The story of AI in the 2020s has been a debate between those who claim it has the power to fix everything, and those who fear it will ruin everything. The tech industry has been fueled by the narrative that it needs vast amounts of power and computing resources to achieve its goals, with many companies and investors investing heavily in the idea. However, this may be about to change.
A Chinese company called DeepSeek has recently dropped a large language model, R1, that is sparking a major shift in the industry’s narrative. R1 is a cheaper, more efficient ChatGPT-like model that was built on a fraction of the budget and resources used by other leading chatbots.
This has sent shockwaves through Silicon Valley and Wall Street, with many industry insiders and investors reevaluating their assumptions about the need for massive computing power and resources to develop AI. The achievements of DeepSeek’s R1 model have forced investors to question whether the vast sums of money being poured into AI projects are necessary.
DeepSeek’s model is built on just 2,000 Nvidia chips obtained before US export restrictions were put in place, compared to the 25,000 chips used by OpenAI to build GPT-4. This has led many to reevaluate the need for such vast amounts of power and resources, and to consider alternative approaches to AI development.
The news has also had a significant impact on the stock market, with Nvidia’s shares falling 17% on Monday and shedding $600 billion in market value. Other tech companies, including Alphabet, Microsoft, and Oracle, also saw their stocks fall.
The shift in the narrative around AI may not necessarily signal the end of the industry, but it does indicate that investors will be more cautious and discerning in their approach to funding AI startups and projects. The era of unbridled enthusiasm and investment in AI may be coming to an end, and a more measured and sustainable approach may emerge in its place.
