Alibaba Is Burning Cash on AI — and That Changes Everything

The Number Nobody’s Talking About

Alibaba reported a 75% drop in quarterly profit — down to roughly $1.6 billion from $6.4 billion the year before. On the surface, that looks like a disaster. But look closer at what’s driving it, and you’ll see something more interesting: a company actively choosing to sacrifice near-term earnings for a long-term bet on agentic AI infrastructure.

This isn’t just Alibaba repositioning itself. It’s a signal of how one of the world’s largest tech companies is navigating the most expensive pivot in its history — and what that means for everyone else watching from Washington, Brussels, and Beijing.

The Real Story Behind the Drop

The profit decline tells only half the story. Revenue grew 9% to nearly $40 billion. AI cloud and compute revenue surged 45% to $7.2 billion. That’s the kind of growth curve that makes investors nervous in the short term but could look like genius in five years — if the infrastructure spend pays off.

Capital expenditures, however, jumped 75% to about $10 billion. Alibaba didn’t lose money to bad bets or operational inefficiency. It spent that money deliberately: on CPU capacity, chip procurement, and the data center infrastructure required to run AI agents at scale.

CEO Eddie Wu framed it plainly during the earnings call. He said revenue growth would accelerate further as supply ramps up, alongside continued profitability improvements. Whether he’s right remains to be seen. What’s clear is that Alibaba has already committed at least $56 billion over three years to this infrastructure push — roughly what the company earned in profit last year alone.

Agentic AI Is the Battlefield

The specific technology driving this spend matters. Alibaba isn’t just building another chatbot. It’s moving toward agentic AI — systems that can plan, execute, and adapt autonomously. That’s a fundamentally different architecture than the question-and-answer models that dominated the early AI gold rush.

Agentic AI requires more compute, more memory, and more complex orchestration. It’s also closer to where enterprises actually want to go: autonomous workflows, not conversational toys. Companies that figure out agentic AI first will have a structural advantage in the next wave of enterprise software.

Alibaba’s Qwen series is part of this play. The company previewed Qwen3.8-Max in July, claiming it ranks “second only” to Anthropic’s Claude Fable 5. That claim needs independent verification, but the direction is unmistakable: Alibaba is competing head-on with the best American models, not hiding in commodity services.

The China Context — What Western Readers Miss

Most English-language coverage treats this as an Alibaba story. It’s really a China story.

Beijing has been pushing its tech giants toward self-reliance in semiconductors and AI for years. US export controls on advanced chips have only accelerated that mandate. Alibaba’s capex surge reflects both commercial ambition and strategic necessity: if you can’t buy the best chips from the US, you build your own ecosystem.

That ecosystem matters beyond China. Alibaba’s cloud platform serves customers across Southeast Asia, the Middle East, and Europe. A successful agentic AI stack built on domestically controlled hardware could become the default infrastructure layer for non-Western markets — exactly the kind of scenario Washington wants to prevent.

The $100 billion annual revenue target within five years — AI plus cloud combined — isn’t vanity. It’s a benchmark for whether Chinese tech can achieve scale without relying on American chip supply chains. If Alibaba gets there, every other Chinese company will try.

Who Wins, Who Loses

Alibaba wins if the infrastructure spend converts into market share before competitors exhaust their own capital buffers. The company has deep pockets, a massive domestic user base, and government backing — three advantages American rivals can’t easily replicate in China.

US hyperscalers win if Alibaba’s spend becomes a trap: too much capital deployed too fast, with returns lagging behind the ambitious timeline. AWS and Azure have deeper margins today. They can afford to subsidize AI infrastructure for years without panic. Alibaba is betting its current profitability on a turnaround that hasn’t happened yet.

Enterprise customers win regardless — competitive pressure between Alibaba, AWS, Azure, and Google Cloud should keep prices down and capability up. That’s the whole point of infrastructure investment: lock in users before they migrate elsewhere.

What Happens Next

Alibaba’s shares fell more than 3% on the report. The market is pricing in uncertainty, not collapse. The real question is timing.

If AI and cloud revenue growth maintains even half the current pace, the capex will start paying off somewhere between 2027 and 2028. That aligns with Alibaba’s own timeline for surpassing $100 billion in combined annual revenue. Miss that window, and the company faces the uncomfortable choice between cutting back — and losing ground — or doubling down on a bet that’s already consumed years of profit.

The broader implication extends far beyond one company’s balance sheet. Every major AI economy is in the same spending spiral: invest heavily now or fall behind. Alibaba’s 75% profit drop is a symptom of that dynamic, not an exception to it. The difference with Alibaba is that it’s happening inside a company that sits at the intersection of e-commerce, cloud, and AI — three of the most valuable segments in tech — and is willing to cannibalize its own profits to lead the next round.

That’s either the most courageous strategy in Chinese tech or the most expensive one. We won’t know the answer for several quarters. But we do know this: the agentic AI arms race has begun, and Alibaba just made one of the largest opening moves.