus on rapid deployment leveraging rocket manufacturing expertise.
- Successful deployment of Starship with V3 satellites, significantly increasing Starlink's bandwidth.
- Expansion of direct-to-cell capabilities via Starlink.
- Continued strong demand for AI compute rental from major AI labs.
- Growth in Grok and Cursor user adoption and revenue.
- The demand for AI compute rental may not sustain at current premium prices.
- The long-term profitability of AI compute rental is uncertain, with potential for price compression.
- Execution risks associated with scaling Starship and the massive data center buildout.
- Competition from other major cloud providers and AI infrastructure players.
- Monitor SpaceX's Starship launch cadence and Starlink V3 satellite deployment.
- Track the growth of Starlink subscribers and ARPU, especially direct-to-cell adoption.
- Analyze the pricing and utilization rates of SpaceX's AI compute rental business.
- Assess the competitive landscape for AI infrastructure and compute providers.
Watchlist
- Google's AI talent retention metrics
- SpaceX's Starlink subscriber growth and ARPU
- AI compute rental pricing trends
- OpenAI and Anthropic ARR growth rates
- Meta's open-source AI model releases
- ByteDance, Alibaba, Tencent, Moonshot AI AI model performance
Open Questions
- What is the long-term sustainability of the AI compute rental market and its pricing?
- Can Google successfully balance its AI infrastructure investments with frontier model development?
- To what extent will open-source AI models continue to close the gap with frontier models?
- What are the geopolitical implications if China significantly closes the AI gap through access to US training data?
- How will the increasing cost of AI training data impact the economics of AI model development?
Key Topics & People
.jpg?width=300){kind=avatar}
.jpg?width=300){kind=avatar}
Host of the All-In Podcast and angel investor.
Host of the All-In Podcast and venture capitalist.
AI models with accessible source code, which Ek advocates for to drive innovation.
A major market for Spotify and Neko Health, noted for its complex and highly expensive healthcare system.
.jpg?width=300){kind=avatar}
.jpg?width=300){kind=avatar}
.jpg?width=300){kind=avatar}
CEO of Nvidia who played a key role in drafting the AI safety accord's internal and external audit mechanisms.
Venture capitalist and co-host of the All-In Podcast who attended and helped shape the White House AI safety accord.
A Milan-based technology company that acquires and scales software products.
The process of showing users relevant products they didn't know they wanted, creating new economic expansion.
Legendary investor who delivered a presentation on failure analysis and COVID-19 origins.
An annual conference hosted by the All-In Podcast where industry leaders and guests present on tech, business, and politics.
.jpg?width=300){kind=avatar}
The massive capital expenditure super cycle driven by the buildout of AI technologies.
Founder and CEO of Altimeter, who presented on the state of the AI market and the broader economy.
Leader of DeepMind prioritizing reliability and alignment in frontier models.
Cutting-edge AI models developed by leading labs that push the boundaries of computational capabilities.
The venue hosting the highly anticipated All-In Summit.
A go-to-market strategy that relies on the product driving user acquisition, which AirTable struggled to scale via traditional sales.
Massive infrastructure facilities being constructed globally to house the GPUs required for AI computation.
An Alphabet subsidiary applying advanced AI to life sciences and drug discovery, led by Demis Hassabis.
A prominent Chinese AI laboratory actively competing in the global AI race.
Highly specialized datasets curated by PhDs, which are reportedly being sold to foreign labs, closing the AI capability gap.
A critical venture capital term dictating that preferred investors receive their capital back prior to common shareholders in a sale.
A widespread macroeconomic contraction negatively impacting growth metrics and valuations across software businesses.
A critical financial metric being utilized to justify the immense infrastructure investments into AI.
_-_img_02.jpg?width=300){kind=avatar}
