Explained: Is DeepSeek about to burst the AI bubble?
The artificial intelligence industry is going through a critical moment, and everything points to DeepSeek, a Chinese startup that has burst onto the scene, being a game-changer. Its advances have shown that it is possible to develop high-performance AI models at a fraction of the cost and resources used by today's tech giants. This has called into question the sustainability of the business model of many companies that rely on expensive hardware and massive energy consumption to train gigantic models.
Why could DeepSeek shake up the industry?
DeepSeek has managed to train its most advanced model, DeepSeek-V3, with 671 billion parameters for just $5.5 million, while models like OpenAI's GPT-4 would have required more than $100 million. This means that the traditional AI model, based on the accumulation of expensive hardware, could be on the verge of a transformation or collapse.
Companies that have invested billions of dollars in data centers packed with high-performance GPUs, such as NVIDIA, Google, Microsoft and Amazon, could face serious problems if DeepSeek proves that its approach is more efficient and scalable at low cost.
The risk for the industry
If DeepSeek's model takes hold, we could see a domino effect throughout the AI value chain:
Flooded used hardware market: Many startups that bet everything on expensive AI servers could be forced to sell off their equipment, creating a glut that collapses the prices of specialized GPUs and chips.
Revaluation of the business model: Tech companies that have based their growth on AI will need to justify huge investments, as investors begin to question whether so many resources are really required to obtain competitive results.
Paradigm shift in AI: If the efficiency demonstrated by DeepSeek is replicated, the need for infrastructure will be reduced
