To write that the topic of AI is on everyone’s lips could well be considered an enormous understatement. One of the biggest beneficiaries of all the hype: GPU manufacturer Nvidia, which has had chips for machine learning and other suitable tasks ready for a long time.
This is also illustrated by statements by the financial expert Frank Lee, who works for one of the largest banks in the world. More specifically, he works for HSBC and Hong Kong & Shanghai Banking Corporation respectively. Here’s what Lee had to say about AI Yahoo Finance said:
In particular, we’re shocked by Nvidia’s pricing power in AI chips, which we think is driving earnings and valuation higher. (…) While the overall (addressable market) of Generative AI is still difficult to predict, Nvidia has the greatest potential leverage from a hardware perspective.
No wonder, after all, the makers of ChatGPT rely on tens of thousands of Nvidia graphics cards. The same applies to Elon Musk’s recently launched AI project:
Microsoft is said to be already working on its own AI chips and Google has its own cloud platform TPU v4 on offer for this purpose, which is to be followed by version v5 at the end of the year. There is also possible competition in the form of old chip acquaintances such as AMD and Intel. It remains to be seen how quickly other providers will be able to catch up with Nvidia.
HSBC’s current forecast for the financial year 2024 is that Nvidia will dominate the AI market with a 90 percent share. It is unlikely to stay that way sooner or later.
How could this affect the gaming market?
The extent to which Nvidia’s current dominance will affect the gaming business is difficult to predict. However, one important aspect cannot be dismissed out of hand: Even Nvidia can only produce chips via suitable manufacturers such as TSMC to a limited extent. And the extremely lucrative AI market is particularly strong competition for gaming GPUs.
According to HSBC analysis, AI chips are sold at prices which are 10 to 20 times higher than standard gaming chips. If the demand is high enough, Nvidia will probably be attracted by particularly high margins in this area.
The situation on the hard-fought chip market should improve as a result of additional factories, the construction of which in many cases began a year or two ago due to the shortage of chips. From our point of view, it is nevertheless quite plausible that the extremely strong internal competition from AI chips ensures that Nvidia’s gaming chips remain very expensive – or even become more expensive.
The PC Games Hardware colleagues seem to share this assumption. So was in hers report on the topic recently by a
gild talking about valuable wafers in gaming GPUs. Let’s hope that, as is so often the case, the wisdom is that in the end it always turns out differently than you think.
How do you assess the situation? Due to the extremely high prices for AI accelerators, can (even) higher prices for gaming graphics cards be expected in the future? Do you instead assume that the price situation will improve again sooner or later? Or will it stay with the status quo for a longer period of time? Feel free to write your assessment in the comments and join the discussion!