Oct 31, 2025 · 1h 14m · bg2-pod
All things AI w @altcap @sama & @satyanadella. A Halloween Special. 🎃🔥BG2 w/ Brad Gerstner · Bg2 Pod
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
In this special episode of the BG2 Podcast, host Brad Gerstner sits down with OpenAI CEO Sam Altman and Microsoft CEO Satya Nadella to discuss the evolution of their historic partnership, massive compute and power infrastructure demands, and the emerging economics of AI agents.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Brad and Bill hold 33.5% of the talking time here. How this is scored →
Brad and Bill as informed peer 6.3 Guest teaching 1.6 Guest disagreement 0.7 Brad and Bill pushing back 0.8
speaking balance: gold is Brad and Bill, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 12:34 Altman fires back on revenue criticisms and short sellers
Sam Altman aggressively shuts down Brad's framing around compute commitments and revenue ratios, offering to find a buyer for Brad's shares and calling out noisy online critics.
Hardest push from Brad and Bill ▶ 10:10 Brad presses Nadella and Altman on AGI jury triggerBrad directly confronts both leaders on their contrasting public statements regarding AGI timing and presses whether an expert jury will be forced to arbitrate their deal terms.
Biggest teaching moment ▶ 18:11 Nadella corrects premise on chip supply versus power constraintsSatya Nadella corrects the popular assumption of chip shortages by explaining that Microsoft's true bottleneck is warm data center shells and power grid availability.
Brad and Bill hold their own ▶ 1:01:01 Brad contrasts search indexing unit economics with inference costsBrad demonstrates deep technical and financial domain expertise by analyzing the stark difference in gross margins between fixed-cost search web indexing and real-time per-prompt LLM compute.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Brad and Bill as informed peer | Guest teaching | Guest disagreement | Brad and Bill pushing back | Why |
|---|---|---|---|---|---|---|
| Welcome & Personal Catch-Up with Sam Altman and Satya Nadella | 4 | 0 | 0 | 0 | Friendly intro banter about Sam's new baby and average company ages. Brad tees up the week's major tech earnings, nuclear power deals, and macro backdrop smoothly. | |
| Equity Ownership & The $130 Billion Non-Profit Foundation | 6 | 2 | 0 | 0 | Brad asks Satya to confirm Microsoft's 27% equity stake and initial investment terms. Satya politely shifts emphasis from the ownership percentage to the creation of the massive non-profit foundation. | |
| Non-Profit Directives: Healthcare, Resilience, and Avoiding Bias | 5 | 0 | 1 | 1 | Brad details the $130B non-profit capitalization and queries Sam on avoiding political bias and focusing on health/resilience. Sam articulates OpenAI's strategy for funding scientific discovery and transition safety. | |
| Azure Exclusivity Terms & Model Distribution Scope | 7 | 2 | 1 | 2 | Brad probes the exclusivity terms on Azure through 2032 and calculates rev share mechanics. Sam clarifies that only stateless APIs remain exclusive to Azure, while products like ChatGPT distribute elsewhere, and Satya humorously notes the accounting is a question for CFO Amy Hood. | |
| Core Mission Alignment Beyond AGI Milestones | 6 | 1 | 2 | 3 | Brad presses on whether divergent AGI timelines will trigger an expert jury panel between the companies. Sam playfully calls out Brad for trying to invent drama before reaffirming the deep partnership. | |
| Microsoft's Perspective on OpenAI's Commercial Execution | 7 | 2 | 3 | 2 | Brad asks how OpenAI can commit $1.4T to compute against $13B in revenue. Sam forcefully pushes back, stating actual revenue is far higher and offering to find a buyer if Brad wants to sell his shares. | |
| Commoditizing Intelligence, Efficiency, and Jevons Paradox | 5 | 3 | 1 | 0 | Sam explains compute elasticity analogous to energy demand curves. Satya introduces Jevons paradox and token efficiency, while Sam lightly corrects the scaling law equation. | |
