BG2 Halloween: Altman + Satya + Brad Gerstner

Category: Expert Interviews · Duration: 74 min · ▶ Watch

Speakers: Brad Gerstner

Switch language → 中文

Segments (14)

  • 00:00:00 · Introduction and Partnership Reflection
    • Sam Altman and Satya Nadella reflect on their partnership, describing it as one of the great tech partnerships ever, built on early conviction.
  • 00:00:32 · Personal Check-in and The Compute Crunch
    • The hosts check in with Sam Altman about his new baby and set the stage by discussing the massive demand for compute across the tech industry.
  • 00:02:23 · The Microsoft-OpenAI Deal: Ownership and Structure
    • Brad Gerstner breaks down the financial terms of the partnership, confirming Microsoft’s investment and resulting equity stake in OpenAI’s for-profit arm.
  • 00:03:04 · The OpenAI Foundation: A Unique Non-Profit
    • Satya Nadella highlights the creation of the highly-capitalized OpenAI Foundation as a unique and significant outcome of the company’s restructuring.
  • 00:04:05 · The Journey of the Partnership
    • Sam Altman discusses the evolution of the partnership, emphasizing the initial uncertainty and the critical role of Microsoft’s early belief in their vision.
  • 00:05:26 · Foundation’s Mission and Initial Focus
    • The discussion covers the new non-profit structure, its $130B capitalization, and its initial $25B commitment to health and AI resilience.
  • 00:07:53 · Deal Deep Dive: Model Exclusivity on Azure
    • The terms of model exclusivity are clarified: OpenAI’s leading ‘stateless API’ models will be exclusive to Azure until 2032, while other products can be distributed elsewhere.
  • 00:08:53 · Deal Deep Dive: Revenue Share Agreement
    • The hosts confirm the existence of a revenue-sharing agreement where OpenAI pays a percentage of its revenues to Microsoft, also running until 2032 or AGI verification.
  • 00:09:32 · The AGI Clause and Its Implications
    • The contract’s clause for verifying AGI is discussed, which would end the exclusivity and revenue share agreements, though both leaders see AGI as not being imminent.
  • 00:11:33 · The Future of Software: Disruption by AI Agents
    • Satya Nadella explains his view that the architecture of SaaS applications will fundamentally change, with AI agents replacing traditional business logic tiers.
  • 00:12:22 · The New Economics: Search vs. Chat
    • The conversation shifts to the differing unit economics of traditional search versus AI-powered chat, and the challenge of finding a new, profitable business model.
  • 00:13:53 · AI’s Impact on Corporate Productivity and Headcount
    • The hosts discuss the trend of corporate layoffs and flat headcount at major tech companies, questioning how much is due to AI-driven productivity gains.
  • 00:16:42 · Microsoft’s Internal AI Productivity Gains
    • Satya Nadella provides a concrete example of how Microsoft’s network operations team is using AI agents to automate complex DevOps pipelines, increasing leverage.
  • 00:18:25 · The Reindustrialization of America and Global AI Policy
    • The discussion broadens to the massive capex investment in AI, its role in reindustrializing the US, and the importance of coordinated, global AI policy.

Specific Prices (11)

Timestamp Item Value Context
00:01:48 Combined Market Cap of MSFT, GOOG, META, NVDA $5 Trillion Brad Gerstner mentions the combined market cap of the four companies just hit $5 trillion.
00:02:16 New Nuclear Fission Funding $80 Billion Brad Gerstner mentions $80 billion in funding for new nuclear fission as part of the reindustrialization of America.
00:02:46 Microsoft’s Investment in OpenAI $13-14 Billion Brad Gerstner states Microsoft has invested in the ballpark of $13-14 billion into OpenAI.
00:02:50 Microsoft’s Stake in OpenAI 27% Brad Gerstner states that for its investment, Microsoft gets a 27% ownership stake in OpenAI’s for-profit business on a fully diluted basis.
00:03:20 OpenAI Foundation Equity Valuation ~$130 Billion The on-screen text states the non-profit holds equity in the for-profit valued at approximately $130 billion.
00:03:22 OpenAI Foundation Initial Focus $25 Billion The on-screen text states the foundation will initially focus on a $25B commitment across two areas.
00:11:58 OpenAI Reported Revenue Target $13 Billion Brad Gerstner mentions OpenAI’s revenues are reported to be $13 billion in 2025.
00:12:06 OpenAI Projected Compute Spend $1.4 Trillion Brad Gerstner mentions a projected compute spend of $1.4 trillion over the next four or five years.
00:12:12 OpenAI Compute Commitments $500M to Nvidia, $300M to AMD/Oracle, $250B to Azure Brad Gerstner lists large compute commitments made by OpenAI.
00:39:52 OpenAI Losses Consolidated by Microsoft $4 Billion Brad Gerstner mentions Microsoft consolidated $4 billion of losses from OpenAI in the last quarter.
00:46:07 Microsoft’s Remaining Performance Obligations (RPO) $400 Billion Brad Gerstner mentions Microsoft’s RPO is $400 billion, representing its booked business.

