# AI costs explained: why 'tokenmaxxing' is ending and cheaper models are winning

> Companies like Uber and Meta are rationing AI use and switching to cheaper Chinese open-weight models.

*Companies spent two years using as much AI as possible — now the bills have landed. Here's what changed, why cheaper Chinese models are winning business, and what it all means for you.*

By The SuggestedTech Team · SuggestedTech
Canonical: https://suggestedtech.com/news/ai-costs-efficiency-shift-explained

For the past two years, the unofficial motto inside a lot of AI-hungry companies was simple: use as much of it as you possibly can. Staff were nudged, and sometimes openly rewarded, for running every task through the priciest, most powerful model going. There was even a nickname for it — 'tokenmaxxing'. In late June 2026, CNBC reported that the mood has flipped. The same firms that were egging everyone on are now quietly counting the cost, and some are switching to far cheaper models made in China. Here's what's happening, in plain English, and why it actually matters to you.

> **Info:** **In plain English.** A *token* is just a small chunk of text — roughly three-quarters of a word. AI models read and write in tokens, and companies pay per token, a bit like paying for a phone call by the second. *'Tokenmaxxing'* was the habit of burning through as many tokens as possible on the assumption that more AI always meant more value. The monthly bill eventually said otherwise.

## Why the 'use all the AI you want' era is ending

The trouble with pay-per-token is that costs can run away from you. If your team leans on a top-tier model for everything — drafting emails, summarising documents, writing code, running automated 'agents' that loop over a task dozens of times — the meter never stops. And because the labs behind the leading American models, OpenAI and Anthropic, have been slower to cut their prices lately, that meter kept spinning at full rate. CNBC reported that big names including Uber, Microsoft, Salesforce and Meta have started rationing how much advanced AI their employees can use, because the pay-per-token model turned out to be 'more expensive than it's worth'.

> Companies including Uber, Microsoft, Salesforce and Meta have moved to ration employees' use of advanced AI, as the token-payment model favoured by Anthropic and OpenAI proved 'more expensive than it's worth' — a broad shift in mood from 'tokenmaxxing' to efficiency.
> — [CNBC](https://www.cnbc.com/2026/06/26/openai-anthropic-new-ai-spending-reality-as-users-shift-to-efficiency.html), 2026-06-26

Uber's story is the one that makes the point stick. Its chief technology officer, Praveen Neppalli Naga, said the company blew its entire annual AI budget in just four months, according to The Information. Uber's fix was to bring in spending tiers — a baseline of around $1,500 a month per employee, with more available if you can justify it. Think of it like a company mobile plan: you get a generous allowance, and if you want to go over it, you have to explain why. It isn't banning AI; it's asking the question that got lost in all the excitement — is this spend actually paying for itself?

## The cheaper alternative: 'open-weight' models from China

The other half of the story is where the money is going instead. A growing number of companies are switching to 'open-weight' models — and many of the best-value ones right now come from China. An open-weight model is one whose 'brain' (the trained parameters, or *weights*) you can actually download. Instead of renting access through someone else's cloud and paying per token forever, you can run the model on your own servers. That matters for two reasons: it can be dramatically cheaper at scale, and your data never has to leave your building.

Take Lindy, a roughly 25-person startup that builds AI agents. Its chief executive, Flo Crivello, moved 100% of the company's traffic off Anthropic's Claude and onto DeepSeek, a cheaper Chinese open-weight model. He told CNBC the cost curve 'crash[ed] to the ground', said he expects to save millions within months, and bluntly called the switch 'a matter of survival'. One analysis put the saving on the migrated work at around 90%. He'd happily switch back if American prices fall — 'until then, we've got options'. And the scale of the move shows up plainly in figures from OpenRouter, a popular service that routes app requests to whichever model you pick:

| What the numbers show | Then | Now |
| --- | --- | --- |
| Chinese models' share of top-10 token use (OpenRouter) | under 1.2% (late 2024) | ~61% (mid-2026) |
| The same share a few months earlier | ~51% (April 2026) | ~61% (June 2026) |

In plain terms: of all the tokens flowing through the ten most-used models on that platform, the Chinese-built ones — **Qwen** (Alibaba), **DeepSeek**, **Kimi** (Moonshot), **GLM** (Zhipu) and **MiniMax** — now account for roughly 61%, about six in every ten. Under two years ago it was a rounding error.

