The Disposable Interface

Infinite App Store

The primary way we interact with computers is through "apps." In essence, an app is a durable tool you acquire and keep on your phone or desktop. If you want to calculate a mortgage, you get a mortgage calculator app. If you want to play a memory game, you get a memory game app. The entire multi-trillion-dollar App Store and SaaS economy is built on this simple premise: software is a product you own or rent.

Last week, Google proposed a different model. Instead of a toolbox filled with static, pre-built applications, you get a single block of programmable matter that can become any tool you need, for exactly as long as you need it, before dissolving back into nothing (or, more accurately, before you close the browser tab).

The technology is called Generative User Interface (UI), and it represents a quiet, fundamental break with the past. The idea is that the AI model generates not just the content, but the entire user interface itself. As Google's own research paper puts it, this enables " custom interactive experiences, including rich formatting, images, maps, audio and even simulations and games, in response to any prompt."

The results are startling. Ask Google's Gemini 3 for a potato soup recipe, and instead of a block of text, it might generate a custom, single-purpose cooking app, complete with interactive ingredient checklists and timers. Ask it to teach your five-year-old math, and it can spin up a "Little Ballers Math Academy" game on the fly. The UX pioneer Jakob Nielsen has dubbed this "the dawn of cheap, disposable UI."

By unbundling the "tool" from the "app," the physics of internet user behavior changes. Inbound, it transforms consumption: a search for shoes or a flight becomes a bespoke, hyper-personalized storefront generated for that specific moment. Outbound, it democratizes creation: a non-technical user can now spin up a complex interface to interact with the world, effectively bypassing the need for a developer to act as a translator between human intent and software execution.

This isn't just a theoretical lab experiment. In a fascinating—and slightly humiliating—exercise, Google paid human web designers on Upwork up to $130 each to spend three to five hours building custom websites for specific prompts. It then had Gemini 3 perform the same task instantly. While human experts were still narrowly preferred, the AI's generated UI was considered "at least comparable in 44% of cases." When compared to a traditional website from a search result, users preferred the AI's disposable interface a staggering 90% of the time.

Technically, Google is generating web interfaces. But functionally, it is generating software tools. This raises an uncomfortable question for the software industry. If a custom, fully functional application can be generated and discarded in the time it takes to answer a single question, what is the value of an "app" anymore?

The App Store and SaaS business models are built on the economics of scarcity and persistence. You pay a monthly subscription for access to a durable tool that a team of software engineers spent months building. But Generative UI operates on the economics of abundance and disposability, as Google's researchers explain:

Generative UI enables us to spin up an instant (AI-based) product management, UX design, and engineering teams, to build an interactive experience, over the course of a minute, for a specific prompt. While not as competent as human experts, Generative UI enables custom experiences for any prompt.

Entire job functions, from product management to UX design, are being compressed into a single API call. The "app" is being atomized, broken down from a product into a momentary service.

This will be the next phase of the internet's evolution. We are moving from a finite library of applications you install to an infinite catalog of interactions you generate. The last era was about building software. This next one might be about forgetting it as soon as you’re done.

More on AI Adoption:

• Microsoft faces uphill climb to turn enterprise dominance into widespread AI chatbot adoption (CNBC)
• Enterprise AI Monetization: 2025 Scorecard (ARPU)

On Our Radar

Our Intelligence Desk connects the dots across functions—from GTM to Operations—and delivers intelligence tailored for specific roles. Learn more about our bespoke streams.

Amazon's Midwest Power Play

• The Headline: Amazon is investing $15 billion to build new data center campuses in Northern Indiana, adding 2.4 gigawatts of capacity to support surging AI demand for its cloud services. (Reuters)
• ARPU's Take: The AI infrastructure race is now a nationwide scramble for the one resource that matters most: massive, reliable power. By striking a deal directly with the local utility to fund new power infrastructure, Amazon is proactively solving the number one bottleneck to AI expansion, planting a major flag in the American heartland.
• The Operations Question: For competitors like Microsoft and Google, this establishes a new operational imperative. The focus shifts from simply acquiring land to developing complex, capital-intensive partnerships with utility providers to secure and fund multi-gigawatt energy access, as grid availability is now the primary bottleneck to growth.

The Bedrock Bottleneck

• The Headline: A leaked internal document reveals Amazon's flagship AI service, Bedrock, suffered "critical capacity constraints" and performance issues over the summer, causing it to lose tens of millions of dollars in revenue as major customers shifted workloads to rivals like Google Cloud. (Business Insider)
• ARPU's Take: This leak confirms that AWS was caught flat-footed by the surge in AI demand. The problem was twofold: they didn't have enough capacity for customers, and the product itself was underperforming on key metrics like latency. This created an opening for rivals like Google Cloud to poach major, high-profile customers.
• The GTM Question:  The AI platform war is a battle of execution, not just technology. For AWS, the loss of key customers due to preventable capacity and performance issues creates an urgent imperative to overhaul its AI go-to-market and infrastructure planning to avoid ceding its market leadership. For rivals, it demonstrates that operational excellence and supply availability have become powerful competitive weapons that can successfully pry away even the most entrenched enterprise accounts.

P.S. Tracking these kinds of complex, cross-functional signals is what we do. If you have a specific intelligence challenge that goes beyond the headlines, get in touch to design your custom intelligence.

You received this message because you are subscribed to ARPU newsletter. If a friend forwarded you this message, sign up here to get it in your inbox.