[Two Cents #86] “Flights of Thought” on Consumer + AI — Part 9: Consumer Bahavioral Shifts — 3. AI Super App: $1T battle for ‘Intent Capture”
Introduction
One of the most important battlegrounds in Consumer AI is quietly forming around the race to become the AI Super App.
This isn’t a contest for the best model, the smartest assistant, or the slickest UI. It’s a structural competition over where consumer intent is captured, interpreted, and executed in the AI era.
We’re already seeing the race unfold across multiple fronts: AI assistants (ChatGPT, Gemini, Claude, Perplexity), AI-native browsers (Atlas, Comet, Dia), agentic platforms (Manus, Genspark), OS-level intelligence from Apple and Google, and new entrants that may emerge with entirely new form factors. It’s still early, but the winner of this race has a real chance to become the dominant consumer platform of the AI era—what people will call the “Google of the AI age.”
China offers a useful precedent. Over the past decade, it already lived through a prolonged battle for mobile Super App dominance, which ultimately converged into a structure where a small number of platforms controlled most consumer time, transactions, and value creation. Today, China is entering a second platform war—this time for the AI Super App.
The U.S. will not replicate China’s outcome 1:1. But the economic logic, competitive behaviors, and distribution dynamics observed in China are highly instructive for understanding how the AI Super App race in the U.S. is likely to play out.
Why the AI Super App matters
The core function of an AI Super App is intent capture. That’s where the largest value capture tends to happen: the platform that understands intent most accurately—and responds most effectively—ends up owning the downstream flows.
History supports this. In every major computing shift, the platform that captured intent earliest captured disproportionate economic value:
Web era: search engines became the starting point for information and commerce.
Mobile era: app stores controlled distribution and monetization by controlling access to apps.
Social era: feed-based recommendation systems (Facebook, TikTok) captured attention at scale—and each translated that control into ~$1T outcomes.
As discussed as the transition from the “Attention Economy” to the “Intent Economy” in [Two Cents #84] , the internet and mobile ecosystems historically struggled to understand user intent directly. The industry therefore relied on the best available proxy: the flow of user attention. Metrics like traffic, CTR, conversion, and A/B testing became the primitives, and the resulting businesses were ad networks, affiliate networks, and performance marketing—collectively, the Attention Economy.
AI changes the form factor of intent expression itself.
Users no longer communicate intent indirectly through keywords, dwell time, or click behavior. They state it directly through prompts and context: “Plan a five-day trip to X with a budget of Y,” or “Recommend a dress for a weekend party within $Z that matches this vibe.”
The interface that becomes the default starting point for these requests gains leverage over everything downstream—services, workflows, commerce, and monetization.
That’s why the AI Super App can become a $1T+ consumer opportunity. It sits above vertical services as a controlling layer: it intermediates across categories, routes execution, and effectively collects a toll on the value created along the way. China’s Super App history already shows that once this layer consolidates, value concentration accelerates quickly.
How China built the Super App and Mini-App ecosystem
China’s Super App ecosystem didn’t appear overnight from a single breakout product.
It emerged as platforms combined (1) massive consumer distribution, (2) durable consumer connection via social channels, and (3) integrated payments—then used those ingredients to internalize the entire “intent flow,” from intent expression to delivery, inside the Super App container.
WeChat is the canonical example. It began as a messaging product, then expanded step-by-step into content (Official Accounts), payments (WeChat Pay), and offline acceptance via QR codes. In 2017, Tencent formalized what had been forming organically by introducing Mini Programs—lightweight “no-install” apps running inside WeChat—turning these components into a coherent platform.
Precursors existed before Mini Programs: HTML5 pages in in-app browsers, QR-driven O2O flows, and service accounts that could interact with users. Mini Programs unified those pieces into a standardized platform, allowing WeChat to internalize the consumer intent flow—and capture most of the value created along that flow.
The result was unambiguous. As of 2024:
~936M MAU engaging with WeChat Mini Programs
1M+ Mini Programs live
RMB 2T+ quarterly GMV flowing through the ecosystem (>$1T annualized)
Mini Programs didn’t just complement the app ecosystem; they redefined it. Many services no longer needed standalone apps. Customer acquisition costs dropped sharply. WeChat evolved from a portal into a massive closed-loop transaction infrastructure.
