• Chinese AI startup, DeepSeek, has seen its assistant overtake ChatGPT as top rated free app in the Apple App Store.
• The DeepSeek-V3 model in particular has proved popular since launching Stateside earlier this month.
• DeepSeek reportedly used NVIDIA’s H800 chips for training, which challenges the US’ restricting on processor exports.

The beginning of 2025 is proving an exciting time for Chinese AI startup DeepSeek. In recent weeks it launched V3 of its AI assistant in the US and in only a couple of weeks, it has already surpassed OpenAI’s ChatGPT is the top-rated free app in the Apple App Store Stateside.

It is an impressive feat, especially considering how pervasive the ChatGPT platform has been since debuting officially to the public a little over two years ago. Since that time it has dominated the consumer-facing generative AI space, with the likes of Anthropic, X (formerly Twitter), Microsoft, Google, and Meta all launching similar offerings.

Instead it is DeepSeek that could present the biggest challenge to ChatGPT.

Much of this success can be attributed to the fact that DeepSeek is open-source, which is not something that OpenAI currently claims about ChatGPT, with many of its advanced features locked behind paid subscription tiers.

Per Reuters, creators at the Chinese startup note that V3-DeepSeek now, “tops the leaderboard among open-source models and rivals the most advanced closed-source models globally”.

Reportedly helping it reach that level is NVIDIA’s H800 chips which are used for the training of the model, although some naysayers contend this fact. DeepSeek said it only spent $6 million in order to do so, which is far less than such a system has cost to implant at other firms.

“The NVIDIA H800 GPU features fourth-generation Tensor Cores and the Transformer Engine with FP8 precision, further extending NVIDIA’s market-leading AI leadership with up to 9X faster training and an incredible 30X inference speedup on large language models. For high-performance computing (HPC) applications, The GPUs triple the floating-point operations per second (FLOPS) of FP64 and add dynamic programming (DPX) instructions to deliver up to 7X higher performance,” outlined Lenovo of its capabilities in 2023 for AI-focused server environments.

“It uses breakthrough innovations in the NVIDIA Hopper architecture to deliver industry- leading conversational AI, speeding up large language models by 30X over the previous generation,” it added.

What is equally interesting about this development is the fact that the United States has actively been looking to restrict the export of AI-focused chips to other parts of the globe, and in recent weeks, looked to intensify those efforts, much to the frustration of NVIDIA.

Following the success of DeepSeek, we could see US government officials look to get even stricter, as the economic tensions between the United States and China only looks to get betting worse.

In the interim, however, it appears as if DeepSeek is looking to make waves, which may prompt OpenAI to deepen its own partnership with the US government as it aims to establish dominance in the pervasive generative AI landscape.

[Image – Planet Volumes on Unsplash]