Ahead of Google: Alibaba’s Edge in the Super AI Agent Race

Analysis for informational purposes only. Capital at risk.

Highlight

• Alibaba’s Lead in the AI Agent Race: While Google is still trying to tackle the API walls created by a fragmented industry, Alibaba has already showcased its Qwen AI agent, which delivers seamless AI shopping across online retail, food delivery and travel.
• The “Silent Killer” Strategy: Qwen intercepts user intents such as booking flight or ordering bubble teas and routes transactions into Alibaba’s ecosystem, effectively disintermediating vertical rivals such as Meituan and Trip.com.
• Strategic Trade Off: Alibaba is consciously willing to cannibalize ad revenue from merchants in order to recapture market shares. It prioritises growing user base over maximizing yield from marketing services.

Agentic AI – The Next Major Leap

Agentic AI, or “Super AI Agents,” is widely seen as the next major step in AI. These agents can execute real‑world tasks such as shopping, booking hotels and making payments from a single user command. The prevailing narrative is that AI agents will displace websites and apps, and users will simply ask and receive.

The “Brain Without Hands” Problem in the US

However, US attempts to build Super AI Agents have run into practical obstacles.

The “API Wall”: Major online verticals such as Amazon are concerned about being disintermediated and therefore limit direct API access to OpenAI, Google, and other third-party agents. Without authorized access, AI agents cannot integrate seamlessly with transactional systems.

The Optical Workaround: Some companies, including Microsoft and Anthropic, resort to “optical simulation”, which takes screenshots, interprets pages, and simulates user inputs. However, it is slower and more error-prone than human interaction, sensitive to UI changes, and prone to breaking on multi-step flows.

Google’s Universal Protocol: Google recently announced the Universal Commerce Protocol, or UCP, a standardized language intended to let AI agents communicate with merchants and platforms. The initiative has attracted support from partners including Shopify, Etsy, Wayfair, Target, and Walmart. However, Amazon has not joined the alliance, so there is still no single entry point for shoppers. Implementation timelines remain unclear, so UCP’s ability to enable seamless agent execution and payments is still unproven.

Alibaba’s “Walled City” Solution

On the other hand, Alibaba already showcased its Qwen AI Agent by leveraging its existing ecosystem. Qwen does not need to simulate mouse clicks. It integrates natively across Alibaba’s services.

Alibaba owns the payment (Alipay), the inventory (Fliggy, Taobao, Ele.me, and others), and the Map (Amap). Qwen operates via native API rather than optical simulation.

User case 1: Instant commerce—buying bubble tea

• User prompt: “Order 40 cups of bubble tea from Bawang Chaji.”
• Qwen action: Searches Taobao Instant Commerce, applies the best coupon, and generates an Alipay order.
• User action: Confirms payment with one tap inside the chat. No app switching, no links.

User case 2: Business trip—flight booking

• User prompt: “Book the first flight to Beijing tomorrow on Air China.”
• Qwen action: Queries Fliggy inventory, auto-fills passenger ID from Alipay and completes the booking.
• User action: Confirms with one tap.

User case 3: Movie night—ticket purchase

• User prompt: “Two tickets for Avatar 3 at the nearest IMAX tonight.”
• Qwen action: Geocodes location with Amap, filters IMAX screens, selects middle seats, and generates the order.
• User action: Confirms with one tap.

With 400+ operational capabilities, Qwen has transitioned from a Large Language Model (LLM) to a Large Action Model (LAM).

Qwen can complete end‑to‑end transactions reliably and securely because it operates through APIs and unified account data. In addition, one‑tap flow reduces friction and keeps users inside Alibaba’s ecosystem, increasing monetisation potential.

Why the US Can’t Achieve This?

Key question: Why Amazon, Apple or Google not able to build a true Super App that executes end‑to‑end transactions via AI?

The Answer:  A true “Super App” requires three key assets under one roof:

1. Brain — proprietary AI that understands intent and controls actions.
2. Wallet—a payment and settlement system that can complete transactions and handle post‑payment flows.
3. Inventory—direct access to goods and services (retail, travel, food, tickets, etc.).

Alibaba has all three. Its native API integration plus unified identity and payments allow it to complete end‑to‑end flows reliably.

Alibaba (The Benchmark)

• The Trinity: Owns the AI (Qwen) + payment (Alipay) + Inventory (Fliggy/Taobao).
• The Verdict: The Only “True Agent.” It commands the entire transaction loop.

