AI training and inference are extremely compute-intensive and therefore extremely energy-intensive. The two main bottlenecks for AI model development today are compute and energy; data centers are 100% powered by an energy source, so compute infrastructure, energy, and data centers are critical AI demand centers.
The AI boom is not a pure dot-com-style bubble because many 1999 companies were not technology-intensive and collapsed when multiples corrected, while real AI infrastructure companies can endure. Nvidia sells the compute infrastructure used for AI training and inference—the picks and shovels of the AI gold rush—and is already one of the world's three most valuable companies. Even if AI spending adjusts, Nvidia is proving to be a durable winner.
AI training and inference are extremely compute-intensive and therefore extremely energy-intensive. The two main bottlenecks for AI model development today are compute and energy; data centers are 100% powered by an energy source, so compute infrastructure, energy, and data centers are critical AI demand centers.
Capital is pouring into AI from cash-rich companies, venture capital, and retail investors. Google said it prefers to invest heavily in AI rather than fall behind, and AI leaders have moved from hundreds of millions to trillion-dollar valuations, showing that the AI sector remains supported by broad investment flows.