What happened
Chinese AI models are undercutting American AI companies by as much as 90% on price, and it's already reshaping who businesses trust to run their day-to-day work. NYU professor Scott Galloway and journalist Josh Tyrangiel recently laid out the numbers on a podcast, pointing to research from Juniper Research showing that American AI labs' share of global usage dropped from roughly 70% to 30% in a single year. That's not a slow decline. That's a landslide.
Here's the plain-English version: running an AI model costs money, mostly for the computing power behind it. US companies like OpenAI and Anthropic charge prices that reflect those costs plus a profit margin. Chinese labs — names like DeepSeek, Qwen (from Alibaba), and Kimi (from Moonshot AI) — are charging a fraction of that for models that, on many everyday tasks, perform close enough that most users can't tell the difference.
Galloway's argument is that the discount itself isn't really the headline. The real story is what happens after companies build their workflows around the cheap option. Once a business rewires its operations, its software, and its habits around a supplier, switching back gets expensive and painful — even if that supplier eventually raises prices. It's the same undercut-then-lock-in strategy China has used in infrastructure deals across Africa and Latin America, just applied to software this time instead of ports and railways.
Why is the price gap so big? Part of it is subsidized development — Chinese tech giants and state-backed funds are willing to lose money for years to win market share. Part of it is a technique called distillation, where a smaller, cheaper model is trained to copy the behavior of a larger, more expensive one, keeping most of the usefulness at a fraction of the running cost. The result: a 2026 buyer can get GPT-4-class performance for pennies on the dollar compared to eighteen months ago.
What it means for you
This isn't just a boardroom story. Cheaper AI changes what's realistic for regular people too.
• At work: If your company pays for AI tools seat-by-seat, expect that cost to keep falling — or expect your employer to add more AI features to the tools you already use, because the underlying engine got cheaper to run.
• Running a small business: A solo e-commerce seller or local service business can now afford AI-generated product descriptions, customer support chatbots, and ad copy that used to require a $200-a-month enterprise plan. The math that made AI "too expensive to bother with" a year ago often doesn't hold anymore.
• Studying: Students who couldn't justify a premium AI subscription now have more free or nearly-free options for tutoring, essay feedback, and language practice, since providers are racing each other on price.
• Creative side projects: Cheaper models mean more free tiers and higher usage limits on tools for writing, brainstorming, and editing — useful if you're building a portfolio, a newsletter, or a YouTube channel without a budget.
• Freelance income: If you sell services like copywriting, virtual assistance, or basic design, AI tools that used to eat into your margin are now cheap enough that they boost it instead — you can take on more clients without hiring help.
• At home: Planning meals, drafting emails, checking homework, or getting quick explanations of paperwork gets easier when the AI behind your favorite app costs its maker less to run — because that savings tends to trickle down as more generous free plans.
How to try it right now
You don't need to pick a side in a US-versus-China debate to benefit from this price war. You just need to know where to look.
1. Start free. Before paying for anything, try the free AI tools at mykreatool.com — a no-cost way to test writing, image, and productivity AI tools without committing to a subscription while prices in this market are still moving.
2. Test a Chinese model directly. DeepSeek offers a free web chat at chat.deepseek.com, and it's genuinely strong at reasoning and coding tasks. Alibaba's Qwen chat is also free to use in a browser. Try the same prompt on both and compare the answers to what you're used to.
3. Check your current subscription's fine print. If you're paying for ChatGPT Plus, Claude Pro, or a similar plan, look at what you actually use each month. If it's mostly basic writing or Q&A, a cheaper model might cover 90% of your needs.
4. Pilot before you switch. For business use, run a two-week side-by-side test: same tasks, two different tools, and compare output quality and time saved — not just the price tag.
5. Watch for the lock-in trap. If you build automation, templates, or workflows around one provider's API, write down what it would take to move to a different one later. A cheap tool that's hard to leave isn't actually cheap.
Upsides and what changes
The obvious upside is cost. Tasks that used to require a paid plan — drafting a contract summary, generating a batch of product photos, translating a document — are becoming close to free. For small businesses and freelancers, that's real money back in your pocket every month, not a hypothetical.
Competition is also forcing every AI company, American or Chinese, to move faster and cut prices to keep users. That benefits everyone: features that were premium-only a year ago (longer memory, faster responses, image generation) are showing up in free tiers now because providers can't afford to hold them back while a rival gives them away.
It also means AI stops being a luxury reserved for companies with big budgets. A two-person startup can now access tools that only a well-funded tech company could afford in 2023.
Limitations
Cheaper isn't automatically better, and it's worth being honest about the trade-offs. Some Chinese models have had documented issues with censoring politically sensitive topics, which matters if your work touches current events, history, or anything China's government treats as off-limits. Data privacy is another open question — sending business or customer information to a server whose privacy practices and legal jurisdiction you haven't checked is a real risk, not a theoretical one. And "good enough" performance on everyday tasks doesn't always hold up for specialized, high-stakes work like legal drafting or medical information, where a mistake costs more than the money you saved. Test any new tool on low-risk tasks first, and read the privacy policy before you paste in anything sensitive.
Conclusion
The price of AI is falling fast, and Chinese models are the biggest reason why — American labs' usage share dropped from about 70% to 30% in roughly a year, according to Galloway and Tyrangiel's discussion of Juniper Research data. That competition is good news for your wallet, whether you're running a business, studying, or just trying to get more done at home. Your one action today: pick one task you currently pay for or avoid doing because of cost, and test it for free at mykreatool.com or on a free-tier chatbot before your next subscription renewal.
Comments 0