The challenge facing the artificial intelligence industry is no longer limited to developing models and increasing their capabilities, but extends to securing the energy needed to run huge data centers, so much so that companies are looking for unconventional sources of electricity generation, including obsolete aircraft engines.
According to an analysis published by Reuters, energy companies and data center developers are buying used gas turbines, restoring equipment that has been put out of use, and even converting aircraft engines to electric generators, in an attempt to overcome long waiting times for new turbines, which may extend for years.
The average wait for a new gas turbine in the US is estimated to be six years, while global orders exceed 100 gigawatts, compared to annual production capacity of only 60 to 70 gigawatts.
This imbalance between supply and demand has led to an increase in the importance of the used equipment market, with prices for good used turbines ranging from $1.5 to $1.7 million per megawatt, compared to about $2 to $2.5 million for new units. In a move that reflects the extent of this trend, Crusoe ordered about 13 turbines based on aircraft engines from GE Aerospace, with a total capacity of 650 megawatts.
But relying on these solutions is not without its challenges; Reconditioned equipment may encounter issues related to maintenance history and remaining operational life, while questions remain about the suitability of converted aircraft engines for continuous operation as sources of power generation.
This reveals a deeper change in the artificial intelligence race, as the competition is no longer only related to the power of the processors and the efficiency of the models, but the ability to provide the necessary electricity and infrastructure has become a decisive factor in determining the future of this industry.