OpenAI just doubled the price of intelligence. GPT-5.5 Pro charges $180 per million output tokens — triple what competitors charge and 10x what we saw two years ago. But here's the thing: the market isn't blinking. Usage is up. Anthropic just hit a $1 trillion valuation on Forge Global, surpassing OpenAI's $880B. The economics are shifting, and fast.
The Compute Floor Is Rising
Texas Instruments posted its best day since 2000 — a 19% jump driven by "high demand for analog chips used in AI data centers." That's not a startup pivot. That's legacy silicon eating the AI boom. When the underlying substrate (compute, memory, networking) is in shortage, the price of the thing built on top (tokens) has only one direction to go.
Intel reported Q1 revenue up 7% YoY to $13.58 billion, beating estimates by over a billion. After years of stagnation, Intel is back because AI doesn't run on vibes — it runs on silicon, and someone has to make the chips.
The token economy isn't about tokens. It's about who controls the meter's dial.
What $30 Per Million Tokens Actually Means
Let's do the math. A typical GPT-5.5 Pro response runs 500-2000 tokens. That means each reply costs between 1.5¢ and 6¢. For a coding agent making hundreds of calls per day? That's dollars per developer, per day. Multiply by 200 million knowledge workers and you're talking real money.
Microsoft just announced its first voluntary retirement program — for employees whose age plus years of service equals 70+. In a 50-year company. The workforce is aging out, and AI is filling the gap. Copilot's agentic features in Word, Excel, and PowerPoint are now generally available for 365 Copilot and Premium subscribers. The office suite is becoming the operating system, and Microsoft's collecting rent on both.
The Infrastructure Play
Google just launched TPU 8t (training) and 8i (inference), generally available later this year. 75% of new code inside Google is now AI-generated and human-reviewed, up from 50% last fall. The cycle time for building is compressing, which means the demand for inference compute is compounding.
Meanwhile, SpaceX's S-1 reveals a $28.5 trillion total addressable market, with $26.5 trillion from AI. They're planning to manufacture their own GPUs. Not rent, not buy — make. That's the next level of vertical integration, and it tells you everything about how scarcity is being priced.
The Signals
The token economy is forming along clear lines:
- Pricing power: OpenAI doubled rates and demand keeps growing. Price elasticity is not the constraint.
- Infrastructure scarcity: TI up 19%, Intel up 7%, Google launching TPUs. The substrate matters.
- Workforce recalibration: Microsoft retiring employees. Meta cutting 10% of staff to offset AI spending. The human capital equation is flipping.
The old economy ran on interest rates and labor. The token economy runs on inference latency and context windows. The rules aren't the same, but the game is just starting.