| Infrastructure Bubbles, Cost Reductions, and Potential Gluts | 6 | 3 | 2 | 1 | Brad cites Jensen Huang saying there is no chance of a compute glut in 2-3 years. Satya clarifies Microsoft's current bottleneck is power and warm shells, not chips, while Sam directly counters Jensen's framing by predicting inevitable supply gluts. | |
| Potential OpenAI IPO Timeline & Retail Investor Access | 6 | 2 | 2 | 2 | Brad brings up Reuters reports on OpenAI's 2026/2027 IPO plans. Sam dismisses the reports as unfounded speculation while agreeing an eventual public listing would provide great access for retail investors. | |
| The Impact of State AI Regulations vs. National Preemption | 7 | 1 | 0 | 0 | Brad provides detailed analysis of federal preemption failures and the Colorado AI Act. Sam confesses he doesn't know how anyone is expected to comply with state patchworks, and Satya notes startups suffer most. | |
| US Competitiveness and International AI Regulatory Risks | 6 | 1 | 1 | 1 | Satya hopes for regulatory harmony across the US and EU, but Sam dismisses the likelihood of European alignment. Brad transitions the conversation to product roadmap horizons for 2026. | |
| Frictionless AI Personal Assistants and Hardware Form Factors | 5 | 0 | 0 | 0 | Sam shares his excitement for multi-day coding tasks and novel AI scientific discoveries before departing. Satya and Brad discuss ambient hardware form factors and micro-steering workflows. | |
| Productization Milestones: GitHub Copilot and M365 Copilot | 6 | 2 | 0 | 0 | Brad asks about the internal 2019 decision to invest $1B into OpenAI. Satya details Microsoft's historic focus on natural language, Bill Gates' initial skepticism, and the breakthrough 1-to-10 moment with early GitHub Copilot. | |
| Monetization Layers: Rev Share, Compute Commitments, and Migration | 7 | 1 | 0 | 0 | Brad breaks down the structural financial pillars of the partnership. Satya explains the long-term strategic benefits of stateless API exclusivity, Azure Foundry app servers, and 7-year royalty-free frontier model IP access. | |
| Accounting for OpenAI Losses and MSFT Shareholder Value | 7 | 2 | 0 | 1 | Brad challenges the market's negative reaction to Microsoft consolidating OpenAI losses on quarterly earnings. Satya breaks down the risk-reward math, noting losses are capped at invested capital while cross-cloud adoption surges. | |
| Managing Infrastructure Constraints and High-Margin Suites | 7 | 3 | 1 | 2 | Brad asks if Microsoft was too conservative on AI capex given persistent power shortages. Satya explains the discipline of building a fungible fleet across GPU generations and prioritizing high-margin Copilot suites over raw bulk training data centers. | |
| Azure's Financial Performance, $400B RPO, and Capacity Planning | 7 | 1 | 0 | 0 | Brad highlights Azure's 39% growth on a $93B run rate and questions the durability of its $400B RPO backlog. Satya explains the 2-year average duration and diversified mix across 1P and 3P workloads. | |
| Compute Expansion, Azure Operating Margins and Scale Economics | 6 | 2 | 0 | 0 | Brad inquires about defending cloud margins against debt-financed competitors like CoreWeave and Oracle. Satya emphasizes that at scale nothing is a commodity, leveraging platform compounding and OpenAI's massive workload. | |
| Circular AI Revenues, Vendor Financing and Balance Sheet Risk | 7 | 2 | 1 | 2 | Brad presses Satya on market worries surrounding circular AI vendor financing and GPU credit swaps. Satya clarifies that Microsoft's $13.5B investment went purely to equity and compute bootstrap, not circular revenue booking. | |
| SaaS Application Disruption and the Emerging Agent Tier | 7 | 2 | 0 | 0 | Brad discusses compressed SaaS valuation multiples and AI disruption threats. Satya explains why low-ARPU, high-usage applications with rich data graphs like M365 and GitHub win as business logic shifts to the agent tier. | |
| Mapping the AI Value Stack: Token vs. Agent Factories | 7 | 3 | 1 | 1 | Brad asks Satya to steelman why value won't entirely accrue to token factories and foundation models. Satya bifurcates the stack into heterogeneous token factories and intelligent agent factories that manage evals and task routing. | |
