Industry

Data center electricity use could quadruple by 2035

A new report from BloombergNEF predicts that U.S. data centers will consume four times more electricity by 2035, accounting for one-fifth of the nation's total power generation. The surge is driven by AI compute demand, with nearly half of new capacity going to training and inference. The forecast is 83% higher than BloombergNEF's previous estimate, reflecting rapid data center development and strained electrical grids.

Neura News

Neura News

Neura Market Editorial

July 21, 20263 min read

Originally reported by techcrunch.com

Data center electricity use could quadruple by 2035

Data centers in the United States are on track to consume four times more electricity by 2035 than they do today, according to a new report from BloombergNEF. That would mean data centers use about one-fifth of all electricity generated in the country.

The report attributes the surge to a boom in AI computing. BloombergNEF predicts that data center capacity will reach nearly 200 gigawatts over the next decade. Roughly half of that capacity will go toward training and running AI models, with most of that activity staying in the U.S. By 2033, the country is expected to host 64% of the world's AI chips measured by power demand.

If anything, those numbers might be conservative based on earlier forecasts.

BloombergNEF's latest estimate for 2035 electricity demand is 83% higher than what the consultancy projected in December. Other organizations have also raised their forecasts. EPRI, a nonprofit focused on the electrical industry, has more than doubled its 2024 estimate. S&P's forecast rose by more than a third between October and April. These revisions reflect the rapid pace of data center construction across the country.

Strained grids face growing demand

Over the next decade, BloombergNEF expects most new data centers to connect to electrical grids that are already under pressure. The PJM Interconnection, which covers an area from Virginia to Illinois, will see 34% of its electricity go to data centers. ERCOT, which serves most of Texas, will need to devote 22% of its generating capacity to data centers.

PJM already hosts a large number of the nation's data centers and has struggled to handle connection requests from both large power generators and large consumers. The grid manager paused applications for new sources to connect to the grid for four years, leaving it in a difficult position as demand kept rising.

The #1 Newsletter in AI

Stay ahead of the AI curve

The most important updates, news, and content — delivered weekly.

No spam. Unsubscribe anytime.

Although PJM reopened the queue for new generating sources in April, the situation has become so serious that American Electric Power, a major utility, has threatened to pull out of the interconnection. The imbalance between supply and demand has pushed electricity prices up 76% over the past year.

Despite the congestion, data centers still want to connect to PJM. They represented 38% of charges in the grid manager's most recent capacity auction.

Global growth in data center electricity

While the U.S. will claim the majority of AI compute, data centers will continue to expand in other parts of the world. By 2033, if AI adoption follows an aggressive trajectory, data centers will create 1,935 terawatt-hours of new electricity demand globally. That is nearly as much electricity as India uses in a year.

The report underscores the tension between the rapid expansion of AI infrastructure and the limitations of existing power grids. As data center development accelerates, utilities and grid operators face mounting pressure to keep up.

Related on Neura Market

More from Neura News

AI Models

Google Unveils Gemini 3.6 Flash, 3.5 Flash-Lite, and Cyber Model

Google has released three new Gemini models: 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. The 3.6 Flash model offers improved coding and knowledge work with 17% fewer output tokens and lower costs. The 3.5 Flash-Lite is the fastest in the series at 350 tokens per second, designed for high-throughput agentic tasks. The 3.5 Flash Cyber model, available only to governments and trusted partners via CodeMender, focuses on finding and fixing cybersecurity vulnerabilities. Google also noted that Gemini 3.5 Pro is being tested with partners and that pre-training for Gemini 4 has begun.

Jul 21·5 min read
AI Models

Alibaba Qwen-Image-3.0 renders infographics and tiny text in one pass

Alibaba's Qwen team released Qwen-Image-3.0, an image generator designed for practical applications like newspaper layouts and complex infographics. The model processes prompts of up to 4,500 tokens and can render legible text as small as ten pixels, mathematical formulas, and twelve languages in a single pass. It is currently available through invite-only API access, with plans to integrate it into first-party apps like Qwen Chat soon.

Jul 21·4 min read
AI Models

Google Unveils Gemini 3.6 Flash, 3.5 Flash-Lite, and Cyber Model

Google DeepMind has introduced three new Gemini models: 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. The 3.6 Flash model offers improved coding and multimodal performance with 17% fewer output tokens and lower cost. The 3.5 Flash-Lite is the fastest in its series at 350 output tokens per second, designed for high-throughput agentic tasks. The 3.5 Flash Cyber, fine-tuned for cybersecurity, will be available exclusively to governments and trusted partners via the CodeMender agent.

Jul 21·6 min read