Alibaba and Baidu Raise AI Cloud Prices by Up to 34% as China’s Discount War Starts to Reverse

Alibaba Cloud and Baidu AI Cloud are raising prices on AI computing and storage services by roughly 5% to 34%, with both companies saying the changes will take effect on April 18 after global AI demand and hardware-related supply costs climbed sharply. The timing matters as much as the percentages. After years in which China’s cloud market was defined by aggressive discounting, near-simultaneous price increases from two of the country’s biggest AI infrastructure suppliers suggest compute is no longer being treated only as a tool for buying market share. It is starting to look like an asset that large Chinese cloud platforms believe they can finally price for scarcity as well as scale.

Two Chinese cloud leaders moved within hours of each other

The basic facts are unusually clean for a China cloud pricing story. Alibaba said on March 17 that prices for AI computing services and CPFS, its parallel file storage offering for AI workloads, would be adjusted from April 18 because exploding global AI demand and rising supply-chain costs had pushed core hardware procurement costs materially higher. Baidu published a similar notice on March 18, saying demand for computing power had continued to climb as AI applications spread and that hardware plus related infrastructure costs had also risen significantly.

The price ranges were large enough to make the move impossible to dismiss as routine fine-tuning. According to Bloomberg, Alibaba’s AI-related computing services will rise by 5% to 34%, while CPFS pricing will go up about 30%. Baidu’s official notice said AI computing services will increase by about 5% to 30%, with parallel file storage also rising roughly 30%. Both companies said existing orders would keep their current pricing through the present billing cycle, with the new terms applying from the next renewal period after April 18.

That parallel structure is the real signal. This was not a single vendor quietly repricing one product family. It was two Chinese cloud leaders acknowledging, in nearly identical language, that the economics of AI infrastructure have changed. China Daily described the shift as a potential turning point in an industry long defined by aggressive discounting, which is a restrained way of saying the old playbook may be breaking down.

This is a sharp reversal from the old price-for-volume era

The contrast with the recent past is what gives the story its weight. Early in 2024, Alibaba cut prices on cloud computing services by as much as 55%, and rivals moved quickly to match or respond. That period reinforced a familiar view of China’s cloud sector: scale first, discounts everywhere, profitability later. Vendors were willing to sacrifice price discipline if that was what it took to win developers, enterprise workloads and ecosystem loyalty.

The latest round of price hikes points in the opposite direction. Dow Jones, via Morningstar, framed the new moves as an inflection point in the race to monetize AI. That interpretation fits the broader Chinese market context. We saw a similar pressure building in our earlier look at the $369M AI infrastructure war reshaping cloud competition, where the economics of compute access were already starting to define competitive positioning. Over the past year, China’s largest internet and cloud groups have accelerated investment in foundation models, inference services and enterprise AI tooling, but those efforts have also raised questions about when AI stops being an expensive strategic project and starts becoming a business that can support margins.

That is why the Alibaba-Baidu story matters beyond a few product pages. In China’s cloud market, pricing has always carried strategic meaning. Cutting prices signaled willingness to subsidize adoption. Raising them now suggests some providers believe demand has become sticky enough, and supply tight enough, to justify charging more without instantly losing the market.

Demand growth and hardware inflation are colliding in the same place

Both official notices put demand and cost on the same side of the equation, and that is what makes the story more than a simple inflation pass-through. Alibaba explicitly pointed to a global AI demand explosion and to higher prices across the supply chain, saying procurement costs for core hardware had risen significantly. Baidu used similar language, saying rapid growth in AI applications had kept demand for computing power climbing while core hardware and related infrastructure costs also moved up.

That pairing matters because AI infrastructure is not just about raw chips. It is also about storage, networking, data movement and the ability to keep training and inference systems running at industrial scale. Parallel file storage products such as CPFS are not decorative extras; they are part of the data backbone for large-model training, checkpointing and high-throughput AI workloads. If both compute and storage are becoming more expensive at the same time, the cost pressure is reaching deeper into the stack.

It also helps explain why China’s cloud giants are acting now instead of waiting for a cleaner market. If demand keeps rising while procurement costs for chips, servers and supporting infrastructure stay elevated, permanent discounting becomes harder to defend. In that environment, price increases are not just a revenue decision. They are also a capacity-allocation decision about which workloads deserve premium infrastructure and which customers can absorb the bill.

The China angle is the whole story, not a local footnote

This is not a generic global cloud pricing story with a China dateline attached to it. The companies involved are Alibaba Cloud and Baidu AI Cloud, two domestic platforms deeply embedded in China’s AI and enterprise technology market. Their customers include Chinese startups, large internet companies, industrial firms and enterprise developers building model training, inference and agent-style applications for the domestic market. Baidu’s enterprise push has also shown up in infrastructure-led deals such as its 837 million yuan AI infrastructure project from China Unicom’s Shandong arm, underscoring how strategically important this layer has become.

That makes the downstream impact particularly important. For China’s AI startups and mid-sized enterprise users, higher prices for compute and parallel storage could raise the threshold for experimentation, deployment and scaling. A company that could previously justify one training cycle or a broader inference rollout under aggressive discounting may now have to optimize harder, narrow model choices or stretch renewal budgets. In other words, the pricing shift does not just change cloud vendor revenue. It can also reshape which kinds of AI projects remain economically viable in China.

At the same time, the move says something about China’s supply chain exposure. Neither Alibaba nor Baidu framed the increases as a domestic policy story. They framed them as the result of global AI demand and rising hardware or infrastructure costs. That is significant because it shows how global pressure on AI components and systems is feeding directly into the pricing of services sold inside China’s domestic cloud market.

A real signal, but not yet proof of a full industry reset

There is still reason to stay measured. Publicly confirmed price increases so far are from Alibaba and Baidu, not every major cloud company in China. Higher pricing also does not automatically mean profits will improve quickly. Capital spending, customer bargaining power and the cost of keeping AI platforms competitive can still absorb much of the upside. A pricing move is evidence of intent, not guaranteed proof of a margin rebound.

Even so, the direction is hard to ignore. Bloomberg, Dow Jones and China Daily, together with the companies’ own notices, all point toward a similar reading: AI demand is no longer only a growth narrative for China’s cloud vendors; it is becoming a monetization test. That change in framing matters because it suggests China’s AI infrastructure market may be entering a more mature phase, where access to compute is priced with more discipline and less desperation.

What changed, and what could happen next

What changed this week is that China’s AI infrastructure story stopped looking like a pure subsidy race and started looking like a pricing-power story. Two of the country’s largest cloud and AI platforms raised prices within hours of each other, using nearly the same explanation: demand is rising faster, hardware and infrastructure cost more, and the old discount logic is no longer enough. That is a meaningful shift in a market that spent the last two years teaching customers to expect cheaper cloud resources, not more expensive ones.

What could happen next depends on whether demand stays strong and whether supply costs keep climbing. If both forces persist, other Chinese cloud providers may try more selective price increases, tiered packaging or tighter discounts around premium AI services. If competition intensifies again or hardware pressure eases, the latest hikes could prove narrower than they now appear. Either way, something important has already changed. In China’s AI market, computing power is starting to be sold less like a promotional product and more like a scarce industrial input whose owners believe they can finally charge for it.