Bottleneck Claims (3)

  • [00:01:55] The entire AI industry is compute-constrained.
    • Evidence: Brad Gerstner cites consistent messaging from Nvidia, Google, Meta, and Microsoft earnings calls about not having enough compute to meet demand.
  • [00:18:27] The primary bottleneck for AI scaling has shifted from chips to power and the ability to build data centers fast enough.
    • Evidence: Satya Nadella states his current problem is not a lack of chips, but a lack of ‘warm shells’ (powered, ready data centers) to plug them into.
  • [00:25:27] A 50-state patchwork of AI regulation would be a big mistake for the US.
    • Evidence: Sam Altman and Satya Nadella argue it would stifle innovation, be impossible for startups to comply with, and that a single federal standard is preferable for competitiveness.

Predictions (5)

  • [00:19:29, 2-3 years or 5-6 years] There will be a compute glut at some point.
  • [00:22:33, Someday, but no specific date is planned.] OpenAI will eventually go public.
  • [00:28:21, Next year (2026)] AI agents will evolve from multi-hour to multi-day tasks.
  • [00:28:42, 2026] AI will make its first, very small, novel scientific discoveries.
  • [01:08:08, Ongoing] Future headcount growth at Microsoft will come with much more leverage than pre-AI.

Key Technologies (7)

  • Artificial Intelligence (AI): The overarching technology enabling the capabilities discussed, from language models to agents.
  • Deep Learning: The fundamental machine learning technique that OpenAI and Microsoft bet on to scale AI capabilities.
  • Stateless APIs: The specific type of AI model interface that is exclusive to Microsoft Azure, processing requests without retaining memory of past interactions.
  • Artificial General Intelligence (AGI): A hypothetical future AI with human-like or superior cognitive abilities across a wide range of tasks. Its verification is a key contractual point.
  • Codex / Agents: AI systems that can understand goals and take multi-step actions over time, such as writing software, performing scientific research, or managing tasks.
  • Azure AI Foundry: A Microsoft platform that allows enterprises to build AI applications using various models, including OpenAI’s, with tools for evaluation and security.
  • Fungible Fleet: A cloud infrastructure design principle where computing resources (like GPUs) are managed as a flexible pool that can be dynamically allocated across different workloads, geographies, and hardware gen

Companies Mentioned (17)

OpenAI · Microsoft · Nvidia · Google · Meta · AMD · Oracle · Samsung · SK Hynix · Softbank · Amazon (AWS) · CoreWeave · Crusoe · TSMC · Micron · Intel · Broadcom

Notable Quotes (6)

I think this is one of the great tech partnerships ever. — Sam Altman @ 00:01:16

One of the largest non-profit gets created. I mean, let’s not forget that. — Satya Nadella @ 00:03:17

Brad, if you want to sell your shares, I’ll find you a buyer… I think we could sell, you know, your shares or anybody else’s to some of the people who are making the most noise on Twitter whatever about this very quickly. — Sam Altman @ 00:12:37

At scale, nothing is a commodity. — Satya Nadella @ 00:48:33

The notion that business applications exist, that’s probably where they’ll all collapse, right, in the agent era. — Satya Nadella @ 00:54:02

Agents are the new seats. — Satya Nadella @ 01:03:43

Key Topics

Microsoft and OpenAI Partnership · AI Infrastructure Economics · Compute and Power Bottlenecks · AI Business Models · SaaS Disruption · AI Regulation and Policy · Future of AI Agents and Scientific Discovery · Corporate Productivity and AI

Takeaways

  • The Microsoft-OpenAI partnership is a deep, multi-faceted strategic alliance involving equity, compute, revenue sharing, and IP access, which both CEOs consider one of the most successful in tech history.
  • The primary bottleneck for AI progress is shifting from the availability of chips to the availability of power and the physical infrastructure (data centers) to house them.
  • The massive capital investment in AI is driving a ‘reindustrialization’ of the US, but is also creating boom-and-bust cycle risks due to the rapid deflation in the cost of intelligence.
  • AI is fundamentally changing the architecture of software; the traditional ‘business logic’ tier of SaaS is being replaced by AI agents, shifting value in the stack.
  • While AI is causing companies to ‘get fit’ and flatten headcount, the long-term expectation is that it will drive massive productivity gains and be a net job creator by enabling new workflows and addressing huge backlogs of work.
  • The economics of AI-driven chat/agents are fundamentally different from traditional search, requiring new business models (like subscriptions) to cover the higher marginal cost per query, but the potential value created is also much larger.
  • There is a strong desire from industry leaders for a single, clear federal regulatory framework for AI in the US to avoid a stifling 50-state patchwork of laws.