| Comparing Unit Economics: Search Indexing vs. AI Chat | 7 | 1 | 0 | 0 | Brad contrasts the near-zero marginal cost of amortized search indexing with expensive per-prompt AI chat inference. Satya agrees that consumer chat unit economics are undergoing transition from freemium to subscription and agentic commerce. | |
| Monetization Models in Consumer vs. Enterprise AI Markets | 7 | 1 | 0 | 0 | Brad and Satya explore whether AI drives corporate margin expansion and headcount leverage. Satya illustrates how automated DevOps pipelines allow network engineering teams to manage massive infrastructure scaling without linear headcount growth. | |
| US Re-Industrialization and AI Infrastructure Capex Buildout | 7 | 1 | 0 | 0 | Brad compares tech's $4T capex cycle to 10x the Manhattan Project. Satya details how US hyperscaler investments and domestic manufacturing hubs like Wisconsin re-industrialize the nation while exporting American technology globally. |
Statements from this episode (69)
Altman: Average employee age at OpenAI is around low thirties
“It's not crazy young. Not, not like most Silicon Valley startups. I don't know, maybe low thirties average.”
Gerstner: Earnings from Google, Meta, and Microsoft all report compute shortages
“Google, Meta, Microsoft, Satya, you had your earnings yesterday, you know, and we heard consistently not enough compute.”
Nadella confirms Microsoft owns approximately 27% of OpenAI on diluted basis
“Yeah, it does. But I would say before even our stake in it, Brad, I think what's pretty unique about OpenAI is the fact that as part of OpenAI's process of restructuring, one of the largest nonprofit gets created.”
Nadella: Microsoft did not expect initial $1B OpenAI bet to return 100x
“It's not what we thought, and as I said to somebody, it's not like when we first invested our billion dollars that, oh, this is going to be the hundred bagger that I'm going to be talking about to VCs about, but here we are.”
Altman: OpenAI could not have succeeded without Microsoft and Nadella's conviction
“I don't think, I think this is one of the great tech partnerships ever and without, certainly without Microsoft and particularly Satya's early conviction we would not have been able to do this.”
Altman predicts OpenAI's nonprofit will become the largest nonprofit ever
“And also as Satya said, create what we believe will be the largest nonprofit ever.”
Gerstner: OpenAI non-profit is capitalized with $130 billion in stock
“The nonprofit is already capitalized with a hundred and thirty billion dollars, a hundred and thirty billion dollars. Of open AI stock.”
Gerstner: California AG will not object to OpenAI restructuring
“The California attorney general said they're not going to object to it.”
Gerstner: OpenAI announced directing first $25B of nonprofit to health and security
“You announced that you're going to direct the first twenty five billion to health and AI security and resilience, Sam.”
Altman: OpenAI will keep stateless APIs exclusive to Azure through 2030
“We are keeping what Satya termed once. And I think it's a great phrase of stateless APIs on Azure exclusively through 20, 30 and everything else we're gonna You know, distribute elsewhere.”
Nadella confirms Microsoft-OpenAI revenue share runs until AGI or term end
“Yeah, we have a rev share, and I think as you characterized it, is either going to AGI or till the end of the term.”
Altman: OpenAI would still rely on Microsoft distribution even with superintelligence
“Brad, to say the obvious, if we had super intelligence tomorrow, we would still want Microsoft's help getting this product out into people's hands and we want them like, yeah.”
Altman: OpenAI will become a major AI cloud and consumer device business
“And that not only will Chachapiti keep growing, but we will be able to become one of the important AI clouds that our consumer device business will be a significant and important thing that AI that can automate science will create huge value.”
Nadella: OpenAI has beaten every business plan presented to Microsoft
“And let me just say one thing Brad as both a partner and an investor, there is not been a single business plan that I've seen from open AI that they're put in and not beaten it.”
