The AI Price War Has Started. Here's Why Cheaper Won't Help.

If you run anything on AI, you'll have seen the headlines: OpenAI is reportedly weighing significant cuts to what it charges for tokens, anticipating that Anthropic will move the same way. Google has already trimmed its consumer pricing. And in the background, a wave of Chinese open-source models is doing comparable work for a fraction of the price — in some cases a tenth or less.

The obvious read is that this is great for buyers. Cheaper tokens, lower bills, everyone wins.

I'd be careful with that. A price war doesn't make your AI spend easier to manage. In a lot of cases, it makes it harder. Here's why.

Cheaper per token isn't cheaper full stop

This is the trap, and it catches almost everyone. Your AI bill isn't high because tokens are expensive. It's high because usage is unmanaged.

There's a word doing the rounds in Silicon Valley for this — "tokenmaxxing" — burning through tokens at volume with nothing measurable to show for it. Uber reportedly went through its entire 2026 AI budget by April. Analysts have started warning that corporate AI bills are simply unsustainable. None of that is a unit-price problem. It's a consumption problem.

And here's the bit that should worry you: cut the price per token and consumption tends to go up. Cheaper feels like permission to use more. So the bill stays exactly where it was, or climbs — you're just paying less for far more of it. Halving the price of something you don't measure doesn't save you money. It removes the one bit of friction that was keeping a lid on it.

You're now budgeting against a moving target

Until recently, AI pricing was reasonably stable. You could set a model, work out a rough cost, and plan around it.

That's over. Prices are about to start moving month to month — and not gently. These cuts aren't being made because the maths suddenly works. They're being made as weapons, by two companies losing billions and racing each other to public markets, each trying to pull customers off the other. A rate set to win a war is not a rate you can build a budget on. It can move again the moment the strategy shifts.

So the business that picked a provider eighteen months ago and stopped looking is now flying blind through the most volatile pricing period AI has had. Cheaper, maybe. Predictable, no.

The temptation to switch has its own bill

When a rival appears at a fraction of the cost, the instinct is to jump. And sometimes you should — routine work especially doesn't need a flagship model.

But switching isn't free. Re-testing, re-integrating, and re-validating output all cost time. And the cheapest provider today carries its own risk: models get deprecated, access gets pulled, terms change. Chasing the lowest headline rate across a dozen providers is how you end up with spend scattered everywhere and no idea what any of it is doing. You can trade a high bill for an unmanageable one.

The only thing that actually helps

You can't ride a price war you can't see. Every move that helps you — switching the right workloads to a cheaper model, capping the wrong ones, catching a provider's price change before it hits the invoice — depends on one thing: knowing what you're actually using, per model, per team, per workflow.

That's the unglamorous truth under all of this. The businesses that come out of the price war ahead won't be the ones who waited for the lowest rate. They'll be the ones who could see their own usage clearly enough to act the moment the numbers moved.

Two practical steps. First, watch the prices — I built a live token-price page so you can see what the major providers charge in one place, side by side, as it changes. Second, watch your own usage. That's what SpendLil is for: it tracks every AI request across your business so that when prices move, you already know exactly which workloads to move with them.

And while you're at it, make sure you're not overpaying regardless of the rate — the simple optimisations that cut your bill work at any price per token.

The bottom line

Cheaper tokens are coming, and that part is genuinely good news.

But cheaper only helps the business that knows what it's buying. A price war rewards visibility, not patience. If you can see your spend, falling prices are a gift. If you can't, they're just a faster way to lose track of a bigger number.

Watch the prices. Watch your usage. Then let the war work in your favour.