London, 2014. A black cab to Heathrow ran £70+. Uber charged £25. We all knew, somewhere in the back of our minds, that the maths didn’t add up—that venture capital was paying the difference. We took the ride anyway.

A decade later, Uber charges airport surcharges and takes a cut up to 50%, at times becoming even more expensive than taxis. The economics caught up. They always do.

I’ve been thinking about this pattern as I experiment with generative AI. The tools feel almost free. OpenAI distributes credits generously. Anthropic, perhaps the most “expensive” of all, bundles usage into accessible tiers. Google subsidises Gemini through search. For individual hobbyists and marketing teams testing content automation or personalisation at scale, the marginal cost of experimentation appears close to zero.

It isn’t. And—here’s the part that matters—it won’t stay that way.

Focus On: The subsidy and the meter

The price is wrong (for now)

Let me be clear: the cost decline is real. Independent benchmarks show inference costs for equivalent capability dropping by roughly 10× per year. Competition among foundation model providers remains fierce. No one wants to cede market share during the adoption sprint.

This creates a genuine sensation of abundance. Marketing teams spin up experiments without triggering procurement scrutiny. The CFO hasn’t asked hard questions yet because the line items are small.

But IBM’s research suggests enterprise compute costs are tracking toward an 89% increase between 2023 and 2025. Stanford’s AI Index finds that realised savings from AI adoption are frequently modest—often below 10%. How do you reconcile “AI is cheap” with “our AI bill keeps climbing”?

Volume. Variance. Duplication. Every team runs its own pilots. Every pilot consumes tokens. Every token multiplies across the organisation. The subsidy masks the volume; it doesn’t eliminate it.

I’ve seen this film before

Uber’s playbook was explicit. Internal documents describe the strategy as “burning the burn”—a deliberate phase where losses funded market capture before pricing discipline arrived. Subsidise supply, buy demand, extract margin once the network locks in.

AI vendors are following the same arc, just faster. The shift from seat-based to usage-based pricing is already underway across enterprise SaaS. Reasoning models and agentic workloads carry higher inference costs; vendors are quietly aligning revenue with marginal compute. More meters. More tiers. More overage clauses buried in the fine print.

The free ride won’t just end with a price-increase announcement. It will also come through contract structure.

So what’s the move?

Here’s where the thinking gets counterintuitive. If AI pricing will tighten—and I believe it will—the rational response isn’t to wait. It’s to consume disproportionately now.

But consume what, exactly?

Not tokens for their own sake. The real prize is organisational learning. Promptcraft is trivial; anyone can write a decent prompt in an afternoon. Process redesign, measurement frameworks, governance structures—those take quarters to build. The companies that thrive when the meter arrives will be those that used the subsidy era to discover which workflows actually generate measurable value per thousand tokens spent.

A suggestion: treat 2026 as an experimentation portfolio. For each use case—lead qualification, content variation, sales enablement, research synthesis—force yourself to quantify the value created against tokens consumed. When pricing tightens, only workflows with demonstrated unit economics survive the budget review.

Marketing is the natural proving ground here. High-volume text generation. Rapid experimentation cycles. Immediate performance feedback. If any function can develop disciplined “AI P&L” thinking before the CIO mandates it enterprise-wide, it’s ours.

Here’s the question I keep coming back to: if your AI costs doubled next year, which workflows would you defend in the budget meeting—and which would you quietly delete within 48 hours?

If you can’t answer that today, the subsidy is buying you time. Use it wisely.

Follow me

To keep up with the latest in generative AI and its relevance to your digital transformation programs, follow me on LinkedIn or subscribe to this newsletter.

Disclaimer: The views and opinions expressed in Chronicles of Change and on my social media accounts are my own and do not necessarily reflect the official policy or position of S&P Global.