OpenAI Expansion Targets AI Escape Velocity

Brad and Bill on the BG2 Podcast describe the AI CapEx bubble. The explosive growth in AI investment—especially in data centers, chips, and model training—shows potential signs of historical overreach. They increase the odds of overbuilding and reduce the quality of revenues.

Gurley identifies round-tripping deals like the original deal between Microsoft and OpenAI, where investments are effectively recycled as cloud credits, creating synthetic revenue. These accounting and financing practices were used excessively by Enron and WorldCom (famous scammy businesses).

Nvidia’s balance sheet is strong but it hides systemic risk in deals such as CoreWeave obligating Nvidia to purchase unused capacity—masking true demand.

AI infrastructure spending could exceed $3 trillion by 2030, with major hyperscalers (Microsoft, Google, Amazon, Meta) deploying over 60 gigawatts of data center capacity. Valuation multiples remain historically modest, though signs of overprovisioning loom.

OpenAI is trying to achieve [AI] escape velocity with aggressive chip partnerships (Broadcom, AMD, etc.), effectively forcing competitors to follow.

AI escape velocity refers to a tipping point in artificial intelligence development where progress accelerates exponentially. There is AI financial escape velocity where AI becomes profitable with high growth and high margins.

AI systems becoming capable of improving themselves faster than humans can will lead to rapid breakthroughs in capabilities. OpenAI CEO Sam Altman has described this as a fast AI takeoff. It could manifest as AI solving complex problems (e.g., curing diseases, optimizing energy) at an accelerating rate, transforming economies and society in unpredictable ways, but also raising risks like job displacement or existential threats if not managed carefully.

There is a computing arms race where hyperscalers plan to spend over $1.5 trillion on data center spending over the next 4-5 years.