KM4BI KM4BI

AMD pushes AI forward, Google posts strong Q2 results, and OpenAI plans $750B in cloud spend

TBPN connects AMD’s AI event, Google’s earnings and OpenAI’s spending plans to the widening race for AI infrastructure.

Broadcasting from AMD Advancing AI in San Francisco, TBPN frames the week around one central theme: AI infrastructure is becoming the strategic battlefield. AMD introduces Helios, a rack-scale AI system built around the MI450 accelerator, and announces a major Anthropic partnership that could deploy up to 2 gigawatts of Instinct MI450s. The point is not only chip supply; AMD is trying to show that its hardware, software stack and direct lab partnerships can become credible alternatives to Nvidia.

Google’s results add another dimension. Alphabet reports strong year-over-year growth, with cloud, search and advertising all contributing, while investors focus on the cash-flow impact of massive data-center spending. The hosts also unpack the debate around TPU system revenue, noting that Google’s internal accelerators are becoming more meaningful without implying that Google Cloud has turned into a pure TPU sales business.

OpenAI represents the demand side of the same race. TBPN cites the company’s plan to raise projected cloud-compute spending to roughly $750 billion through 2030, alongside a new data-center project in Georgia. In the context of coding agents and continued belief in scaling, the number reads as a bet that capacity will remain a critical constraint.

The episode then broadens to DeepSeek’s reported inference economics and AGI focus, Cognition’s acquisition of the Poke team, the rise of search funds, Meta’s AI-optimism campaign, Paramount’s interest in microdramas, and hints of OpenAI releases around Codex, voice, desktop agents and health. The common signal is that AI competition is moving beyond model quality alone toward control of compute, software integration and product distribution.

Source