In the US, the necessary assets are split across different companies with conflicting incentives, creating an API wall and preventing a single firm from becoming the de facto execution layer.

The practical result is that Alibaba can offer a one‑tap agentic experience today, while US agents remain constrained by fragmented ownership and commercial frictions.

Amazon — The shopping assistant

• Strength: Owns retail inventory and has Amazon Pay.
• Gap: Limited service inventory (travel, local on‑demand transport) and Amazon Pay behaves like a credit‑card wrapper rather than a unified settlement layer.
• Limit: Great for buying commodities, but it struggles to fully automate multi‑step service transactions such as booking flights or hailing taxis.

Apple — The middleman

• Strength: Tight integration across devices and a widely used wallet (Apple Pay).
• Gap: No owned inventory; third‑party apps must opt in to deeper control.
• Limit: Most partners provide only read‑level access to Siri for status queries, not write‑level access for booking or ordering, preserving their app traffic.

Google — The ad trap

• Strength: Unmatched intent data from search and maps.
• Gap: No owned inventory and a structural conflict with its core ad business.
• Limit: If Google’s agent executes transactions directly, it risks cannibalising search ad revenue.

The “Silent Killer” of Vertical Apps

If Qwen becomes the new browser for China, who will lose? We believe vertical super apps such as Meituan (food and services) and Trip.com (travel) are at risk.

How disintermediation works

• Intercept: Users no longer open vertical apps. They tell Qwen, “Book a hotel.”
• Routing: Qwen resolves intent upstream and routes the transaction to Alibaba’s subsidiaries (Fliggy for travel, Ele.me for food).
• Result: The vertical app never sees the user and is effectively disintermediated.

Margin Expansion for Alibaba

In our view, such disintermediation creates margin expansion opportunities for Alibaba.

• Old model: Alibaba spent heavily on marketing and subsidies to acquire users, compressing margins in food delivery and quick commerce.
• New model: If Qwen becomes the default entry point, traffic is generated organically within the ecosystem, lowering customer acquisition cost.
• Impact: Lower customer acquisition cost (CAC) translates into meaningful margin improvement potential across Alibaba’s consumer businesses.

The Regulatory Twist: While Alibaba suffered from China’s antitrust law several years ago, China’s Anti-Unfair Competition Law (Oct 2025), which tightens rules on illegal data crawling and scraping, effectively strengthens Alibaba’s competitive edge.

• Kills the “Aggregator”: Third-party AIs cannot simply scrape Taobao or Fliggy to execute transactions.
• Forces “Native Ownership”: Legal execution requires authorised access via APIs and permissions. As Alibaba owns the data and APIs, Qwen retains native execution rights by default, reinforcing Alibaba’s market position.

The Strategic Balance: Cannibalization vs. Conquest

Like Google, Alibaba faces a trade‑off between cannibalisation and conquest. Qwen can streamline discovery and execution so effectively that it risks displacing Alibaba’s own advertising and value‑added services for its merchants on Taobao and Tmall. If Qwen handles intent and fulfillment directly, its marketing services could see lower yield.

That said, Alibaba appears to accept short‑term yield dilution in pursuit of long‑term share conquest.

• Upstream capture: Qwen acts as a universal interface that intercepts user intent before competitors such as Pinduoduo, Meituan and Trip.com.
• Volume over yield: Capturing higher transaction volume, even at lower monetisation, can be more valuable than preserving high margins on a shrinking traffic base.
• Lifecycle monetisation: Owning the end‑to‑end relationship increases opportunities to monetise across services (payments interchange, financial products, instant commerce, loyalty), partially offsetting lower ad yields.
• Competitive defense: Gaining share today establishes barriers to rivals, creating a durable advantage over time.

Alibaba is prioritising ecosystem control and sustained user engagement over short‑term yield, betting that long‑term economic value from native transactions, payment flows, and cross-sell will exceed the lost yield.

This article is a “periodical publication” for information only and is not investment advice or a solicitation to buy or sell securities. This article does not constitute a “personal recommendation” or “investment advice” under UK FCA regulations. Investing in equities involves significant risk. The author holds NO position in the securities mentioned. There is no warranty as to completeness or correctness. Please do your own due diligence or consult a licensed financial adviser. Please read the Full Disclaimer before acting on any information. Images created with the assistance of Gemini AI.

Article provided by Asia Pulse.