China\u2019s cheaper AI tokens a double-edged sword for Asian\nbusinesses<\/p>\n
In the second of a two-part series on AI tokens, CNA explores how\nlower-cost Chinese models are emerging as an attractive option across\nbusinesses in Asia, even as experts warn that price is only one part of\nthe equation.<\/p>\n
SINGAPORE: Artificial intelligence (AI) is often described as a race\nto work smarter. But for businesses in Asia, the more urgent question\nmay be simpler: who can afford to use it at scale?<\/p>\n
At the heart of that question is the cost of AI tokens - a\nlittle-known building block that determines how much companies pay when\nAI systems read, process and generate information.<\/p>\n
This is where Chinese AI models could excel and gain more traction\nover American AI models among businesses in Asia - especially India and\nSoutheast Asia - experts told CNA, citing their ability to offer cheaper\nAI tokens.<\/p>\n
For instance, models from Chinese companies such as MiniMax and\nMoonshot charge about US$2 to US$3 per million output tokens.<\/p>\n
In comparison, Google\u2019s Gemini 3.5 Flash model charges about US$9,\nAnthropic\u2019s Claude Sonnet 4.5 costs about US$15 and OpenAI\u2019s GPT 5.5\nmodel is priced at US$30, according to a Financial Times report and\nGoogle and OpenAI\u2019s pricing documents.<\/p>\n
Charges are based on the number of input and output tokens\nconsumed.<\/p>\n
Input tokens come from the prompt or material sent to the AI, while\noutput tokens come from the response it generates. Output tokens usually\ncost more.<\/p>\n
A small sales team of 50 employees could use about 450 million tokens\nmonthly, including both input and output tokens, according to estimates\nfrom Amit Verma, founding head of technology at US-based AI services\nfirm Neuron7.ai.<\/p>\n
This can amount to a cost of about US$3,150 monthly and US$38,000\nannually using GPT 5.5 model, which is around two to three times more\nthan the costs of Chinese AI models.<\/p>\n
Chinese AI token costs are cheaper because of a mix of efficient\nmodel designs, lower energy and data infrastructure costs, government\nsubsidies, and aggressive pricing strategies, said experts.<\/p>\n
As companies move from simple AI chatbots to AI agents that can plan,\nsearch, verify information, connect to other software systems and repeat\ntasks in the background, token usage can surge - and so can costs.<\/p>\n
\u201cToken costs multiply across every step, making the unit price of\neach token far more consequential,\u201d Wong Qi Han, an independent AI\nresearcher and builder, told CNA.<\/p>\n
Signs are emerging: companies and organisations such as Airbnb,\nThinking Machines Lab - founded by former OpenAI chief technology\nofficer Mira Murati - and AI Singapore have incorporated Alibaba\u2019s Qwen\nmodels.<\/p>\n
Experts said AI token prices from Chinese AI firms such as Alibaba\u2019s\nQwen, DeepSeek, Kimi, Zhipu\u2019s GLM and MiniMax are giving startups and\nenterprises a cheaper way to run high-volume AI tasks, especially in\nprice-sensitive markets such as India and Southeast Asia.<\/p>\n
They added that cheaper AI tokens could make its adoption far more\naffordable across call centres, software development, e-commerce,\neducation, legal research, manufacturing and back-office operations.<\/p>\n
But observers also added that the cheaper route comes with\ntrade-offs, including quality, latency, trust, regulation, data security\nand geopolitical risk.<\/p>\n
WHY CHEAPER TOKENS MATTER TO ASIAN BUSINESSES<\/p>\n
AI token-based pricing mainly affects companies and developers that\nbuild AI into products, apps and internal workflows.<\/p>\n
While ordinary users may access AI through free or fixed-fee\nsubscriptions, businesses running AI at scale typically pay by usage,\nbased on the number of input and output tokens their systems\nconsume.<\/p>\n
Tokens also act like a billing meter: the more a system reads and\ngenerates, the more it costs.<\/p>\n
Experts said this pricing model is now common because corporate\nbusinesses and enterprises use AI at a much larger scale than ordinary\nusers - across millions of customer chats, coding requests, research\ntasks, document summaries and background AI agent actions.<\/p>\n
Every chatbot reply, code suggestion, translation, document summary\nor AI agent action consumes tokens.<\/p>\n
According to a joint report by McKinsey, Singapore Economic\nDevelopment Board and Tech in Asia in February, 46 per cent of companies\nin Southeast Asia had gone beyond AI experimentation to include them in\ntheir workflows and products.<\/p>\n
In India that number is 47 per cent, according to an Ernst &\nYoung-Confederation of Indian Industry report.<\/p>\n
For Asian companies, the cost of AI is becoming a business problem,\nespecially with the rise of AI agents, experts told CNA.<\/p>\n
AI agents go beyond answering prompts. They can plan steps, check\ninformation, use apps or company systems, and repeat actions in the\nbackground to complete a task.<\/p>\n
Verma told CNA that AI use is shifting from \u201csingle-turn prompts\u201d -\nsimple one-step AI requests - to agentic workflows that may require 50\nto 100 internal operations for a single output.<\/p>\n
These background steps, including prompting, verification,\nreflection, code execution and the use of other external software tools,\nall consume more tokens, he said.<\/p>\n
Based on Anthropic\u2019s estimates, the average AI token cost of a\nsoftware developer in an enterprise using Claude Code was US$13 per day,\nwith monthly costs of roughly US$150 to US$250 per developer, said a\nBusiness Insider report in April.<\/p>\n
For large tech businesses employing 500 developers, the AI token\ncosts would be roughly US$75,000 to US$125,000 a month, or US$900,000 to\nUS$1.5 million a year, before any discounts or enterprise deals.<\/p>\n
Most AI providers do offer discounts for enterprise customers.<\/p>\n
However, these are typically negotiated privately and based on usage\nvolume commitment, contract length, models being used, customer support\nrequired and if whether cloud services are included.<\/p>\n
According to media reports, OpenAI has offered some enterprise\ncustomers 10 per cent to 20 per cent discounts on multi-year or bundled\ndeals.<\/p>\n
Verma added that Asia \u201cmay become the first region where AI b<\/p>", "url": "https:\/\/jawawa.id\/newsitem\/chinas-cheaper-ai-tokens-a-double-edged-sword-for-asian-businesses-1781563808", "image": "" }, "sponsor": "Okusi Associates", "sponsor_url": "https:\/\/okusiassociates.com" }