The AI Race Sparks Unprecedented Demand for Gas Turbines
- The rapid growth of AI and data centers is causing a significant increase in demand for electricity, leading to a global shortage of gas turbines needed for new power generation capacity.
- Despite the urgent need, many gas turbine manufacturers are cautious about dramatically increasing production due to past experiences where oversupply led to market crashes, and the long lead times for building these precision machines.
- The imbalance between supply and demand is driving up costs and extending wait times for new gas-fired power plants globally, highlighting a critical challenge in meeting the energy requirements of the expanding AI industry.
There are not enough gas turbines being manufactured in the world right now. The message has been flashing on and off in the media for months now as the AI race drives demand for electricity sharply higher and, with it, demand for new generation capacity. Despite the spike, however, not all gas turbine makers are in a rush to go all in on the AI hype. They’ve been burnt before.
In September, the U.S. Energy Information Administration reported that there were 114 GW of new natural gas generation capacity under construction and in pre-construction in the country. This amount represented a more than twofold increase over last year and came in response to the mad dash to build data centers to handle the growing demand for artificial intelligence. But there is a problem: the supply of gas turbines for those power plants is falling far behind demand.
Bloomberg was one of the publications warning about a turbine shortage earlier this year. The outlet reported that some data center operators, notably Stargate, were deploying single-cycle gas turbines to secure generation for their facilities in the absence of enough combined-cycle turbines being manufactured fast enough.
“There is a high level of urgency in the industry to get power fast,” a senior executive from Crusoe, the company that is building the first Stargate project, told Bloomberg. “We have tried to be creative about the energy component of data centers.”
However creative one might try to get, there is a limited pool of options. And turbines take a long time to be built because they are precision machines, and building precision machines cannot be rushed. What compounds the supply problem is the fact that data center operators are not the only ones in need of more gas turbines. Gas, for all the bad rap it has been getting from climate activists, is what the developing world is replacing coal with, so that growth in demand for turbines is a global trend.
Transition advocacy outlet the Institute for Energy Economics and Financial Analysis recently reported that the imbalance between supply and demand for gas turbines was driving costs for new gas-fired power plants higher in Vietnam and the Philippines and extending wait times to seven or even eight years. The outlet recommended “the rapid expansion of low-cost renewables and storage”, which “could ultimately limit the long-term role for gas and liquefied natural gas (LNG).”
The problem is, they can’t. If they could, data center operators would be causing a shortage of solar panels and inverters. Instead, the shortage is looming in gas turbines because there is one fundamental difference between wind/solar and gas: the latter generates dispatchable electricity, and that is the kind that data center operators need.
In such a context, turbine makers should be rushing to boost their production capacity, or so some would assume. Yet that is not what they are doing—at least not all of them. “How do we manage what’s real, what’s not?” the chief executive of Mitsubishi Power Americas, Bill Newsom, said recently, as quoted by the Wall Street Journal, expressing concern that all the bombastic forecasts about AI and the related electricity demand surge may turn out to be wrong.
It has happened before, the Wall Street Journal noted in its report. About 20 years ago, gas turbine makers rushed to boost output in expectations that the internet would drive a surge in electricity demand. That surge never materialized, sinking some manufacturers and leaving others with surplus capacity. What makes things even more uncertain is the possibility that it would only take one manufacturer to boost capacity to tank prices for everyone, even with all the bombastic demand forecasts.
“If nobody increases their output, prices will stay really high. If one person raises their output, it could crash the prices for everybody,” Enverus’s director of energy transition research, Ryan Luther, told the Wall Street Journal.
Yet this may be too cautious an outlook. According to Bloomberg, some $400 billion worth of new gas-fired generation capacity could face delays or even cancellation because of the shortage in gas turbines. This does not, on the face of it, sound like a situation where one manufacturer raising capacity could tank prices for all the rest of them. Indeed, one turbine major is trying to ramp up capacity, and it has not tanked prices—because the ramp-up will take a while.
The major trying to boost capacity is Mitsubishi Heavy Industries, one of the three companies dominating the global gas turbine market alongside Siemens and GE Vernova. “It’s not so easy to ramp up,” Joern Schmuecker, a vice president at Siemens Energy told Bloomberg in an interview earlier this month. “The whole supply chain is struggling to also keep pace.”
Challenging as this may be for Big Tech and adjacent industries, the time factor could provide the ultimate test of how serious the world is about AI and whether reality would live up to the hype or not.
By Irina Slav for Oilprice.com
No comments:
Post a Comment