When one of the smartest money managers alive says AI extinction risk is real and proposes taxing tokens to save jobs, you should probably pay attention.
The Summary
- Greg Jensen, Bridgewater's CIO and early backer of OpenAI and Anthropic, says AI poses genuine human extinction risk and recently proposed a "token tax" in the New York Times to offset AI-driven job displacement
- Jensen draws explicit parallels between current AI discourse and the pre-Covid months of early 2020, when warnings were visible but widely dismissed
- Bridgewater is actively deploying AI internally while simultaneously calling for stronger regulatory frameworks, a rare position for institutional finance
The Signal
Greg Jensen isn't some tech blogger with a Substack and an opinion. He runs money at the world's largest hedge fund, wrote checks to OpenAI and Anthropic before most people knew what a transformer was, and now leads Bridgewater's AI strategy. When he says extinction risk is real, he's not catastrophizing. He's calculating.
The token tax proposal is the interesting part. Jensen wants to tax AI model inference at the token level, using revenue to fund transition support for displaced workers. It's a carbon tax for compute. The logic: if AI eliminates jobs faster than the economy can create new ones, you need a mechanism to redistribute the productivity gains. Tax the thing doing the displacing, fund the people being displaced.
"Jensen draws explicit parallels between current AI discourse and the pre-Covid months of early 2020, when warnings were visible but widely dismissed."
This isn't about slowing AI down. Jensen clearly believes in the technology, Bridgewater is using it, and he's invested in the companies building it. But he's also watching model capabilities accelerate and seeing the same pattern he saw in January 2020: smart people raising alarms, institutions moving slowly, public discourse stuck in denial or hype cycles.
The Covid comparison matters because it's about preparedness, not prediction. In early 2020, the data was there. The exponential growth curves were visible. The warnings were clear. What failed wasn't foresight, it was action. Jensen is saying we're in that window again, but for labor displacement and potentially worse outcomes.
Key points from Bridgewater's AI position:
- Active internal deployment of AI tools across investment research and operations
- Simultaneous advocacy for stronger regulatory frameworks, unusual for institutional finance
- Focus on economic transition mechanisms, not just abstract safety concerns
The regulatory call is notable because hedge funds don't typically beg for oversight. But Jensen sees asymmetric risk: if AI development outpaces governance and employment infrastructure, you get social instability that's bad for markets, bad for democracy, and bad for the future his portfolio depends on. He's pricing in second-order effects.
The Implication
When the people building and funding AI start sounding like the people warning about AI, the gap between accelerationism and doomerism is closing. Jensen's position suggests a new emerging consensus: AI capabilities are real, economic displacement is coming faster than expected, and current institutions are not ready.
The token tax won't happen soon, but the idea that we need economic mechanisms tied directly to AI inference volume is now in the conversation. Watch for more proposals like this from people who actually understand both the technology and the money. If you're building in this space, plan for a world where compute gets taxed, model deployment gets regulated, and job displacement becomes a political crisis by 2028.