- Gartner study suggests AI data center power requirements will grow 26% by 2026
- This is a 13% increase compared to a previous forecast that limited growth to 500TWh.
- AI data centers currently account for 31% of total data center energy consumption, but are expected to surpass the energy needs of conventional servers by 2027.
Demand for AI chips has skyrocketed in recent years, and all major industry players have invested in infrastructure, training and inference hardware to build their own data centers and computing clouds.
Better, faster chips were supposed to be the key to unlocking both Artificial General Intelligence (AGI) and AI-infused efficiency gains as the world shifts its focus from AI agents to AI operators.
The bottleneck that many anticipated but was arguably minimized is now back in the spotlight: power constraints may limit future data center growth globally.
Is it not a chip problem, but an energy enigma for 2030?
A recent report from Gartner indicates that AI servers may not have a chip supply problem, but they do have power limitations that could decisively determine future data center expansion, stopping it completely by 2030 if not addressed.
Gartner estimates that while current data center power needs are capped at 132 GW, they could reach 290 GW by 2030, indicating that power constraints will undoubtedly dominate future AI data center planning.
“Rising demand for compute-intensive AI workloads is driving unprecedented growth in data center power, while AI capacity is now limited by power availability, making data center power security the new battleground for scaling and protecting margins in the global AI race,” said Linglan Wang, director analyst at Gartner.
The current estimate makes even the most extreme case described by electrical infrastructure provider Schneider Electric look tame.
That’s why Nvidia CEO Jensen Huang has already started pointing to power efficiency as the reason its chips are superior to the competition.
In a recent interview with BloombergHuang said both data centers and enterprise consumers would want the most “tokens per watt” to get maximum value in a power-constrained future.
Scaling up power generation or upgrading networks is arguably a more complex or time-consuming task than simply building an AI data center; Goldman Sachs estimates that up to $720 billion in network spending could be needed by the end of the decade to account for the additional burden that AI data centers will bring to the table.
It remains to be seen if this plays out exactly as Gartner projected; However, with all industry players indicating that they intend to increase spending on AI infrastructure, the projection that sees current energy needs (565 TWh) doubling (1,200 TWh) by 2030 is a very possible scenario, and the industry’s focus could shift over time to delivering both energy and efficiency versus raw computing over time to account for the shift.
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds.
