Nvidia raises prices and takes over a new market share that no one talked about - until today!
The supply chain shortage for artificial intelligence demand continues to seep deeper - and this time it reaches a place most investors didn't even think about:
AI chip testing interfaces.
Nvidia raises prices to secure significant quantities of production capacity from key suppliers of test cards and test sockets.
The meaning is simple, the problem is no longer limited to GPU production itself - the way the chip is tested and validated also becomes a bottleneck.
Why does Nvidia need to perform tests?
New generation AI chips are becoming more complex, larger, and require more advanced testing processes,
(Test cards are used to create electrical contact with the chip, while test sockets allow testing of chips and packages during production phases)
As chips move to more advanced technologies - testing becomes more complex.
The interesting story is that Nvidia is not only ordering more production capacity - it is willing to pay more to get it.
This means the company treats testing capability as a strategic resource,
And that makes sense - Nvidia is currently in the midst of a significant acceleration in the production of its artificial intelligence systems.
And when huge quantities of chips and AI components need to undergo testing and validation, even a tiny bottleneck - can turn into a significant delay.
Then TSMC comes with plan B
Here the story becomes even more interesting...
TSMC and other chip companies are starting to explore additional suppliers of testing interfaces - with the aim of reducing dependence on suppliers whose capacity is being taken up by Nvidia.
This does not mean TSMC is abandoning existing suppliers.
It means something much more interesting:
When Nvidia starts to turn off the tap - the industry starts looking for another tap:
And this is precisely why secondary suppliers in the field can now become strategic.
This event reinforces a picture that is already becoming clear - the demand for AI is starting to encounter bottlenecks at every possible layer:
First there was an HBM shortage, then advanced packaging, and then production equipment,
And now we are seeing pressure on testing and validation equipment as well.
The market is no longer just dealing with the question - "How many chips can be produced?"
The question becomes - "How quickly can they be produced, packaged, tested, and validated?"
And that's a big difference,
Because if these bottlenecks continue to expand - pricing power will gradually shift to smaller companies that are in critical stages of the production chain.
The artificial intelligence game is becoming a race where it's not enough to build the fastest engine - you need enough roads, gas stations, and mechanics to keep it moving.
Which companies are we following?
$COHU Cohu - direct exposure to test sockets, test cards, and advanced testing interfaces for AI and HBM chips.
$TER Teradyne - automated test equipment and test systems for HBM and advanced AI chips.
$KLAC KLA - measurement and process control equipment that benefits from the increased complexity of chip manufacturing and testing. $AMAT Applied Materials - broad exposure to chip manufacturing equipment, which can benefit from increased capacity due to supply chain bottlenecks.