Alibaba raises AI cloud prices by 34%, adding pressure on the compute market

Alibaba Group has raised prices for its cloud services for data storage and artificial intelligence compute by 34%. It is another signal that the race for AI infrastructure is moving from an investment stage to a stage of hard monetization.

Why prices rose

The reason is fairly straightforward: AI turned out to be much more expensive to operate than many expected at the start of the mass boom. Cloud providers are spending billions on servers with accelerators, networking infrastructure, new data center capacity and storage for ever-growing volumes of data. While the market was growing, part of these costs could be covered by expectations of future demand. Now, though, providers increasingly want to recoup their investments not at some point down the road, but through current tariffs.

Alibaba's move fits the global logic of the market. Big technology companies no longer sell AI infrastructure at the lowest price as a ticket into an experimental phase. They are starting to treat it as a premium resource with high costs and steady demand. The more actively businesses adopt generative models, search over large data sets and LLM-based automation, the easier it is for cloud platforms to revise prices without fearing that customers will sharply cut their workloads.

Who it will affect

A 34% increase is most painful for those whose AI has already left the pilot stage and become part of daily processes. When a model runs not on a test bench but in a product, a support service, analytics or internal search, any tariff adjustment quickly shows up in the P&L. The effect will be especially noticeable for teams that pay for both compute and storage of large datasets, logs, embeddings and intermediate results.

• Startups where AI workloads make up a large share of product cost
• Enterprise teams scaling pilots to production
• Services with constant inference and a large number of user requests
• Projects that need to store data, processing results and technical logs for a long time

It is important that the price rise does not always mean the whole cloud bill automatically grows by exactly 34%. The final amount depends on what share of spending goes to the affected services. But even so, the news changes how teams plan: companies will have to count the unit economics of AI features more carefully, drop weak scenarios and look for a balance between model quality, response speed and the cost of a single operation.

A new stage for the market

Alibaba's price increase shows that the AI cloud market is entering a more mature phase. Early on, the priority was to grab share quickly, attract developers and lock customers into one's own infrastructure. Now the focus is shifting to margin, payback and managing capital-intensive growth. This changes the logic of procurement itself: instead of asking "where is it cheapest to test an idea," companies increasingly ask "where is the cost of scaling more predictable six or twelve months from now."

For customers, this also increases interest in optimization. Routing requests between models, hybrid setups with part of the workload outside the public cloud, tighter control over storage and deletion of unused data come to the fore. In other words, the era when teams could endlessly expand AI experiments without serious financial control is ending quickly. The price of infrastructure is becoming as strategic a factor as model quality or the speed of shipping new features.

What it means

For business, the takeaway is simple: the AI budget now has to be counted as a full infrastructure line item, not as a temporary experiment. Alibaba's decision shows that the period of relatively cheap AI cloud is shrinking, and companies will have to assess more strictly which scenarios really pay off, where to store data and on what capacity to run models.