The frontier of AI capital concentration is narrowing while the infrastructure beneath it fractures. Amazon's shutdown of Mechanical Turk eliminates one of the last remaining pools of on-demand human labor that powered early model training and data annotation, a move that coincides with the consolidation of compute resources and training data in the hands of a few well-capitalized players. OpenAI and Anthropic face the paradox of unsustainable burn rates that make going public dangerous yet falling behind competitively more dangerous still, forcing them deeper into dependency on venture capital and strategic investors rather than market discipline. That capital concentration flows upward through a single veteran investor who now sits at the center of the global AI boom, raising questions about concentration of control that regulators are beginning to notice. Britain and Australia are both grappling with the infrastructure question from opposite angles: London wants AI sovereignty but risks remaining a US outpost, while Australia debates whether to build more data centers at all, each recognizing that compute and data location determine who owns the frontier. Meanwhile, financial regulators are sounding alarms about an arms race in AI adoption that they lack the power to oversee, and traditional media like ABC is moving to integrate AI into newsrooms following the historical pattern of early tech adoption. The real story is not whether AI will transform industries but whether the power to build and train frontier models will remain concentrated enough to be controlled by a handful of entities and their backers.
Sloane Duvall