Other Chinese platforms—Alipay, Douyin (ByteDance), Meituan—replicated the model in their own domains. The end state was not a single monopoly, but a multi-polar oligopoly: a small number of Super Apps, each anchored in a core consumer habit, each internalizing the full intent flow within its sphere.
Common success factors among China’s Super App winners
China’s Super App winners didn’t win by packing in more features. They won by positioning themselves at the earliest point of consumer intent, then pulling the entire downstream intent flow inside the platform. This created a compounding structural advantage: value creation and value capture strengthened over time.
1. Owning distribution
Chinese Super Apps became the default entry points for everyday actions—messaging, paying, scanning, searching, consuming content—effectively controlling the top of the funnel across categories. The critical point wasn’t any single feature; it was establishing a default behavior: “If I want to do something, I open this app first.”
WeChat illustrates this clearly. By stacking payments, content, offline QR flows, and service access on top of messaging, it reduced the need for users to leave the platform. Once “the service lives inside WeChat” became internalized, WeChat’s coverage and value capture expanded rapidly.
At the ecosystem level, this created a structure where new services could not bypass the Super App. Consumer access costs (CAC) converged inside the container, and standalone apps became increasingly disadvantaged in marketing and distribution. The Super App became the market’s gate.
2. Embedding trust anchors
The decisive factor that allowed Super Apps to move beyond discovery into execution was embedding trust infrastructure—especially payments and identity. Payments aren’t just a feature; they are the anchor that connects intent to transaction with minimal friction.
WeChat Pay and Alipay tied together real-name identity, financial accounts, and social context. Users could transact immediately without needing to rebuild trust for each new service. Conversion speed inside the Super App accelerated dramatically.
This collapsed the boundary between “finding” and “doing.” Search, recommendation, and comparison increasingly led directly to payment. The Super App stopped being a traffic intermediary and became the transaction infrastructure itself—leaving far less room for value to remain outside the container.
3. Minimizing developer and merchant friction
Mini Programs were the ecosystem’s key structural innovation. The point wasn’t just new functionality; it was removing friction across distribution, installation, login, and payment.
Developers and merchants no longer needed to fight app-store review processes, downloads, updates, or re-engagement challenges. They could ship instantly inside WeChat’s runtime and leverage built-in distribution through search, QR codes, sharing, and location-based discovery.
This created an ecosystem where long-tail services could exist economically. Many services that couldn’t justify standalone apps could thrive inside the Super App, while the platform strengthened its value capture by becoming the infrastructure layer for all activity.
4. Maximizing habit stacking
Perhaps the most powerful weapon was that Super Apps did not try to invent new behaviors. They layered minimal additional actions on top of existing habits.
Messaging flowed into payment; payment flowed into booking; booking flowed into content—naturally, without users feeling like they were learning “new products.” The Super App expanded the user’s behavioral surface area with very little incremental cognitive cost.
Over time, the Super App became less like an application and more like an OS for daily life. The lock-in was not to a single service, but to the platform itself—and that lock-in strengthened as the ecosystem grew.
The outcome: a power-law oligopoly, not a single monopoly
These forces did not produce a single winner-take-all monopoly. They produced a power-law oligopoly, where a small number of Super Apps (WeChat, Alipay, Douyin, Meituan) captured most consumer value, each anchored around a core habit.
The key lesson is that the battle was not feature-driven. The winners were the platforms that owned the intent entry point and internalized the entire intent flow—creating structural advantages that compounded over time. The same framework applies directly to the AI Super App era.
Act II: China’s AI Super App war
China is now replaying this dynamic in the AI era.
The dominant Super Apps are moving aggressively to become AI Super Apps as well. Incumbents like Tencent are embedding AI directly into WeChat’s core surfaces—search, chat, and workflows—sometimes integrating third-party models. It’s the classic incumbent play: defend the container, modernize the surface.
At the same time, challengers such as ByteDance are attacking more aggressively. Doubao is leveraging Douyin’s attention graph and creator ecosystem to open a new AI front door, while expanding into devices to capture consumer intent more directly. Competition is already turning hostile—e.g., ByteDance’s mobile AI agent app reportedly faced rapid blocking by platforms like WeChat and Taobao within days of launch.
Once again, the same principle holds: the AI Super App race is being driven less by “AI capability” and more by control of distribution.
The AI Super App race in the U.S.
What’s similar to China
Despite surface differences, the underlying economics are similar. In both markets, the central prize is intent capture at scale. If you control intent, you sit upstream of decisions and actions—and gain leverage over everything downstream. That leverage is what translates into outsized value capture.
