Of Lotus, Netscape, Napster, Friendster... And OpenAI

Welcome to The Sunday Prompt, where we explore the AI-driven shifts redefining work.

The early movers, the AI specialists, have dominated the headlines in AI’s breakout era. Heck, ChatGPT is AI’s Coca-Cola. But their dominance is an illusion.

AI is not a product. It is becoming the fabric of software itself. And in this game, the specialists will not own the future.

Let’s dig in. 👇

Anthropic (makers of Claude) and OpenAI (makers of ChatGPT) laid out their visions last week. National security. Economic transformation. Productivity gains. Infrastructure investments. Export controls. Grand ideas. AI’s first movers have been ruling the headlines for a while now. But are they calling the shots?

Not really. Microsoft and Google are. They have the customers, capital, and distribution that matter. And they have the bundle. OpenAI and Anthropic can keep innovating their asses off, pushing AI’s boundaries, but without deep integration into existing tools and workflows, they remain expensive, niche products.

Employees overwhelmingly prefer ChatGPT and Claude to whatever Microsoft and Google are slinging. Copilot? Just Clippy AI. Still useless. Still pissing everyone off. But it doesn’t matter. Because they have the bundle.

Microsoft proved this in the ‘90s. Netscape built the best browser. But Microsoft gave Internet Explorer away for free, baked it into Windows, and crushed the competition. Even if Navigator was better, it did not matter. Distribution, not quality, decided the outcome.

Today, OpenAI and Anthropic are where Netscape once stood. They are cutting-edge, but their viability depends on API integrations and standalone subscriptions. They are aftermarket bolt-ons in someone else’s machine. Meanwhile, Microsoft and Google are weaving AI into everything: Office, Windows, Search, Workspace, Android, and Cloud.

AI is no longer a standalone product. Soon, it will be indistinguishable from the software it lives inside. It will be ubiquitous. Like a utility. Google and Microsoft will be the electricity companies supplying the grid. And OpenAI? They’re selling batteries out of the trunk of a Prius on El Camino Real.

Bundling does not make the best product. It makes the default product. And the smartest AI will not win. The most seamless will.

Meanwhile, OpenAI and Anthropic are playing an expensive game. Their research and compute costs are massive. OpenAI’s financial outlook is staggering. Reports suggest OpenAI could lose $14 billion in 2025 alone, and $44 billion by 2028.

Profitability? A meme coin fantasy.

And as if the headwinds weren’t strong enough, a new threat has emerged. No matter whose numbers you trust, DeepSeek’s R1 release sent a chill through OpenAI and every closed-source AI shop. A highly capable, open-weight model with performance rivaling GPT-4 at a fraction of the cost is an existential problem for companies already struggling to make the numbers work.

Microsoft and Google, with $200 billion in cash, highly profitable and diversified revenue streams, and entrenched enterprise relationships, can absorb AI costs while expanding their market position. Meanwhile, OpenAI’s partnership with Microsoft, once considered a lifeline, is showing signs of strain. Microsoft is currently testing models from OpenAI’s competitors (including DeepSeek) while developing its own reasoning model in-house, signaling a move away from OpenAI’s expensive, slow-to-scale technology.

No, AI will not be won in the lab. It will be won in the marketplace. Microsoft and Google don’t need the best models. They need models that are good enough, cheap enough, and integrated enough to become the pragmatic choice and therefore the standard.

OpenAI and Anthropic may lead in research, but Microsoft and Google are positioning themselves to own AI’s next chapter. OpenAI may have the mindshare. Microsoft and Google have the moat.

If this email was forwarded to you and you’d like it delivered directly to your inbox each day, subscribe below.