DeepSeek is not a threat but an opportunity

As the big guys look to repair the damage, small manufacturers see DeepSeek as an opportunity to scale up their operations.

“There’s been a huge increase in developer demand to replace expensive proprietary OpenAI models with open source models like DeepSeek R1,” said Andrew Feldman, CEO of Cerebras Systems, an AI chip startup. Cerebras Systems competes directly with Nvidia and offers cloud services on private clusters.

DeepSeek opens up AI development opportunities for small companies

The launch of DeepSeek R1 has created one of the largest spikes in demand for its services in the company’s history, according to CNBC. This shows that the growth of the AI ​​market will not depend on a single company, Feldman said, because open models are not tied to specific hardware or software. DeepSeek claims that its reasoning model requires less computing power than its American competitors and can be trained without the need for advanced GPUs.

DeepSeek has helped accelerate the adoption of new technologies in the area of ​​AI acceleration, including model training and deployment. Nvidia currently dominates the AI ​​training hardware market, but DeepSeek’s emergence has led many competitors to believe they have an opportunity to expand their presence in running trained models that promise higher performance at lower costs.

AI training requires a lot of computational resources, but weaker hardware can still be sufficient to run trained systems. This has created a high demand for alternative program developers as more customers are willing to seek solutions from DeepSeek models.

DeepSeek Helps AI Get More Widely Adopted

The cost of Nvidia GPUs to serve the needs of AI servers is not cheap

Analysts and industry experts believe that Chinese AI system training and deployment labs will influence the growth of the entire industry. If the services of deploying trained models become cheaper, AI technology will be widely adopted, leading to increased demand - a phenomenon known as the "Jevon paradox".

The surge in demand has been confirmed by companies like d-Matrix and Etched, which specialize in developing AI programs for startups. “The availability of small-scale models has acted as a catalyst for the inferential AI era,” d-Matrix said. Meanwhile, the company has attracted dozens of enterprise customers since using DeepSeek R1, Etched said, “Companies are shifting their spending from training clusters to inference clusters.”

Finally, it’s worth noting that Nvidia’s resources aren’t limitless. Even a tech giant like Nvidia can’t meet the entire global demand for AI accelerators, which makes DeepSeek’s approach open up opportunities for smaller companies in the industry.