Altman targets $20B per gigawatt for AI infrastructure
OpenAI chief executive Sam Altman says the five-year cost of building and operating one gigawatt of AI infrastructure should fall from about $70 billion to roughly $20 billion. The target covers chips, data centres, power systems and electricity; it is not a confirmed cost reduction.
Altman says cheaper energy and lower supply-chain margins are needed, while warning that power availability remains a constraint. Semiconductor analyst Dylan Patel estimates current construction costs exceed $60 billion per gigawatt, up from about $40 billion in 2024, and says removing supplier margins could bring costs closer to $20 billion.
The issue matters to OpenAI, whose infrastructure commitments exceed $1.4 trillion. Its planned Project Camellia campus in Georgia is valued at $20 billion and has contracted capacity of 3.2 gigawatts, with a phased rollout planned for 2028–2032. Lower AI infrastructure costs could support expansion, but persistent energy and capital costs would put pressure on the industry’s growth plans.
Neutral
The article has no direct cryptocurrency catalyst, so the likely effect on crypto trading and market stability is neutral. Altman’s cost target is an industry ambition, not a completed reduction, and the article does not report changes to crypto regulation, token economics, or digital-asset demand.
In the short term, the remarks could influence sentiment around AI-linked technology stocks and chip suppliers, but any effect on crypto would likely be indirect and limited. Traders may watch for moves in AI-related equities, data-centre investment and electricity prices, as well as broader risk appetite. There is no clear basis in this report alone to expect a sustained shift in Bitcoin or other major token prices.
Over the longer term, cheaper data-centre capacity and improved access to power could benefit crypto miners and other operators that use energy-intensive computing, if they gain access to the infrastructure. Conversely, competition for electricity could keep power costs high for miners. Similar periods of strong AI infrastructure investment have sometimes encouraged market interest in both AI-related assets and high-performance computing, but spillovers to crypto have tended to depend on broader liquidity and risk sentiment. These effects are uncertain and do not change the neutral overall assessment.