
The AI industry is undergoing a seismic shift, and at the center of this transformation is DeepSeek, a Chinese AI model that has sent shockwaves through the market. Here I will highlight how this small but innovative company is challenging the dominance of tech giants like NVIDIA and OpenAI and reshaping the landscape of artificial intelligence. Here’s a breakdown of what’s happening and why it matters.
The AI Bubble and the Rise of DeepSeek
The AI industry has been riding a wave of investor excitement, with inflated stock prices and speculative narratives driving a bubble reminiscent of past financial crises. Major players like the Federal Reserve have fuelled this trend, but the emergence of DeepSeek has exposed the fragility of this bubble.
DeepSeek’s breakthrough lies in its ability to achieve performance comparable to, or even superior to, established AI models like ChatGPT, but at a fraction of the cost. While companies like OpenAI spend upwards of 100 million to train their models, DeepSeek has managed to do it for just 5 million. This drastic reduction in costs is a game-changer, signalling that smaller, agile teams (nearly 200 PhD graduates) can now compete with tech giants.
How DeepSeek is Disrupting the Market
DeepSeek’s success isn’t just about cost savings—it’s about innovation. Here’s how they’re shaking up the industry:
Cost Efficiency
Traditional AI models require massive resources, including hundreds of thousands of GPUs and enormous data centers. DeepSeek, on the other hand, has reduced the number of GPUs needed from 100,000 to just 2,000.
By cutting training costs from 100 million to 5 million, DeepSeek has proven that AI development doesn’t have to be prohibitively expensive.
Innovative Approaches
DeepSeek employs novel techniques like reducing memory usage and utilising specialised expert systems. These innovations allow the model to operate efficiently on less powerful hardware.
Instead of relying on massive, resource-intensive models, DeepSeek uses a multi-token system that optimises processing and reduces costs.
Open Source Model
Unlike traditional AI companies that guard their technology closely, DeepSeek has made its code and technical papers open source. This transparency allows the public to verify their claims and methodologies, fostering trust and collaboration.

The Impact on Major Companies
The ripple effects of DeepSeek’s emergence are already being felt across the industry. NVIDIA, a leader in AI hardware, lost nearly $600 billion in market value in a single day following DeepSeek’s launch. This dramatic shift underscores the vulnerability of traditional
AI companies to disruptive innovations.
The implications are clear: the dominance of large corporations in the AI space is no longer guaranteed. Smaller entities like DeepSeek can now compete effectively without the need for massive data centers or exorbitant budgets.
Scepticism and the Road Ahead
While DeepSeek’s achievements are impressive, some industry experts remain sceptical. Questions have been raised about the legitimacy of their reported GPU usage and whether their claims can be replicated at scale. However, even sceptics acknowledge that DeepSeek’s approach represents a significant shift in how AI development can be approached.
What This Means for the Future of AI
DeepSeek’s success is reshaping the future of AI by making development more accessible and affordable, enabling smaller companies to innovate and compete. This challenges tech giants to prioritise efficiency over scale and highlights the growing influence of Chinese innovation, potentially balancing the global tech landscape. By challenging the status quo, this Chinese firm has proven that innovation and efficiency can disrupt even the most entrenched players. As the AI bubble begins to deflate, the industry is poised for a new era of competition, accessibility, and creativity.
Whether DeepSeek’s claims hold up to further scrutiny or not, one thing is clear: the future of AI is no longer in the hands of a few tech giants. The game has changed, and the possibilities are endless. Looks good to me.