Both markets also tend to converge toward oligopoly at the consumer layer. Even if many services exist beneath, durable moats are built at the top—through ecosystems that form around the dominant container.
For that reason, the AI Super App race in Consumer AI is plausibly a $1T competition, comparable in scale to what Google represented in the web era.
The structural differences that matter
The U.S., however, is structurally different in ways that materially shape how the market can form.
China entered the AI era with entrenched Super Apps controlling messaging, payments, and identity. The U.S. consumer ecosystem is more fragmented—payments, messaging, and commerce are split across many players.
As a result, OS-level gatekeepers (Apple, Google, Microsoft) exert far more influence. Defaults, permissions, and background execution are powerful constraints on how an AI Super App can emerge. Apple’s App Store, Siri, and default search placements are obvious choke points. Even Safari’s default search placement alone represents meaningful economic value (on the order of tens of billions in revenue-equivalent economics over time).
Privacy and regulatory expectations are also higher. That can slow automation, but it increases the value of platforms that can provide trust, auditability, and governance.
These factors make a single WeChat-style monopoly unlikely in the U.S. But they do not eliminate power-law outcomes. Instead, they push the market toward a multipolar structure, where dominance is distributed across contexts: OS surfaces (Siri), default search boxes, app stores, AI assistants, AI browsers, and more.
Signals the U.S. is entering its “mini-app moment”
Recent moves suggest the U.S. is entering its own version of a mini-app moment.
With the introduction of an App SDK and app-store-like distribution, ChatGPT is evolving from an assistant into a platform. Natural-language intent can be routed into third-party execution layers, while discovery and monetization remain controlled upstream by the platform.
The expansion of skills and agent libraries in systems like Claude and ChatGPT points in the same direction. Over time, these modules could function like Mini Programs—allowing AI platforms to aggregate and monetize specialized tasks.
Separately, AI-native browsers (Atlas, Comet, Dia) and agentic platforms (Manus, Genspark) are attempting to capture intent at the moment of navigation. By embedding agents into browsing workflows, they challenge the browser as the primary unit of interaction.
Across these efforts, the direction is consistent:
Own the layer where intent is expressed and action begins.
Potential scenarios for the U.S. AI Super App market
The U.S. market may evolve through several paths:
AI assistant–first platforms (ChatGPT, etc.): become the default interface and capture value via intent routing and app marketplaces
OS-native AI (Apple, Google): dominate through defaults and system permissions and extract rents from AI workflows
Social AI (Meta): capture intent through social context inside feeds and messaging, monetizing via creators and commerce
Enterprise AI spillover (Microsoft): control high-ARPU intent through workflows, identity, and governance
Below is a deeper look at each.
1. Assistant-first platforms (ChatGPT, etc.)
The core value capture is intent interpretation + intent routing.
Natural-language requests become the transaction starting point, and the layer that decides which service/agent/tool fulfills that intent gains economic power. Where the App Store controlled app distribution, the AI assistant controls intent distribution.
If this path becomes dominant, the market forms as an AI-native agent/app marketplace. Vertical agents and tools compete for discovery and execution inside the assistant. The platform captures take rates through recommendation logic, distribution control, payments, and identity. Structurally, it resembles a unified layer combining search, app distribution, and payments.
Strengths: fast iteration, model-UX coupling, and relative platform neutrality.
Weaknesses: distribution fragility—lack of direct control over OS/browser defaults. The biggest difference vs. China is that payments, identity, and daily-life infrastructure are not pre-integrated, so the product may look more like a “Super Gateway” than a true WeChat-style Super App.
If this scenario leads, Consumer AI becomes agent-first and API-first. Standalone consumer brands weaken, and execution agents capture value downstream while the assistant controls routing upstream.
Key contenders: OpenAI (ChatGPT), Google (Gemini), Anthropic (Claude). The most likely leader is OpenAI, but OS dependence remains a persistent risk.
2. OS-native AI (Apple, Google)
The value capture here is defaults + system permissions + background execution.
Intent is detected and acted on at the OS layer, often without explicit user initiation. The economic outcome is less about ads and more about extracting workflow rents—similar in spirit to app-store economics.
In this world, the “Super App” is not a separate app; the OS becomes the Super App. Siri/Gemini are interfaces, but the real advantage is control over notifications, permissions, sensors, payments, and identity. The market forms as an AI-enabled workflow layer over the existing app ecosystem.