Altman: OpenAI scaled compute roughly 10x over the past year
“We've scaled our compute probably to 10 X over the past year”
Altman: 10x more compute would yield near 10x revenue for OpenAI
“If we had 10 X more compute, I don't know if we'd have 10 X more revenue, but I don't think it'd be that far.”
Altman: A 100x drop in compute cost would trigger >100x demand surge
“If the price of compute per like unit of intelligence or whatever, however you want to think about it. Fell by a factor of a hundred tomorrow, you would see usage go up by much more than a hundred. And there'd be a lot of things that people would love to do wi…”
Altman: Customers will pay premium rates for breakthrough-level AI models
“Now, on the other hand, as the models get even smarter and you can use these models to cure cancer or discover novel physics or drive a bunch of humanoid robots to construct a space station or whatever crazy thing you want, then maybe there's huge willingness …”
Nadella: Commoditizing intelligence will make AI a key driver of GDP growth
“You keep reducing it, commoditizing, in some sense, intelligence so that it becomes the real driver of GDP growth all around.”
Altman: AI scaling currently equates log of intelligence to log of compute
“Unfortunately, it's something closer to log of intelligence equals log of compute, but we may figure out better scaling laws, and we may figure out how to beat this.”
Nadella: Microsoft's bottleneck is power and warm shells, not chip supply
“Quite frankly, the biggest issue we are now having is not a compute blood, but it's a power, and it's sort of the ability to get the bills done fast enough close to power. So if you can't do that, you may actually have a bunch of chips sitting in inventory tha…”
Altman: An AI compute infrastructure glut is guaranteed to happen
“There will come a glut for sure. And whether that's like in two to three years or five to six, Satya and I can't tell you, but like it's gonna happen at some point, probably several points along the way.”
Altman: Cheap Mass Energy Will Burn Existing Power Contract Signers
“A very cheap form of energy comes online soon at mass scale, and a lot of people are going to be extremely burned with existing contracts they've signed.”
Altman: Cost per unit of intelligence has dropped roughly 40x annually
“This unbelievable reduction in cost per unit of intelligence, let's say it's been averaging like 40 X for a given level per year”
Nadella: AI software improvements are far more exponential than Moore's Law
“You know, we talk about the Moore's law improvement on one end, but the software improvements are much more exponential than that.”
Altman: OpenAI will eventually build low-power devices running GPT-6 locally
“Someday we will make an incredible consumer device that can run a GPT-Five or GPT-Six capable model completely locally at a low power draw.”
Nadella: Cloud AI success requires an efficient token factory and high fleet utilization
“When I look at sort of in the cloud infrastructure business, one of the key things you have to do is have two things. One is an efficient, like in this context, in a very efficient token factory, and then high utilization. That's it. There are two simple thing…”
Altman: OpenAI has no IPO date or formal decision to go public
“No, no, no, we don't have anything that specific. I'm a realist. I assume it will happen someday, but that was I don't know why people write these reports. We don't have like a date in mind. We don't have a decision to do this or anything like that. I just ass…”
Altman: Retail access is the most appealing part of an OpenAI IPO
“Honestly, that, that is probably the single most appealing thing about it to me. That would be really nice.”
Altman: OpenAI Does Not Know How to Comply with Colorado AI Act
“I don't know how we're supposed to comply with that California, sorry, Colorado law. I would love them to tell us you know, we'd like to be able to do it, but that's just, from what I've read of that, that's like I literally don't know what we're supposed to d…”
Altman: 50-State Patchwork of AI Regulations Is a Big Mistake
“I'm very worried about a 50 state patchwork. I think it's a big mistake. I think it's, there's a reason we don't usually do that for these sorts of things. I think it'd be bad.”
Nadella: State AI Regulatory Patchwork Disproportionately Harms Startups
“I mean, I think the fundamental problem of you know, this patchwork approach is quite frankly, I mean, between OpenAI and Microsoft, we'll figure out a way to navigate this, right? I mean we can figure this out. The problem is anyone starting a startup and try…”
Altman expects coding AI to execute multi-day tasks by 2026
“Codex has been a very cool thing to watch this year, and as these go from multi-hour tasks to multi-day tasks, which I expect to happen next year, what people be able to do to create Software at an unprecedented rate.”
