American companies switch to Chinese AI — OpenAI and Anthropic have become too expensive

DeepSeek Instead of OpenAI: American Companies Switch to Cheap AI

According to CNBC, published in July 2026, American companies are increasingly switching to Chinese AI models, primarily DeepSeek, due to rapidly growing expenses for solutions from OpenAI and Anthropic.

Why American Companies Are Leaving for Chinese Suppliers

The main reason is the price gap. Using the OpenAI API or Claude from Anthropic costs organizations significantly more than comparable services based on Chinese models. Meanwhile, the quality of the latter, which recently lagged behind market leaders noticeably, is steadily converging with American flagships.

Small and medium-sized businesses that need mass data processing, content generation, or customer support automation have proven especially sensitive to API request costs. This segment began switching first—especially those using AI not as an experiment, but as operational infrastructure.

An important nuance: this is not about a complete departure from American suppliers. Many companies are transitioning to a hybrid model—tasks requiring maximum quality or work with sensitive data remain with OpenAI or Anthropic, while high-volume routine operations are delegated to cheaper Chinese models.

What Changed in the Quality of Chinese Models

In 2024, Chinese AI models had a noticeable disadvantage over GPT-4 and Claude 3 on complex tasks—multi-step reasoning, programming, long document analysis. By mid-2026, the picture has changed: a number of Chinese developers—including teams from DeepSeek, Alibaba, and ByteDance—released models whose results on standard benchmarks are comparable to leading Western flagships.

For most applied business tasks—document summarization, customer chatbots, processing incoming requests, basic code writing—the quality difference between the best American and Chinese models has shrunk to insignificant. This is precisely what makes the price argument particularly compelling.

• Chinese models cost several times less than American ones in terms of API token pricing
• Quality on applied tasks (documents, support, code generation) has significantly converged with OpenAI and Anthropic
• Companies with high request volumes are switching first—retail, service platforms, SaaS
• Some companies are taking a hybrid approach: expensive models for complex tasks, cheap ones for routine work

How OpenAI and Anthropic Respond to Competition

American AI labs are building differentiation not only on model quality. The key arguments in favor of American suppliers are data privacy and compliance with U.S. regulatory requirements: for corporate clients from finance, healthcare, and defense sectors, this is critically important.

Additionally, OpenAI and Anthropic are actively investing in deep integration with corporate systems, developer tools, and partner ecosystems—areas where open or Chinese models still lag in maturity. Nevertheless, price pressure is mounting: since 2023, token costs for leading American suppliers have dropped tenfold, and the trend continues.

What This Means

The transition of American companies to Chinese AI models is a sign of structural market maturity: when basic quality converges between providers, price becomes the main selection argument. For OpenAI and Anthropic, this is a signal to force differentiation through enterprise tools, data security, and industry-specific solutions. For the market as a whole—acceleration of commoditization of basic AI services and further price declines across the sector. Companies that were overpaying for AI infrastructure two years ago are now getting comparable quality for less money—and they are using it.

Why Are Companies Switching to DeepSeek?

The main reason is the price gap. Using the OpenAI API or Claude from Anthropic costs organizations significantly more than comparable services based on Chinese models.