Back to Dwarkesh Patel

Dylan Patel – Two labs will soon control most of the world's workforce

Dwarkesh PatelAugust 25, 20261h 16m
Topics68
Current State of AI Lab Compute and Revenue0:00AI Labs Taking Increasing Share of Compute0:30Transition from Venture Funding to Profitability1:30Dramatic Improvement in Margins2:03The Compute Investment Flywheel3:03Centralization of Compute at Frontier Labs3:34New Entrants Building Compute4:31Timeline for Labs Controlling Most Compute5:00Exponential Growth Projections5:30Efficiency Improvements Offsetting Compute Growth6:03Labs Capturing Higher-Performance Compute6:33Physical Constraints on Compute Expansion7:01Value Discrepancy and Supply Chain Bottlenecks9:01Supply Chain Reaction Lag10:33Capital Constraints vs. Lab Revenue12:00100 Gigawatts for Labs by 202813:00Questions About Market Disruption13:34Anyone Can Make Money on Compute14:31Labs Must Outbid Everyone Else15:00Labs Must Pay Premium Prices16:03Regulatory Impact Scenarios17:02AGI Revenue Projections17:38Value Distribution Across the Stack19:32Shifting Value Capture Over Time20:05The New Power Structure23:02Grok Bot Workflow for Candidate Screening25:16Revenue Per Gigawatt Projections for 202725:33Compute Pricing and Supply Chain Dynamics26:01AI Progress Rate and Model Capability Leap27:02Regulatory Constraints on Model Deployment28:00Compute Allocation Strategy as Public Companies29:01Training vs Inference Compute Allocation30:02Internal vs External Inference Prioritization31:04Evidence of Shifting Compute Allocation32:01Global Compute Growth Projections33:32US vs China Compute Deployment34:02China's Domestic Semiconductor Development Timeline35:31China's Future Compute Addition Capacity37:01Export Controls and Geopolitical AI Competition38:00Chinese AI Lab Compute vs US Labs40:00Compute Budget Breakdown40:35Training Limitations and Future Shifts42:03Infrastructure Investment Requirements42:34CapEx Funding Sources and Economic Impact44:00Sovereign Debt Crisis Implications47:00US Tax Base and Data Center Revenue50:10Debt Servicing Costs and Interest Rate Sensitivity50:32Global Debt Vulnerabilities52:01Debt-Heavy Industries at Risk52:34Amazon's Debt Strategy and Returns53:02Impact on Equity Valuations58:03Second Volcker Shock Scenario59:04Post-Singularity Interest Rate Projections59:34Opportunity Cost of Capital1:00:30AI-Pilled Market Multiples1:01:33Centralization of Compute and Labor1:03:00Slow Takeoff Factors1:03:32Recursive Self-Improvement Timeline1:04:32Government Intervention Scenarios1:05:00Centralization of AI Labor Supply1:06:31Concentration of Influence1:09:04Forces Driving Centralization1:10:06Decentralization Challenges1:11:31Value Capture Distribution1:14:32Compute Reallocation Logic1:15:27Power Concentration in AI1:15:33Economic Rationale for Internal Compute Allocation1:16:04Closing Remarks1:16:35
In a Nutshell

Two frontier labs—OpenAI and Anthropic—are absorbing 40-50% of new compute deployed each year and will control most usable global AI compute within two years as their gigawatt-scale clusters triple annually while total world compute merely doubles. Their revenue per megawatt has already surged from negative margins to $30-50M, enabling self-funded expansion toward 100 combined gigawatts by 2028, while supply-chain bottlenecks (EUV mirrors, turbines, power) and regulatory constraints on model releases create the only real limits to their dominance. This concentration means virtually all AI labor and most of the world’s high-value cognitive work will soon reside inside two organizations, fundamentally restructuring power, capital allocation, and economic returns worldwide.

AI-Generated Notes

These notes were generated by AI and may contain inaccuracies.

The conversation opens with Dylan Patel of SemiAnalysis discussing how the world economy increasingly depends on lab economics and compute markets. Most GDP growth in America last year came from AI infrastructure.

About a third of compute coming online this year is for the labs (OpenAI and Anthropic), though it may be built by others and rented to them. Compute numbers are ballooning from over a trillion dollars in CapEx this year to more than $2 trillion by 2028. The labs are taking an increasing percentage of this, moving from spending tens of billions annually to hundreds of billions, with forecasts of spending trillions by the end of the decade.

This requires reshaping the labs' economics. Until recently, these were companies that mostly lost money. Anthropic started turning a profit in Q2, and OpenAI is believed to potentially turn a profit in Q3 with the rise of Codex and 5.6. A year ago, all their money was venture-funded losses. Even at the beginning of this year, it remained venture-funded losses. They've now turned the corner and are starting to profit, though new capital continues coming in to accelerate growth. More of their business is being funded from their own revenue rather than capital injections.

Over the last year and a half, their margins have skyrocketed. The base cost of compute tends to be around $10-15 million per megawatt. Previously, serving models like GPT-4 on Nvidia Hopper GPUs generated negative gross margin for OpenAI. Now, when OpenAI serves GPT-5.6 or Anthropic serves Opus 5 or Fable 5, their revenue generation has passed well beyond the incremental $10-15 million per megawatt. In Anthropic's case, revenue has gone as high as $50 million per megawatt.

Sign in to read the full notes

Get access to AI-generated notes, topic timestamps, and more.