Strengths: unmatched distribution and lock-in.
Weaknesses: slower innovation and constrained openness; stronger regulatory and privacy pressure than China. Apple, in particular, may face limitations on how directly it can push into commerce-like value capture.
If this scenario dominates, Consumer AI becomes OS-centric and incumbent-friendly. New entrants become extensions, plug-ins, or infrastructure providers rather than top-layer interfaces.
Key contenders: Apple, Google. Long-term, Apple may have the strongest position—but the outcome is closer to an “AI-powered OS” than a classic AI Super App.
3. Social AI (Meta, TikTok)
The value capture is intent capture through social context.
Many consumer desires are socially formed: taste, identity, validation, and trend-following. Platforms that sit inside social graphs can capture intent earlier and more richly than systems that only see explicit prompts.
An AI Super App here likely manifests as ambient AI inside feeds and messaging: the system proactively suggests, curates, and routes commerce through creators and advertising. The market becomes a unified arena where creators, brands, and AI agents compete inside the same attention surface.
Strengths: massive social graphs and proven ad monetization engines.
Weaknesses: trust and execution. China’s Douyin integrated deeper into payments, logistics, and O2O; Meta is stronger in discovery and influence than end-to-end execution. Regulatory exposure is also significant.
If this scenario leads, Consumer AI evolves into influence-driven commerce + AI recommendation, strongest in low- to mid-consideration decisions rather than high-stakes, high-complexity tasks.
Key contenders: Meta, TikTok. This is less a “Super App” than an “AI-augmented social OS.”
4. Enterprise AI spillover (Microsoft)
The value capture is workflow, identity, and governance.
This path captures intent not from consumer leisure but from work. Copilot may appear consumer-like, but enterprise contracts remain the economic center.
The “Super App” here becomes a work-life crossover platform. Habits formed in Outlook, Teams, and Windows extend into personal contexts, sustaining a high-ARPU structure. The market forms as a SaaS + consumer hybrid.
Strengths: durable revenue, enterprise lock-in, and strong governance/audit capabilities.
Weaknesses: weaker consumer-native distribution and cultural gravity. China has no true analogue here—this is a distinctly U.S.-shaped pathway.
If this scenario strengthens, Consumer AI becomes a two-tier market: a high-ARPU enterprise-driven layer and a separate, lower-ARPU pure consumer layer.
Key contender and likely winner: Microsoft—though it’s unlikely to become the universal consumer interface.
Multipolar equilibrium most likely
The most likely end state is not a single universal winner but a multipolar equilibrium, where power-law dominance emerges within contexts.
ChatGPT could own intent routing; Apple/Google could control execution infrastructure; Meta could dominate social discovery; Microsoft could control high-value workflows. The “AI Super App” outcome may be a set of super layers rather than one app.
The AI Super App race: a new $1T opportunity
The AI Super App race is not fundamentally about building “better AI.” It’s about owning the default interface where intent is captured and converted into execution. This is not a product competition—it’s a market-structure competition that rewires where traffic and revenue flow.
The winners won’t just dominate a single vertical. They will control the top layer that interprets intent, routes execution, and captures economics across categories. In that sense, the AI Super App is less a category and more a defining layer that will shape how value accrues in Consumer AI over the next decade.
China’s Super App history makes one point clear: once a platform becomes the default destination for consumer intent, discovery, payments, and supply all get pulled into the container—and value capture consolidates faster than most expect.
The U.S. is unlikely to converge into a single WeChat-like monopoly. But the same economic gravity still applies: intent capture drives value capture. The structure may differ, but the rule remains.
The key implication is that founders and investors must focus less on surface-level products and more on market structure. The right question is not “Who has the best demo?” but “Where will the choke points and toll booths form in the intent flow?”
For founders, the opportunity is not necessarily to build the AI Super App directly, but to find the indispensable positions in the stack—where intent must pass, where execution must be verified, where trust must be anchored. For investors, the bet is less about short-term traction and more about the layers that compound value over time.
Every major technology wave includes a moment where control of the consumer layer is renegotiated. Web search produced Google. The AI Super App race is that moment for AI—and it will define where the next $1T of consumer value is created and captured.
We are still in the phase where standards and rules are being written. The goal isn’t to declare a single “answer,” but to understand the direction of change and explore new opportunities along that direction.
I hope this framework helps as a starting point for imagining—and defining—meaningful opportunities in the next Consumer AI cycle.