Alibaba's AI spending is shifting from a phase of heavy capital expenditure into a period where returns are becoming tangible. According to the latest notes from Goldman Sachs, Alibaba's management expects its AI infrastructure investments to reach breakeven within three years, and as the business scales further, that payback period could potentially shorten even more.

Goldman Sachs believes this outlook rests on three key pillars: strong visibility into AI demand, accelerating cloud adoption, and steadily improving cloud profit margins. As AI applications move from model training toward inference and real-world enterprise production, demand for cloud-based computing power is expected to expand further.

Meanwhile, Alibaba Cloud is emerging as a key beneficiary of these AI investments. The cloud division is seeing faster revenue growth and expanding margins, with management setting a medium-term target of 20% for cloud operating margins. They also project annualized recurring revenue (ARR) from MaaS to reach RMB 30 billion by fiscal year 2027.

Looking further ahead, Alibaba is stepping up its push into overseas cloud markets. The company has set a target of $100 billion in external cloud revenue by 2030, with a focus on expanding in Southeast Asia, the Middle East, and Latin America. Goldman Sachs notes that with its full-stack capabilities and a massive existing customer base, Alibaba Cloud is well-positioned to amplify its first-mover advantage in the AI era. AI Investments to Break Even in Three Years, Backed by Strong Cash Flow

Goldman Sachs' Asia research team, in notes from the GS Asia Leaders Conference held on September 1, highlighted that Alibaba's management is confident about converting AI investments into revenue growth and shareholder returns. The three-year payback period provides a clear return expectation that supports the company's continued heavy spending on computing power.

Importantly, Alibaba has sufficient internal funding sources and does not rely heavily on external financing. Its core e-commerce business remains the most significant source of cash flow. Excluding investments in instant retail (Quick Commerce), the core business generates approximately $25 billion in cash annually. At the same time, the scale of instant retail-related investments is shrinking by about half each year, which should gradually ease cash flow pressures.

The cloud business is also expected to contribute incremental cash. With faster revenue growth and improving margins, Alibaba Cloud's ability to generate cash on its own is strengthening. Additionally, customer prepayments can partially offset the prepayments Alibaba makes for leased computing capacity, while improved working capital efficiency should further enhance overall cash flow.

Management also cautioned that the capital expenditure run rate from the June quarter should not be simply extrapolated across the full year. Due to quarterly fluctuations in computing capacity supply, AI investment pacing is uneven, and a single quarter's data may not directly represent the annual capex level. Doubling Down on Computing Power to Secure AI Entry Points and Extend First-Mover Edge

For Alibaba, the core purpose of this round of high-intensity capital spending is not just to expand computing capacity, but to secure a strategic position in AI infrastructure and cement its first-mover advantage. The company has already built a full-stack capability spanning IaaS, PaaS, and MaaS, while also holding a vast existing customer base. According to Goldman Sachs' notes, Alibaba Cloud currently accounts for roughly 40% of China's AI cloud market.

Goldman Sachs sees this customer base as a key advantage for Alibaba's continued AI investment. As enterprise AI demand expands from model training to inference, development, and actual production, the links between computing power, platforms, and model services will become even tighter. Vendors with a complete technology stack are better positioned to capture this incremental demand.

Alibaba's current strategy is to embed more AI capabilities into its cloud service system using its existing infrastructure and customer network. In this way, new computing capacity not only supports model training but also converts into inference, development, and application demand from enterprise clients.

Thus, AI capital expenditure and Alibaba Cloud's existing business are not separate tracks but are mutually reinforcing: computing power investments expand supply capacity, model and platform capabilities strengthen customer stickiness, and growing customer demand in turn boosts utilization of computing resources. Cloud Revenue Accelerating, Approaching 20% Margin Target

Beyond AI infrastructure spending, Alibaba Cloud itself has entered a phase of accelerated growth. The company projects that MaaS annualized recurring revenue (ARR) will reach RMB 30 billion by fiscal year 2027. Over the longer term, it targets $100 billion in external cloud revenue by 2030.

Overseas markets will be a vital growth driver toward this goal. International business currently contributes roughly 10% to 20% of Alibaba Cloud's revenue, with the company focusing on Southeast Asia, the Middle East, and Latin America. Its clients span AI models, new energy vehicles, biotech, traditional manufacturing, and embodied intelligence sectors.

On profitability, management has set 20% as the medium-term margin target for the cloud business and indicated that "progress is on track." The impact of previously implemented price adjustments will gradually show up in revenue growth and margins over the coming quarters. Meanwhile, a rising share of the pay-per-use (PPU) model is expected to become a key driver of margin expansion over the medium to long term.

This suggests that Alibaba Cloud's future growth is not solely reliant on expanding revenue scale; revenue growth and margin improvement are expected to work together as dual drivers. AI Commercialization Prioritizes B2B, Qwen Betting on the Agent Era

At the model level, Alibaba is not simply chasing ever-larger parameter counts but aims to build a comprehensive model matrix covering different application scenarios, including coding, world models, and video models. However, when it comes to commercialization priority, Alibaba's choice is clear: the B2B segment comes first.

Compared to consumer-facing (B2C) markets, enterprise clients show more defined willingness to pay for AI, and the path from cloud infrastructure and model services to enterprise applications is more direct in terms of revenue conversion. Thus, Alibaba is currently placing greater emphasis on enterprise-side commercial deployment.

For the consumer segment, the approach is more long-term. The Qwen app is unlikely to achieve profitability in the near term, but Alibaba has no plans to abandon investment in it. Management values its potential in the future era of Agentic Commerce. As AI agents gradually take part in consumer decisions, users may eventually complete product searches, comparisons, purchases, payments, and fulfillment through intelligent agents.

For Alibaba, which owns a complete ecosystem including e-commerce, Qwen is therefore not just a consumer AI product but could become a crucial gateway connecting users with the commercial ecosystem through future agents. Goldman Sachs Maintains Buy Rating as AI Enters a Return-Verification Phase

Overall, the key takeaway from Goldman Sachs' latest exchange with Alibaba's management is that the company's AI investment thesis is shifting from a "spending logic" toward a "return logic." On one hand, the core e-commerce business provides stable cash flow, underpinning AI capital expenditure. On the other hand, AI computing power, model capabilities, and the cloud business are forming a tighter commercial loop.

More importantly, management has provided a fairly clear payback period for AI infrastructure—three years. If cloud revenue growth and margin expansion continue to deliver, this payback cycle could potentially shorten further. Goldman Sachs maintains its "Buy" rating on Alibaba. For the market, what remains to be verified is no longer just "how much Alibaba will spend," but whether AI investments can translate along the three-year payback path into cloud revenue growth, margin improvement, and ultimately shareholder returns.