Off-Prem

CFO says GPU rentals are ‘structurally higher margin than CPU cloud’

Chinese web giant Baidu has told investors its rare ability to build and operate AI infrastructure at scale represents a new high-margin business that its customers can’t avoid.

Speaking on the company’s Q1 2026 earnings call, CEO, chairman and co-founder Yanhong Li said GPU cloud revenue increased by 184 percent year-over-year which represented “growth well above the broader market.”

CFO Haijian He said that Baidu’s GPU cloud “is structurally higher margin than traditional CPU cloud, driven by stronger demand, tighter supply chain, higher technical barriers and pricing power.” He added his view that AI applications are “naturally high-margin business, driven by sticky and subscription-based models and operating leverage over time.”

Dou Shen, the president of Baidu’s AI Cloud Group, remarked “While high-quality supply is relatively tight, customers prioritize proven stability and availability, not just cost.”

“For enterprises, it's not only about the peak chip performance,” he said. “What matters more is the stability at scale, compatibility with mainstream models and frameworks, migration costs and friction, support for a large-scale cluster deployment and ultimately, cost efficiency.”

He thinks the AI market will "increasingly consolidate around players who can deliver on all of these dimensions” and thinks Baidu is nailing them.

“We have seen remarkably strong enterprise demand for AI infrastructure, both training as well as inference,” he said. “Inference is showing particularly strong momentum, which is a pretty healthy signal. It tells us that customers have moved beyond training models and are now running AI across more parts of their business at an accelerating pace.”

That Baidu creates its own Kunlunxin AI chips means he thinks the company will emerge in a strong position.

“Our Kunlunxin AI chips and full stack AI capabilities give us more room to optimize costs and continued improvement in our customer mix further supports margin expansion,” he said.

Baidu is one of many hyperscalers building its own AI chips and ecosystems, so if the Chinese company’s experience is universal the enormous sums of cash US-based clouds are spending on AI infrastructure may well pay off over time.

Shen also shared his views on Chinese AI chips, which he admitted “are still catching up with the most advanced global products in certain frontier training scenarios.”

He added his opinion that Chinese chips can handle inferencing workloads, but said local buyers and chipmakers “still face near-term challenges around the capacity and supply chain maturity, partly because demand is growing faster than supply.”

CEO Yanhong Li proudly revealed increased use of Baidu’s Apollo Go robotaxis but said as the company deploys them more widely “we have encountered a broader and increasingly complex range of real-world scenarios, including system and operational complexities that only emerge at larger scale.”

“We are addressing a new frontier centered on how robotaxi services fit more naturally into public transportation, city operations and everyday life,” he said. Once Baidu figures that out, he expects robotaxis will “coexist more seamlessly with the broader transportation ecosystem over time and ultimately to become a more convenient and trusted service for the people we serve.”

The CEO also discussed Baidu’s “Digital Human” business, which offers interactive avatars-as-a-service that customers often use to interact with their clients online or host online infomercials.

Yanhong said Baidu has reduced the cost of operating digital humans by 80 percent in the last two quarters, taught them 24 languages and even added “presentation styles culturally adapted to resonate with local audiences.”

“This helps merchants run around-the-clock digital human live streams that feel authentically native, unlocking new levels of efficiency and conversion potential across global markets,” he said.

Baidu’s AI revenue numbers remain modest – even the massive growth mentioned above saw its AI cloud revenue reach RMB 8.8 billion ($1.3 billion). But the company was pleased that AI-related products accounted for over half of all revenue for the first time, accounting for RMB 13.6 billion $2 billion) of the quarter’s RMB 26 billion take ($3.8 billion).

Without the spike in AI-related sales, Baidu’s quarterly revenue would have gone backwards. ®