Sam Altman told Business Insider he fears a small group of companies or countries controlling advanced AI would be 'very, very bad.' The irony is acute. OpenAI, Microsoft, Google and Amazon already dominate AI infrastructure and model access. Microsoft’s $13B investment in OpenAI, Google’s $2.3B in Anthropic, and AWS’s bedrock of cloud AI tools prove concentration is accelerating. NVIDIA’s 80% GPU market share for AI workloads cements the hardware chokehold.

Altman’s warning reads like a hostage negotiating with his own shadow. The AI stack is vertical integrating at speed. Microsoft’s Azure AI, Google’s Vertex, and AWS’s SageMaker control the rails while OpenAI, Anthropic, and Meta set the model agenda. This isn’t hypothetical. The barrier to entry now isn’t just talent or data but $100M+ compute budgets only hyperscalers can underwrite. The second-order effect is regulatory capture. As AI becomes a national priority, the incumbents will write the rules to preserve their lead.

Winners: Microsoft, Google, Amazon, NVIDIA, OpenAI. Losers: Startups without cloud partnerships, sovereign nations without domestic AI champions, and consumers facing homogenized innovation. Europe’s AI Act already exempts open-source models but the compute to run them is still gated by US giants.

Watch for Altman’s next move to square this circle. Expect OpenAI to push for ‘democratized’ access via APIs that still run on Azure. Regulators will target NVIDIA’s GPU monopoly first but the real battle is cloud lock-in. If Meta open-sources Llama 3 but only AWS can run it at scale the oligopoly tightens.