Nobody Wants Your AI Subscription — But Everyone's Getting One Anyway

Something strange is happening in the AI industry. While engineers obsess over benchmarks and developers argue about which coding assistant writes better Python, the business people have already made their decision: everything AI touches becomes a subscription.

Consider the past week alone. OpenAI partnered with Malta to give every citizen a year of ChatGPT Plus — a government literally subsidizing a private company's subscription service. xAI launched Grok Build at $300 per month for "SuperGrok Heavy" subscribers, a pricing tier that sounds like it was named by someone who's never had to explain a budget to a manager. Databricks rolled out GPT-5.5 access for enterprise workflows, naturally gated behind their platform pricing. Even Netflix's new INKubator studio, focused on AI-generated animation, fits the pattern: more content to justify that monthly fee you're already paying.

The subscription model isn't new, but its application to AI represents a fundamental shift in how we're expected to access intelligence itself. We've gone from paying for software licenses to paying for cloud storage to paying for the ability to ask questions and get answers. It's cognitive infrastructure as a service, and the meter is always running.

What makes this particularly interesting is how quickly it's becoming normalized across completely different sectors. Government AI programs look like enterprise AI deals, which look like consumer AI products. Malta's initiative requires citizens to complete an AI literacy course before accessing their free year of ChatGPT Plus — essentially building vendor lock-in at the national level. Whether that's forward-thinking digital policy or an unprecedented public-private entanglement depends on your perspective, but either way, it's happening.

The economics are compelling for the companies. Recurring revenue is king in Silicon Valley, and AI subscriptions offer margin expansion that traditional software never could. Training costs are high but largely fixed; serving costs drop over time. Every additional subscriber is nearly pure profit once you've built the infrastructure. It's why every major AI announcement now comes with a pricing tier you've never heard of.

But there's a tension here that nobody's really addressing. When AI becomes subscription-based infrastructure, what happens to the people and organizations that can't afford the monthly fee? We're not talking about premium features or faster service — we're talking about access to tools that are increasingly necessary for basic participation in the digital economy. The gap between ChatGPT free and Plus isn't just about response limits; it's about who gets to use the AI that writes better, codes faster, and analyzes more effectively.

The open-source community sees this coming. Projects like Asimov's $15,000 humanoid robot kit exist partly as a reaction to corporate control of robotics development. Similar efforts are emerging in AI, though they face the brutal economics of compute costs and training data access. You can open-source a model, but you can't open-source a data center.

Meanwhile, the subscription machine keeps building. Every partnership announcement, every enterprise deployment, every government deal adds another layer to the assumption that AI access is something you rent, not something you own. We're building a future where thinking assistance costs $20 a month — or $300, or whatever the market will bear.

The question isn't whether this model works. Clearly it does, or everyone wouldn't be doing it. The question is what kind of AI ecosystem we want, and whether we're making that choice deliberately or just letting the economics decide for us. Because right now, the answer seems to be: subscribe or be left behind.