The global AI industry has been waiting weeks for DeepSeek’s next major model release, and the silence is becoming as revealing as any announcement. DeepSeek V4, the Hangzhou-based startup’s anticipated successor to the model that rattled markets in early 2025, has not appeared, and the central question circling the industry is whether its absence signals a problem with chips.
What Happened
More than a year after DeepSeek’s R1 reasoning model sent technology stocks tumbling and prompted U.S. President Donald Trump to call it a “wake-up call” for American firms, the company’s next-generation model, reportedly multimodal, capable of generating text, images, and video, has not materialized despite months of public anticipation. As AFP reported, industry watchers have been expecting the launch for weeks, and DeepSeek, Huawei, Alibaba, ByteDance, and Tencent have all declined to comment on the matter.
The most consequential question surrounding V4 is not performance but hardware. The Information reported that V4 can operate on the latest chips produced by Huawei, a claim, sourced from five people with direct knowledge of large orders for Huawei’s Ascend processors, that positions the model as a potential milestone for China’s chip self-sufficiency push. Alibaba, ByteDance, and Tencent are said to have placed significant orders for Huawei chips in preparation for the launch.
Yet the picture is complicated. Some reports allege that DeepSeek attempted to train V4 using thousands of Nvidia Blackwell chips, dismantled in third countries and smuggled into China to circumvent U.S. export controls. Nvidia told The Information it had seen no evidence of such activity and called the smuggling scenario “farfetched.” Meanwhile, other accounts suggest that an earlier attempt to train V4 on Huawei’s Ascend chips encountered serious technical setbacks, prompting a quiet return to restricted Nvidia hardware.
Wei Sun, principal AI analyst at Counterpoint Research, framed the stakes directly: “It’s important to know because at one level, it is a signal of China’s AI self-sufficiency trajectory.” As the Japan Times reported, the transition from Nvidia to domestic silicon would require “substantial re-engineering” and introduce performance trade-offs, especially for V4, a model expected to be state-of-the-art.
Stephen Wu, founder of Carthage Capital, predicted that V4 will be “a highly capable, open-source model that handles massive context windows at a fraction of the cost” but added a telling qualifier: if DeepSeek has successfully trained V4 entirely on Huawei silicon, “it signals a material shift in the geopolitical tech landscape.”
Why It Matters
The delay carries asymmetric meaning depending on the eventual chip disclosure. If V4 runs on Huawei Ascend chips at competitive performance, it would demonstrate that China can train frontier-class AI without Nvidia, a development that would undermine the central premise of U.S. semiconductor export controls. If V4 relies on restricted or smuggled Nvidia hardware, it would confirm that the controls are working and that China has not yet achieved meaningful AI chip independence.
This matters beyond DeepSeek’s competitive position. The broader implication is that the effectiveness of Washington’s chip export strategy, and by extension, the timeline of China’s AI catch-up, may hinge on a single model release. Markets, policymakers, and competitors are all reading the same silence.
For investors, the wait also underscores a structural tension in China’s AI buildout: the country’s most capable AI companies remain dependent on foreign silicon to train the models that most loudly signal their independence.
What Comes Next
When V4 does arrive, the chip question will dominate the immediate narrative. Analysts will scrutinize any available performance benchmarks against V3, which was trained on Nvidia hardware, for signs of degradation or the absence of it. The U.S. Congress, which passed the MATCH Act through committee in early April to tighten controls on semiconductor manufacturing equipment exports to China, will be watching closely. The release could accelerate legislative momentum in either direction.