NVIDIA Confirms DeepSeek’s Breakthrough Technology Fully Complies with US Export Regulations

NVIDIA Confirms DeepSeek’s Breakthrough Technology Fully Complies with US Export Regulations

This content is not intended as investment advice. The author does not hold any positions in the stocks mentioned.

DeepSeek’s Revolutionary R1 Model Raises Eyebrows in Tech Industry

DeepSeek has recently stirred the global technology landscape with the unveiling of its innovative R1 model, which can effectively compete with OpenAI’s o1 model while incurring only about 1/50th of the usual expenses associated with training complex large language models (LLMs).This dramatic advancement has ignited extensive discussion online, particularly regarding the methods employed by the Chinese startup to accomplish such a remarkable feat. The situation escalated further when NVIDIA voiced its support for DeepSeek’s initiatives.

Controversial Claims about NVIDIA’s H100 GPUs

Alexandr Wang, the founder and CEO of Scale AI, recently generated significant buzz by allegedly revealing that DeepSeek possesses “about fifty thousand H100s”– NVIDIA’s powerful GPUs. However, US law prohibits the sale of H100 chips to China, leading to a divided reaction within the tech community regarding the validity of Wang’s assertions.

Elon Musk subsequently contributed to the conversation, seemingly endorsing Wang’s comments about DeepSeek’s supposed inventory of H100 GPUs.

Speculations Regarding Supply Channels

The online discourse soon shifted into a whirlwind of speculation, with many suggesting that Singapore could serve as a potential conduit, allowing DeepSeek to acquire NVIDIA’s chips. One analyst highlighted that while NVIDIA reported a significant portion of its sales going to China, its filings indicated that shipments to Singapore were deemed “insignificant.”

NVIDIA’s Response to Speculations

To offer clarity amid the swirling rumors, NVIDIA issued a statement on Sunday. According to the GPU leader:

“DeepSeek is an excellent AI advancement and a perfect example of Test Time Scaling. DeepSeek’s work illustrates how new models can be created using that technique, leveraging widely-available models and compute that is fully export control compliant. Inference requires significant numbers of NVIDIA GPUs and high-performance networking. We now have three scaling laws: pre-training and post-training, which continue, and new test-time scaling.”

NVIDIA contends that DeepSeek achieved its notable success while adhering strictly to US export controls, effectively dispelling the rampant speculation surrounding the startup’s alleged GPU holdings.

NVIDIA Faces Market Reaction

Meanwhile, NVIDIA’s stock experienced a significant downturn, plummeting 17% during trading in response to fears that DeepSeek’s breakthrough could minimize the demand for hyperscalers, subsequently impacting NVIDIA’s order volumes. This decline has resulted in NVIDIA slipping to third place among the world’s most valuable companies.

For further details, visit the original source: Source & Images.

Related Articles:

Notepad++ Version 8.7.6 Release Notes
NVIDIA GeForce RTX 5090 Release Excitement Grows as Fans Camp Outside Retailers

Leave a Reply ▼

Your email address will not be published. Required fields are marked *