> US startups are increasingly building *core* products — code generation, autonomous agents — on Chinese open-weight models rather than just kicking the tyres, drawn by lower cost, competitive capability and the freedom to run the models on their own hardware.
> — [CNBC](https://www.cnbc.com/2026/06/30/white-house-ai-china-crackdown.html), 2026-06-30

## What this all means for you

If you use AI tools day to day — or you're the one paying for them — this shift is mostly good news. More competition between models means downward pressure on prices, and you're already seeing it: Anthropic's new Claude Sonnet 5 launched at a cut-price $2/$10 per million tokens, and OpenAI previewed a cheaper 'Terra' tier, both aimed squarely at this pressure. More choice means you're not locked into a single provider. And the return of the ROI question — *does it actually pay off?* — is healthy: it pushes tools to earn their keep instead of dazzling you with horsepower you don't need.

There's a sensible flip side worth keeping in mind, though. Cheaper isn't automatically better for every job — a bargain model that gets things subtly wrong can cost you more than it saves. Running an open-weight model yourself takes real engineering muscle, so it's not a switch most individuals or small teams will flip themselves. And the tug-of-war between American and Chinese labs is tangled up with export controls and politics, so today's cheapest option might not stay the cheapest. The practical takeaway is refreshingly old-fashioned: pick the tool that does your job well for the least money, check now and then that it's still the best deal, and don't pay for horsepower you're not using.

> **Key:** **The one-line version:** the 'spend without thinking' phase of AI is over. Companies are rationing usage and shopping around, cheaper open-weight models (many of them from China) are winning real work, and the likely winner from all this price competition is you — as long as you keep asking whether a tool actually earns its cost.

## Key takeaways

- Big companies including Uber, Microsoft, Salesforce and Meta are now rationing staff AI use because the pay-per-token model got 'more expensive than it's worth', CNBC reported.
- Uber brought in spending tiers starting around $1,500 a month per employee after it burned through its entire annual AI budget in just four months.
- AI-agent startup Lindy moved 100% of its traffic off Anthropic's Claude to China's DeepSeek, expects to save millions, and called it 'a matter of survival' — one analysis put the saving at about 90%.
- On the OpenRouter platform, Chinese open-weight models are now about 61% of token use among the top 10, up from under 1.2% in late 2024.
- For you it means cheaper AI tools, more choice of models, and the return of a healthy question: does this actually pay off?

## FAQ

### What does 'tokenmaxxing' mean?
It's the recent habit of encouraging staff to use as much AI as possible — always reaching for the priciest, most powerful model — on the assumption that more usage meant more value. Because you pay per token (a small chunk of text), the bills ballooned, and CNBC reported in June 2026 that companies are now swinging the other way, towards efficiency.

### What is a token, and why did AI bills get so big?
A token is roughly three-quarters of a word; models read and write in tokens, and you're charged per token used. Costs exploded because teams ran everything through top-tier models and automated 'agents' that loop over tasks many times — so the meter never stopped. Uber reportedly burned its entire annual AI budget in four months before capping spend at tiers starting around $1,500 a month per employee.

### What are open-weight models, and why are Chinese ones like DeepSeek cheaper?
An open-weight model is one whose trained 'brain' (its weights) you can download and run on your own servers, rather than renting it through someone else's cloud and paying per token forever. Chinese models such as DeepSeek and Qwen combine that openness with low cost and strong capability, which is why startups like Lindy have moved core work onto them — Lindy expects to save millions after cutting its costs by an estimated 90%.

### What does the OpenRouter 61% figure actually mean?
OpenRouter is a service that routes app requests to whichever model a developer chooses, so its data is a useful snapshot of what people actually run. Of the tokens flowing through the ten most-used models on the platform, Chinese-built ones (Qwen, DeepSeek, Kimi, GLM, MiniMax) now make up about 61% — up from under 1.2% in late 2024 and roughly 51% in April 2026. In short, a huge amount of real AI work has shifted to cheaper Chinese models fast.

### What does this cost shift mean for me?
Mostly good things: more competition is pushing prices down (Anthropic's Sonnet 5 launched at a discounted $2/$10, and OpenAI previewed a cheaper 'Terra' tier), and you get more choice of provider. The honest caveats are that cheaper isn't always better for every task, running open-weight models yourself needs real engineering, and politics could reshuffle who's cheapest. The smart move is old-fashioned: pick the tool that does your job well for the least money, and keep checking it still pays off.

## Sources

- [OpenAI and Anthropic face new AI reality as users shift from 'tokenmaxxing' to efficiency](https://www.cnbc.com/2026/06/26/openai-anthropic-new-ai-spending-reality-as-users-shift-to-efficiency.html) — CNBC, 2026-06-26
- [White House AI crackdown opens door for Chinese model makers to close gap](https://www.cnbc.com/2026/06/30/white-house-ai-china-crackdown.html) — CNBC, 2026-06-30
- [Palantir's Karp bashes token-based AI model as 'completely wrong'](https://www.cnbc.com/2026/07/01/palantir-karp-open-ai-anthropic-tokens.html) — CNBC, 2026-07-01