The huge cost of AI

Companies like Jensen Huang's Nvidia are busy investing in supercomputers to power the AI of the future. Photo: Nvidia Blog .

A new study published by Epoch AI – a San Francisco-based research institute – shows that the energy demand of supercomputers is increasing at a dizzying rate, in parallel with the development of artificial intelligence.

According to Epoch AI, if the annual doubling of electricity demand continues through the end of the decade, the world's top supercomputers in 2030 could need up to 9 gigawatts (GW) – the equivalent of power to power 7-9 million households.

Currently, the world's most powerful supercomputer consumes about 300 megawatts (MW), equivalent to the amount of electricity used by 250,000 households. Compared to this figure, the predicted energy consumption in the future is "huge", according to the research team.

One of the main reasons why AI supercomputers will become more power-hungry is because they are getting bigger. Epoch AI estimates that if growth trends continue, a top AI supercomputer in 2030 could require as many as 2 million AI chips, costing up to $200 billion to build.

The number of operations per second (FLOP/s) of today's top systems is doubling every nine months. Source: Epoch AI

For comparison, the Colossus system - a supercomputer built by Elon Musk's xAI in 214 days - is currently one of the largest systems in the world with 200,000 chips and a total investment of about 7 billion USD .

Big tech companies are racing to build computing infrastructure to support increasingly powerful AI models. OpenAI announced earlier this year its Stargate project, worth more than $500 billion over the next four years, to develop a key AI supercomputer.

According to Epoch AI, supercomputers are no longer simply research tools as they once were. They have become “industrial machines” – providing real economic value, the key infrastructure of the AI era.

Even politics have gotten in on the act. Earlier this month, President Donald Trump took to the social media site Truth Social to tout Nvidia’s $500 billion investment in AI supercomputers in the US, calling it “big and exciting news” and emphasizing the promise of a “Golden Age of America.”

Epoch AI 's research is based on data covering about 10% of global AI chip production in 2023-2024, along with 15% of major companies' chip inventories as of early 2025. The team of experts asserts that while energy efficiency is improving, the current increase is still not enough to offset the overall growth rate in electricity demand.

This is also the reason why many technology giants such as Microsoft, Google and data center operators are considering alternative solutions such as nuclear power to provide stable, long-term energy.

If current trends continue, not only will AI grow more powerful, but the size, cost, and energy demands of supercomputing systems will also increase exponentially.

Source: https://znews.vn/chi-phi-khong-lo-cua-ai-post1548789.html