The shift from compute to tokens is the inflection point. For years, cloud economics were defined by selling GPU hours and storage-linear, commoditized infrastructure. But as agentic AI takes hold, the unit of value is changing. Alibaba sees this clearly. At the Siemens RXD Summit in Beijing, Chairman Joe Tsai framed the opportunity in terms that would make any exponential investor sit up: agents are "virtual knowledge workers" capable of disrupting a TAM bigger than all other industries combined. With nearly $50 trillion of the global $110 trillion economy tied to white-collar knowledge work, the addressable market for agentic AI is not just large-it's civilization-scale.
This is the S-curve moment. The transition from compute-based pricing to token-based economics creates non-linear revenue potential because tokens capture value at the point of action, not just computation. As Citi Research noted in a recent report, "the shift from selling resources to selling intelligent capabilities creates ample room for non-linear revenue growth and profitability improvement." Alibaba is not waiting for the curve to rise-it has reorganized to ride it.
The structural proof is the Alibaba Token Hub (ATH) Business Group, newly established and structured around a three-layer framework: creating, distributing, and applying tokens. At the application layer, tokens power Wukong, the enterprise-grade AI-native agent platform that moves AI from "thinking partners" to "doing partners." On the consumer side, the Qwen App has become China's first all-in-one personal AI assistant, reaching 300 million monthly active users as of February. Underlying both is the elevation of Model-as-a-Service (MaaS) to an organizational priority-projected by CEO Eddie Wu to become the largest revenue product within Alibaba's Cloud Intelligence Group.
The flywheel is self-reinforcing: superior models attract more applications to the MaaS platform, which generates data that strengthens those models, which in turn attracts more customers. The momentum is already visible in the numbers. Alibaba Cloud has recorded over RMB100 billion in revenue from external customers in just 11 months (through February FY2026). The five-year target? Surpassing $100 billion in that same revenue metric, including MaaS.
But here's where the strategy gets interesting-and where Alibaba diverges from the open-source orthodoxy that has dominated Chinese AI. The company has released its third proprietary model in as many days, with Qwen3.6-Plus joining closed-source upgrades to image generation and multimodal models. This is a deliberate pivot. While competitors like MiniMax and DeepSeek open-source to drive adoption, Alibaba is choosing to retain control where it matters most: monetization. As the company stated, going proprietary in select instances allows it to "retain greater control and charge more users directly."
The logic is straightforward. Open-source builds ecosystem gravity. Proprietary captures value. By pairing its most popular open-source Qwen platforms with closed-source, high-value models integrated into Wukong and the flagship Qwen app, Alibaba creates a model supermarket where users can start with free or low-cost access and upgrade to premium, action-oriented intelligence. This is the infrastructure layer play: not just selling compute, but selling the tokens that execute work.
For investors, the implication is clear. Alibaba is not positioned on the linear tail of cloud infrastructure. It is building the rails for the agentic AI economy-the place where intelligence becomes action, and where revenue scales with usage, not capacity. The $50 trillion TAM is not a prediction. It is the starting line.
Revenue Acceleration Evidence: The S-Curve Is Bending Upward
The strategic pivot is no longer theoretical. The numbers prove Alibaba Cloud is riding the upward bend of the AI S-curve.
Cloud revenue surged 36% year-over-year to 43.3 billion yuan ($6.2 billion) in the October-December quarter, outpacing overall group revenue growth of just 2% in the October-December quarter. That's the first signal: AI-native revenue is scaling faster than the legacy business. More critically, Alibaba Cloud's external commercial revenue crossed the RMB 100 billion threshold in February 2026 as of February 2026-a milestone achieved in just 11 months. The five-year target is to surpass $100 billion in that same metric, including MaaS revenue.
Market position confirms the momentum. Alibaba Cloud retained its #1 IaaS position in Asia Pacific with 22.5% market share, up from 20.8% in 2024 in the latest Gartner report. The growth isn't stagnant-it's accelerating. Singapore posted triple-digit year-over-year growth, and the global IaaS market share expanded to 7.7% from 7.2% in 2024. These aren't incremental gains; they're the signature of a company capturing disproportionate demand as AI-native workloads become the dominant source of new cloud consumption.
The monetization model is shifting in real time. During the Spring Festival 2025, Qwen deployed a RMB 3B subsidy program-the "Spring Festival Treat Plan"-integrating AI across Alibaba's entire on-demand ecosystem connecting its AI assistant. This wasn't a promotional campaign; it was a structural experiment in agent-driven commerce. By embedding AI into instant retail-where decision cycles are measured in minutes rather than days-Alibaba is testing a revenue model that moves beyond advertising and commissions toward infrastructure-level monetization as AI agents autonomously manage household replenishment.
