Why AI, for all its advances, still relies on diesel generators

AI infrastructure is creating fresh growth opportunities for traditional manufacturers.

In the past, diesel generators were mostly deployed in factories and telecommunications base stations. Today, the massive expansion of AI data centers, which demand far higher power‑supply reliability, has thrust this decades‑old piece of conventional equipment back into the spotlight.

Generac, founded in the late 1950s, was long best‑known for its residential backup generators. Now this established US manufacturer is devoting an increasing share of its production capacity to data‑center solutions.

In Q2 this year, Generac’s commercial and industrial segment hit USD 556 million in sales, a 29 % year‑on‑year rise. By the end of July, its backlog of data‑center‑related orders stood at roughly USD 1.6 billion. Similar shifts are unfolding in China.

Weichai Power, a national manufacturing champion focused on engines, powertrains and power‑generation equipment, saw its sales of backup power supplies for data centers surge 259 % year‑on‑year in 2025.

Taihao Technology, specializing in intelligent emergency power supplies and diesel generator sets, secured its first entry for Chinese‑made diesel generator sets into North‑American data‑center markets in the same year.

AI‑driven data‑center growth is benefiting far more than just generator manufacturers.

Booming data‑center construction fuels greater demand for power distribution, thermal management and building structures. Many legacy industrial‑, energy‑ and construction‑sector products are opening up new market space.

Why do AI data centers require so many generators?

Diesel generators do not serve as the primary power source for data centers.

Under normal operating conditions, data centers draw grid power. When grid outages occur, UPS (uninterruptible power supply) systems instantly take over critical loads to keep servers online. If the blackout persists, backup generators kick in to sustain power over longer durations.

Vertiv, a global major supplier of data‑center power and cooling infrastructure, notes that in its power‑supply architecture, UPS acts as a short‑term bridging power source, while diesel generators remain a widely‑adopted backup‑power option.

AI has brought two key changes, starting with vastly higher power ratings.

According to the International Energy Agency (IEA), global data centers consumed around 415 TWh of electricity in 2024, accounting for roughly 1.5 % of worldwide power use. Driven by AI training and inference workloads, data‑center power consumption is projected to approach 945 TWh by 2030 — more than doubling the 2024 figure.

Higher‑power servers demand correspondingly scaled‑up backup infrastructure. Legacy data centers of tens of megawatts required matching backup systems. As AI‑focused data centers evolve toward hundred‑megawatt and even gigawatt‑scale campuses, backup systems must deliver greater capacity and reliability.

This explains why diesel generators have not been fully phased out. The technology is mature; high‑power units can be paralleled for capacity expansion; fuel can be stored on‑site; and a complete industrial ecosystem exists covering engines, alternators, control systems and field‑service networks. For mission‑critical data‑center operations, proven performance and maintainability constitute major advantages.

Alternatives including energy‑storage systems, natural‑gas generators and fuel cells are also being trialed for data‑center deployments. Vertiv’s research indicates UPS, energy‑storage and power‑generation hardware will each fulfill distinct resilience roles, pointing toward a more diversified backup‑energy mix for future data centers.

In short: AI is not creating a boom for one single power‑generation technology. It is lifting demand across the entire data‑center energy ecosystem.

 Which traditional enterprises are joining the AI‑infrastructure supply chain?

As AI data centers scale from tens of megawatts toward hundreds of megawatts and gigawatt‑size sites, equipment suppliers are seeing dramatic order shifts. Generac exemplifies this trend, with AI‑focused data centers emerging as major new customers. In Q2 2026, its commercial‑industrial revenue rose nearly 30 % year‑on‑year. Comparable momentum is visible at established industrial players such as Caterpillar.

Unlike Generac, Caterpillar did not build its position via residential‑grade backup generators. Instead, it leverages decades‑worth of powertrain and energy‑equipment expertise to supply AI‑data‑center power infrastructure.

In 2026, the company partnered with American Intelligence & Power to deliver 2 GW of fast‑response natural‑gas generator sets paired with battery‑energy‑storage systems for hyperscale AI‑infrastructure projects.

In Weifang, Shandong Province, established power‑equipment manufacturers are tapping these emerging use cases. Weifang hosts a dense power‑equipment industrial cluster. Xinhua News Agency has reported the region is home to multiple national manufacturing champions — Weichai Power among them — alongside a full ecosystem of engine, generator‑set and component subcontractors.

Weichai Power was an early mover in this market. Data centers became a key vertical for its power‑energy business in 2025, when its data‑center backup‑power sales multiplied approximately 3.6‑fold year‑on‑year. To meet large‑scale data‑center requirements, sales of its M‑series large‑bore engines exceeded 10 000 units for the first time (up 32 % YoY), alongside the launch of its 5 MW high‑speed diesel generator product, the 20M61.

Taihao Technology has long specialized in intelligent emergency power. Its high‑power intelligent emergency power systems serve internet‑firm and carrier‑operated data centers, and the firm achieved the landmark 2025 breakthrough of landing Chinese‑built diesel generator‑set deployments in North‑American data centers.

These enterprises were not born out of the AI boom. Rather, their long‑accumulated manufacturing capabilities in engines and powertrains are finding new application in AI‑physical infrastructure.

Demand ripples well beyond generators to broader manufacturing segments. A July Xinhua field investigation into the computing‑infrastructure industrial chain in Jiangsu Province noted that Changzhou Huapeng Group and Sumec Electromechanical, among others, have benefited from surging overseas demand for computing‑infrastructure hardware.

The expansion of AI data centers is driving demand not only for semiconductors and electricity but reshaping production footprints for upstream equipment makers.

Generators are only one example. Larger‑scale data centers create new market opportunities for transformers, liquid‑cooling systems and modular building solutions. Overseas industry surveys of US manufacturing supply chains report rising orders for Timken (engineered bearings and industrial transmissions), Siemens, plus makers of cables, industrial hoses and prefabricated building panels.

For Chinese manufacturers, power‑generation hardware represents just one segment of AI‑infrastructure demand. Ongoing data‑center build‑outs unlock new commercial potential across transformers, liquid cooling, optical communications, servers and modular construction.

Notably, some Chinese firms are already building overseas production bases and supply chains targeted at these needs. We will continue to track how Chinese manufacturing integrates into global AI‑data‑center supply chains and the new export opportunities unlocked by this wave of infrastructure construction.