Artificial Intelligence and the Semiconductor Market
Lately having started looking at artificial intelligence and hardware there has been a series of question that have come up. However one big question is regarding the network of resources that are required to build and maintain systems running artificial intelligence not only from a software perspective – additionally examining the supply chain.
Looking at it from a critical perspective is important, and I have taken a look at some thoughts by the AI Now Institute with their mapping of these aspects.
Revisiting the Anatomy of An AI system
The Amazon Echo as an anatomical map of human labor, data and planetary resources
If one would imagine some sort of scale of academic critical mindset for implementation related to production and the financial consulting interest of facilitating these very same actors/producers. Well, then one might say McKinsey is on that opposite side of the spectrum.
McKinsey writing about semiconductors focused on artificial intelligence and hardware. This insight report is aimed at semiconductor companies.
It would likely be best if you read the whole report, yet this article sums up some of the main points.
On behalf of these companies they argue the following:
- “AI could allow semiconductor companies to capture 40 to 50 percent of total value from the technology stack, representing the best opportunity they’ve had in decades.
- Storage will experience the highest growth, but semiconductor companies will capture most value in compute, memory, and networking.
- To avoid mistakes that limited value capture in the past, semiconductor companies must undertake a new value-creation strategy that focuses on enabling customized, end-to-end solutions for specific industries, or ‘microverticals.’”
In their report they wrote a general outline of the historic backdrop in the AI industry beginning in the 1950s and rapid expansion in the 2010s.
They talk too of the great increase in the ‘next-generation accelerator architectures’. I wrote about this previously in an article about AI accelerators.
However the article from McKinsey mentions the technology stack for AI as having nine layers.
Semiconductors are an important part of this ecosystem.
“Semiconductors are materials which have a conductivity between conductors (generally metals) and nonconductors or insulators (such as most ceramics). Semiconductors can be pure elements, such as silicon or germanium, or compounds such as gallium arsenide or cadmium selenide.”
McKinsey argues that AI will drive a large portion of semiconductor revenues for data centers and the edge.
They did so by estimating the potential value for semiconductor companies related to compute functions and found the following:
“Our research revealed that AI-related semiconductors will see growth of about 18 percent annually over the next few years — five times greater than the rate for semiconductors used in non-AI applications.”
There are several units that are used for compute that requires different AI architectures, however in general McKinsey lists the following:
“Compute performance relies on central processing units (CPUs) and accelerators — graphics-processing units (GPUs), field programmable gate arrays (FPGAs), and application-specific integrated circuits (ASICs).”
They outline as well that the infrastructure of data centres may change.
“GPUs are now used for almost all training applications. We expect that they will soon begin to lose market share to ASICs.”
An application-specific integrated circuit (ASIC) is a chip customised for a particular use.
There has been a growth of dedicated AI accelerator ASICs.
According to the report AI generates a lot of data:
“AI applications generate vast volumes of data — about 80 exabytes per year, which is expected to increase to 845 exabites by 2025.”
As I described earlier there is an interplay between software and hardware outside of the direct storage/compute.
Thus it becomes important to keep an eye on what is happening within the semiconductor market as it will affect the overall field of artificial intelligence, and the other way around.
This is #500daysofAI and you are reading article 311. I am writing one new article about or related to artificial intelligence every day for 500 days. My focus for day 300–400 is about AI, hardware and the climate crisis.