Part 3/10:
By 2028, the scale will reach 10 gigawatts, surpassing power output of most US states and costing hundreds of billions.
By 2030, these clusters could consume 100 GW of power with 100 million GPUs, approaching a trillion-dollar valuation.
This rapid scaling signals not only technological but also significant economic commitments—billions, if not trillions, of dollars—to fuel the development of powerful AI systems capable of achieving human-level and beyond intelligence.