The announcement that a team of Chinese researchers has created an open-source large language model has caused deep declines in tech shares traded on the Nasdaq.
On the surface, it’s understandable. The new DeepSeek model delivers high-quality results at a fraction of the cost. Investors are concerned that DeepSeek’s ability to process results 10-30 times cheaper could pop the AI bubble.
I expect the opposite to happen.
If anything, DeepSeek's success shows that we can expect AI models to deliver extraordinary results at an affordable price faster than ever before.
I expect every tech firm to copy the "efficiency gains" championed by DeepSeek. It will mean that every $1 billion investment in AI in the U.S. will result in an extra 10-30 fold gain in intelligence output.
We’ve already read reports that Meta now has four "war rooms" working on replicating all the efficiency techniques published by DeepSeek, and we expect to see the benefits in its upcoming Llama 4 or 5 with a crazy jump in performance.
Short-term, this may be negative for Nvidia. That’s because people won’t need to buy as much hardware as originally planned. Instead, they can turn to engineering optimization.
Longer-term I am bullish on the ability of LLM enhancements to boost productivity and encourage new AI fueled innovation.
Here is how I expect many of the tech giants to be affected:
⬆️Apple benefits from lessons learnt in optimizing for memory bandwidth in “edge inference”
⬆️Meta benefits because it wants to deploy AI services everywhere
⬆️Microsoft benefits from selling cheaper Azure LLM services
⬆️Amazon benefits because it didn’t build an LLM and can now host frontier open source models
⬇️Google will be hurt as their advantage in TPU hardware is eroded and new AI competition will emerge for Google services such as search
⬇️AI labs that don't open source will suffer from brutal price competition, i.e. OpenAI, Anthropic and Cohere
There’s certainly an irony that China - known as a country that mass produces consumer goods – has emerged as the source of innovation that may mass produce intelligence!!
It’s a good indication that the focus of capital and attention on training LLMs is coming to an end and will shift to products and services that are being built on top of the technology.