Altman predicts AI will make a novel scientific discovery by 2026
“I hope for very small scientific discoveries in 20, 26, but if we can get those very small ones, we'll get bigger ones in future years. That's a really crazy thing to say is that like AI is going to make a novel scientific discovery in 20, 26, even a very smal…”
Altman confirms OpenAI is developing new computing device form factors
“That's one reason I'm very excited that we're doing new form factors of computing devices, because computers were not built for that kind of workflow very well.”
Nadella: Microsoft Azure was OpenAI's first sponsor in 2016
“We were involved even in 2016 when initially OpenAI started, in fact, Azure was even the first sponsor, I think. And then they were doing a lot more reinforcement learning at that time. I remember the Dota two competition, I think, happened on Azure, and then …”
Nadella: Bill Gates was skeptical of Microsoft's initial $1B OpenAI investment
“Obviously you go to the board and say, hey, I have an idea of taking a billion dollars and giving it to this crazy structure, which we don't even kind of understand. What is it? It's a nonprofit, blah, blah, blah. And saying, go for it. There was a debate. Bil…”
Nadella: Seeing Codex in GitHub Copilot triggered Microsoft's 1-to-10 OpenAI investment
“Then seeing the early codecs inside of Copilot, inside of GitHub Copilot, and seeing just the code completions, and seeing it work, that's when I would say we, I felt like I can go from one to 10. Because that was the big call, quite frankly. One was controver…”
Nadella: Kevin Scott's conviction drove Microsoft's big AI bet
“Quite frankly Kevin's conviction that this is worth going after is one of the big things that drove this.”
Gerstner: OpenAI revenue share sends billions annually to Microsoft
“So, you have the value in the rev share, billions per year going, going to Microsoft.”
Gerstner: OpenAI has a $250B Azure compute commitment with Microsoft
“You have the profit you make off the two hundred and fifty billion dollar of the Azure compute commitment from OpenAI.”
Nadella: Microsoft has royalty-free access to OpenAI IP for seven years
“Not only we have the exclusivity of the model in Azure, but we have access to the IP. I mean, having a royalty free, let's even forgetting all the know how and the knowledge side of it, but having royalty free access all the way till seven more years gives us …”
Nadella: Microsoft cannot lose more than $13.5B invested in OpenAI
“If you've invested, let's call it, 13 and a half billion dollars, you can, of course, lose 13 and a half billion dollars. But you can't lose more than 13 and a half billion dollars. At least the last time I checked, that's what you have at risk.”
Nadella: Microsoft does not plan to sell its OpenAI stake
“You could also say, hey, the hundred and thirty five billion dollars that is, you know, today are equity stake, you know, is sort of illiquid, what have you. We don't plan to sell it, so therefore it's got risk associated with it.”
Nadella: Copilot adoption pace is bigger than any Microsoft product in decades
“In fact, one of the things about Microsoft three six five was what was the next big thing after E five? Guess what? We found it in co-pilot. It's bigger than any suite. Like, you know, we talk about penetration and usage and the pace. It's bigger than anything…”
Nadella: Microsoft is currently bringing up NVIDIA GB300 hardware
“We now have GB 300 bringing, you know, that we're bringing up. So you don't want to have ordered a bunch of GB 200 that are getting plugged in only to find the GB two, 300 are in full production.”
Nadella: Microsoft refused OpenAI requests for dedicated multi-gigawatt compute clusters
“Sometimes you may have to say no to some of the demand, including some of the open AI demand, right? Because sometimes, you know, Sam may say, hey, build me a dedicated, you know, big, you know, whatever, multi-gigawatt data center in one location for training…”
Nadella: Copilot Suite Is Microsoft's True High-Margin Business Over Azure
“Some of times our investors are overly fixated on the Azure number. But remember, for me, the high margin business for me is co-pilot. It is security co-pilot. It's GitHub co-pilot. It's the healthcare co-pilot. So we want to make sure we have a balanced way t…”
Nadella: Azure Growth Would Have Hit 41-42% With More Compute
“Absolutely. Absolutely. There's no question. There's no question.”
