Nikkei reporting shows Indian startups turning to DeepSeek, Qwen, and Kimi as US frontier token costs become unsustainable. Cost is rewriting model selection.
Via ZeroHedge: Indian Companies Increasingly Turn To Chinese LLMs Due To "Unsustainable" US Token Bills
Indian companies are increasingly leaning on Chinese large language models from DeepSeek, Alibaba, and Moonshot AI to contain AI spend, according to Nikkei reporting summarized by ZeroHedge. The driver is not ideology. It is the bill.
Puneet Kumar of Mirae Asset Venture Investments India said several consumer technology startups he has met since mid-2025 use Chinese open-weight LLMs that help drive costs down by an "order of magnitude." A Trilegal partner put it more bluntly: token bills are "increasingly becoming unsustainable."
If you run an owner-led firm in the US, this is not an India-only story. It is a preview of what happens when "tokenmaxxing" meets a real budget — and when "good enough" models undercut frontier pricing hard enough that procurement stops pretending every task needs the most expensive stack.
The price gap is concrete. For DeepSeek models Microsoft makes available in southern India through Foundry, ZeroHedge cites charges between 19 cents and $1.74 per million input tokens and 51 cents to $5.40 per million output tokens. Moonshot's Kimi goes up to about 95 cents input and $4 output. OpenAI's GPT 5.5 series, by the same comparison, runs roughly $5 to $12 per million input tokens and $30 to $54 output.
That is the "sports car on a crowded city road" problem: expensive models used for basic work. Globally, the same pressure shows up as caps — Tesla, Amazon, Uber, and Walmart among firms reported as limiting AI usage — while Open Router data cited in the piece says Chinese LLM usage more than doubled to 25 trillion tokens in the final week of June, 78% more than US models. Coinbase, DoorDash, and Airbnb have publicly said they have begun using Chinese models.
Open weights also tempt firms that want local hosting so data stays in-country. That is a residency win on paper. It is not free of risk: the same legal source flags unverified deployment artifacts (malware or trojans) as a concern, and analysts warn access can be shut overnight amid geopolitics — including reports that Beijing may restrict overseas access to advanced Chinese models. Cheap tokens with no continuity plan are still a vendor-risk story.
Vidya Madhavan of dating app Schmooze described a familiar path: start with Google, OpenAI, Anthropic, and similar tools to learn what "great" looks like, then shift to open source plus tuning where the outcome holds and the savings matter. That is model selection in practice — not brand loyalty.
Do not copy India's startup stack blindly. Copy the discipline: match the model to the job and the bill.
The token bills are a serious issue -- it is increasingly becoming unsustainable. - Nikhil Narendran, Trilegal, via ZeroHedge / Nikkei
This story maps to Model Selection & Continuity Planning and runaway token cost control. The question is not "Chinese or American?" It is "which model fits which job, at what cost and residency, with what fallback when pricing or geopolitics move?"
AgentsROI stays vendor-neutral. I help owner-led firms choose deliberately — cloud, hybrid, or local — measure whether AI spend is returning value, and keep continuity so a cheaper model of the month does not become an accidental single point of failure. If usage is already sprawling across personal accounts and uncapped APIs, a Shadow-AI Risk Assessment or Workflow ROI Audit is the low-risk first step.
Primary next step: get a model map that assumes token prices and model menus will keep shifting — because they will.
India's shift toward cheaper Chinese LLMs is a cost signal, not a fashion tip. Frontier models still have a place. So do cheaper bases. The firms that stay solvent are the ones that decide which is which — before the token invoice does it for them.
Ready to pressure-test your model map and AI spend? Start with a Model Selection & Continuity review at AgentsROI.ai.
This article summarizes publicly reported information and is for general informational purposes only. It does not constitute legal, tax, financial, investment, security, or compliance advice. AgentsROI.ai is not a law firm, accounting firm, or registered investment adviser. Facts, pricing, statistics, and product capabilities cited here reflect the sources listed at the time of writing and may change. Readers should verify current information independently and consult qualified professionals regarding obligations specific to their industry, jurisdiction, and circumstances—including applicable New York State and New York City requirements. AgentsROI.ai may have commercial relationships with vendors mentioned; where material, such relationships are disclosed. Nothing in this article is an endorsement of any specific AI product, model, or provider.
