AI and Induced Demand
Why selling Nvidia now is probably short sighted...
When LED bulbs were invented and suddenly lighting was more efficient, did that mean less people wanted light bulbs?
With the release of DeepSeek AI claiming it achieved better performance for a fraction of the training cost, many investors in US tech companies decided to sell shares. Nvidia in particular took a large hit. Yet this is likely short sighted, as the actual reality of fielding AI systems means that more chips are needed.
Although the training step is made more efficient, the usage of the AI models (called inference) will increase as more folks want to utilize the systems.
Coupled with the rise of robotics demanding lots of high powered edge computing, it is unlikely that Nvidia will reduce in value. Instead the induced demand will mean that chips will be needed more than ever. Nvidia themselves have said much the same so I will not belabor the point.
Yet there are still risks of disruption to Nvidia and other established chip manufactures and fabs. The same pressures of induced demand will likely lead to the emergence of entirely new chip architectures including the much-needed and under explored Reversible Computing. However large firms can buy their way into continued relevance and Nvidia’s position of software platform lock in is not trivial. In future posts I will explore the relationship between vertical and horizontal integration that firms can use to stay relevant.
This is not financial advice, just an observation. Let’s see what happens when everyone tries to use the new ‘roads’ laid down by the DeepSeek highway…