US firms test DeepSeek as AI costs climb

Some US companies are adopting China's DeepSeek AI models as expenses for Silicon Valley tools increase. The trend appears on corporate spending data from Ramp, though adoption remains limited compared with OpenAI and Anthropic.

DeepSeek ranked first on Ramp’s June list of trending software vendors. Ara Kharazian, lead economist at Ramp Economics Lab, noted that firms are sending payments directly to the Chinese startup and routing data to its hosted servers. Adoption of DeepSeek reached 0.1 percent by April 2026 after a brief rise to 0.3 percent in January 2025. OpenAI and Anthropic held shares of 32.3 percent and 34.4 percent in the same period. DeepSeek is also reported to be raising 7.4 billion dollars in its first external funding round at a valuation between 52 billion and 59 billion dollars, with investors including Tencent and CATL.

관련 기사

Sam Altman in talks with government officials about OpenAI equity stake
AI에 의해 생성된 이미지

OpenAI in early talks for US government stake

AI에 의해 보고됨 AI에 의해 생성된 이미지

OpenAI chief executive Sam Altman is in early discussions with the Trump administration about giving the US government a 5 percent equity stake in the company. The proposal would extend to other leading AI firms including Google, Meta, Anthropic and xAI. Any agreement would require congressional approval and remains far from certain.

Chinese AI firm DeepSeek is moving to design its own chips for running trained models. The effort aims to cut reliance on suppliers including Nvidia and Huawei amid US export controls.

AI에 의해 보고됨

Chinese companies have released advanced AI models that rival US offerings, yet the latest ones sparked little of the earlier market panic. DeepSeek's R1 in January 2025 triggered a trillion-dollar selloff in US tech stocks, but Z.ai's GLM-5.2 last month did not. The shift reflects China's growing role in open-source AI.

Tesla has begun deploying a localized dual-AI voice assistant in its vehicles in China. The system replaces the Grok model used elsewhere with two Chinese models to meet local regulations and preferences.

이 웹사이트는 쿠키를 사용합니다

사이트를 개선하기 위해 분석을 위한 쿠키를 사용합니다. 자세한 내용은 개인정보 보호 정책을 읽으세요.
거부