Nadella: Microsoft Is Supply Constrained Rather Than Demand Constrained
“We're not demand constrained, we're supply constrained. So we are shaping the demand such that it matches the supply in the optimal way with a long-term view.”
Nadella: Microsoft's $400B RPO Backlog Averages Two-Year Duration
“Yeah, that, that four hundred billion it has a very short duration as Amy explained. It's the two year duration on average.”
Nadella: At massive scale, nothing in cloud infrastructure is a commodity
“Everything is a commodity, right? Compute, storage. I remember everybody saying, wow, how can there be a margin? Except at scale, nothing is a commodity.”
Nadella: Running OpenAI workloads accelerates Microsoft's cost curve reduction
“When you have the biggest workload there is running on your cloud, that means not only are we going to learn faster on what it means to operate with scale, that means your cost structure is going to come down faster than anything else. And guess what? That'll …”
Nadella: Microsoft manages compute capex as a single unified outlay
“My capital allocation is for the cloud from whether it is Xbox cloud gaming or Microsoft three, six, five, or for, Azure. It's one capital outlay. And then everything is a meter, as far as I'm concerned, from an MSFT perspective. It's a question of, hey, the b…”
Nadella: Microsoft's $13.5B OpenAI investment was equity, not booked as Azure revenue
“Our investment of let's say that 13 and a half, which was all the training investment, that was not booked as revenue. That is the reason why we have the equity percentage. That's the reason why we have the 27% or a hundred and thirty-five billion. So that was…”
Nadella: AI circular financing will ultimately be tested by end-user demand
“And I think circularity ultimately will be tested by demand. Because all this will work as long as there is demand for the final output of it. And up to now, that has been the case.”
Nadella: Business apps will collapse as AI agents replace business logic
“That's probably where they'll all collapse, right, in the agent era, because if you think about it, right, they are essentially CRUD databases with a bunch of business logic. The business logic is all going to these agents.”
Gerstner: Public software multiples are at 5.2x forward revenue vs. 7x average
“Public software companies are now trading at about 5.2 times forward revenue. So that's below their 10 year average of seven times, despite the markets being at all time highs.”
Nadella: High-ARPU, low-usage SaaS apps face disruption from AI
“And so I think the next generation of SaaS applications will have to sort of, if you are high ARPU, low usage, then you have a little bit of a problem. But if you are, we are the exact opposite. We are low ARPU, high usage.”
Nadella: GitHub replicated over a decade of activity in just one year
“What GitHub did in the first, I don't know, 15 years of its existence or 10 years of its existence, it was basically done in the last year, just because coding is no longer a tool. It's more a substitute for wages somewhere.”
Nadella: Hyperscalers will run heterogeneous hardware fleets across multiple chipmakers
“And it's not, and by the way, it's going to be heterogeneous. Obviously, Jensen's super competitive. Lisa is going to come. You know, Hawk's going to produce things from Broadcom. We will all do our own. So, there's going to be a combination. So, you want to r…”
Nadella: AI gives software significantly higher marginal costs, forcing business model adjustments
“Overall, it is going to be true that there is a real marginal cost to software this time around. It was there in the cloud era too. When we were doing, you know, CD-ROMs, there wasn't much of a marginal cost. You know, with the cloud, there was. And this time …”
Nadella: AI chat unit economics differ fundamentally from search due to per-query GPU costs
“Search was pretty magical in terms of its ad unit and its cost economics, because there was the index, which was a fixed cost that you could then amortize in a much more efficient way. Whereas this one, you know, each chat to your point, you have to burn a lot…”
Nadella: AI agents will replace seats in enterprise software monetization
“Whereas in the enterprise, one is it's not winner take all. And two, it is going to be a lot more friendly for agentic interaction. So it's not like, for example, the per seat versus consumption. The reality is agents are the new seats.”
Nadella: Enterprise AI monetization is clearer than murky consumer AI
“And so you can think of it as, ah, the enterprise monetization is much clearer. The consumer monetization, I think, is a little more murky.”