The implication for investors is clear: the S-curve is bending upward, and the inflection point is now. Cloud revenue growth at 36% YoY, external commercial revenue exceeding RMB 100 billion, and dominant IaaS market position in Asia Pacific-these metrics collectively signal that Alibaba has transitioned from selling compute hours to selling intelligent capabilities. The $100 billion five-year target isn't a prediction. It's a trajectory.
Valuation Implications: $100B Target and the Path to Exponential Returns
The $100 billion target isn't just ambitious-it's a declaration that Alibaba sees itself on the steep part of the S-curve. To reach $100 billion in cloud and AI revenue from the current ~$6.2 billion quarterly run rate, the company needs to sustain roughly 60%+ compound annual growth over five years. That's not linear extrapolation. That's the signature of exponential adoption kicking in.
But here's the disconnect: the market is not pricing this as an exponential play. At 24.09 PE with a 0.49 beta, Alibaba trades like a moderate-growth infrastructure name-low volatility, steady expectations. That's the gap.
The 34% price hikes signal confidence that AI-driven demand is inelastic. When you can raise prices that aggressively and expect adoption to hold, you're no longer selling compute hours-you're selling tokens that execute work. The unit of value has shifted. Customers aren't buying capacity; they're buying outcomes. That's the monetization leverage that transforms an infrastructure business into an exponential one.
The strategic pivot explains the profit trade-off. Net profit fell 67% to 16.3 billion yuan as Alibaba invests heavily in infrastructure and AI capabilities during the October-December quarter. This is the classic S-curve investment pattern: front-load capital to capture the steep ascent, accept margin compression today for non-linear revenue tomorrow. The $100 billion target includes MaaS revenue-the model supermarket is the infrastructure layer, and tokens are the unit of account.
For investors, the question is whether the S-curve is already bending upward in the numbers. Cloud revenue at 36% YoY, external commercial revenue crossing RMB 100 billion in just 11 months, dominant IaaS position in Asia Pacific-these are the early signals. The current valuation doesn't reflect the exponential trajectory. It shouldn't yet. But if the model supermarket thesis holds, the gap between where BABA trades and where it should be is where the asymmetric upside lives.
The key insight: Alibaba is being priced on the linear tail of cloud infrastructure, not the exponential rise of agentic AI. The 34% price elasticity test, the $100 billion five-year target, the proprietary model strategy-these all point to a company that has already crossed the inflection point. The market just hasn't caught up.
Catalysts and Risks: What Could Break the Thesis
Every S-curve play faces binary outcomes. For Alibaba, the difference between a transformative win and a costly detour comes down to three specific watchpoints.
HappyHorse-1.0: The Benchmark Catalyst
The anonymous AI video model that stormed global leaderboards isn't just a curiosity-it's a potential inflection point. HappyHorse-1.0's dominance in blind-test rankings for both text-to-video and image-to-video generation arriving without identifying its affiliations validates Alibaba's capability to produce frontier models. With OpenAI exiting Sora and ByteDance's Seedance 2.0 facing copyright headwinds, the competitive landscape has cleared. If Alibaba sustains this benchmark lead, it gains asymmetric positioning in the next generation of AI-generated content infrastructure. The market's 2.12% rally on confirmation of Alibaba's involvement after news of its involvement is just the opening move.
The Profit Collapse: 7% Drop as Warning or Investment Signal?
The 67% profit decline during the October-December quarter triggered a 7% selloff-but this is the classic S-curve trade-off. The question isn't whether profit takes a hit; it's whether the investment converts to exponential revenue. Alibaba's $100 billion cloud target demands front-loaded capital. The market's reaction reflects linear thinking applied to an exponential play. What matters: does cloud revenue growth at 36% YoY in the October-December quarter accelerate or decelerate in coming quarters? That's the signal.
Qwen3.6-Plus + Wukong: The Monetization Watchpoint
The integration of Qwen3.6-Plus with Wukong represents the critical test of Alibaba's proprietary model strategy. This isn't just another model release-it's the mechanism through which Alibaba validates whether closed-source, high-value models can drive monetization. The 34% price elasticity test for AI services combined with this integration will show whether the model supermarket thesis holds. If Wukong adoption accelerates with Qwen3.6-Plus powering enterprise agents, the revenue per token thesis validates. If not, the proprietary pivot risks becoming an expensive experiment.
The thesis holds if benchmark dominance translates to enterprise adoption, if profit compression converts to revenue acceleration, and if the model supermarket captures value at the action layer. Any break in this chain-benchmark erosion, profit decline without revenue offset, or integration failure-breaks the thesis. The next two quarters will show which path Alibaba takes.