Member-only story

THE END OF NVIDIA’S FAIRY TALE? Lessons from the DeepSeek Disruption

Oliver Inderwildi
6 min readJan 28, 2025

By Oliver Inderwildi

Chinese startup DeepSeek has slashed AI training costs by 95%, threatening NVIDIA’s silicon empire and sparking fears of a tech power shift.
But could this efficiency revolution actually fuel AI adoption and demand for computing power, turning NVIDIA’s apparent crisis into an unprecedented boom?

Key Takeaways:

  1. DeepSeek’s Efficiency Breakthrough: Cuts training costs by 95% and boosts efficiency 45 times reducing the demand for advanced microchips
  2. The Compute Shift: Less compute (computing power demand) works, but reasoning models thrive on more test-time compute, shifting focus to inference.
  3. Nvidia’s Competition Heats Up: Giants like Google and Amazon, plus startups like Cerebras, threaten Nvidia’s dominance.
  4. AI Demand Will Soar: Efficiency boosts will drive exponential compute demand, not reduce it.
  5. Innovation Disrupts Constantly: Rapid AI innovation makes adaptability key for investors and stakeholders.
  6. Ultimately, its about Power: Recent breakthroughs, this time by the Chinese, are part of a race for technological dominance and geopolitical power.

If you’re an AI geek and investor like myself, the recent news from DeepSeek probably hit you like an existential crisis with a spreadsheet. On the one hand, the results were staggering: this…

--

--

Oliver Inderwildi
Oliver Inderwildi

Written by Oliver Inderwildi

Dr Oliver Inderwildi is Managing Director of CMPG, Senior Fellow at the University of Cambridge and Editor of several books published by Springer Nature.

Responses (7)