Since 2022, everyone is talking about Artificial Intelligence. Every week we find new models, more capable than the previous ones to create documents, extract data, answer questions, generate images, videos or whatever comes to mind.
The main beneficiaries of Artificial Intelligence, at least economically, are, as expected, large graphics card manufacturers like NVIDIA and AMD. Intel, for now, still has a long way to go if it wants to be a real alternative in the GPU market for training artificial intelligence.
According to Bloomberg, the Philadelphia Stock Exchange Semiconductor Index saw a significant rise throughout 2023. growth of more than 65%which is the biggest of the last decade and which only confirms everything related to semiconductors, especially with graphics dedicated to AI, they have a fantastic future ahead of them because as the years go by it will become a fundamental part of many jobs.
It is not a question of replacing them but rather of complementing them and we find a clear example in Microsoft’s Copilot. One of the uses of this AI is to prevent workers from wasting work hours creating documents based on others, since it takes care of extracting the data and writing the corresponding document, making the workers and businesses are much more productive.
Continuing the news, throughout 2023, NVIDIA has entered the select club of companies with more than one capitalization in billions of dollars, where we also find giants like Apple, Microsoft, Amazon and Google. Precisely, with the exception of Apple, it is these companies that have allowed NVIDIA and AMD to get where they are today, thanks to the GPUs that they offer for their data centers, although some companies , like Microsoft, are considering the possibility of creating their own GPUs with the help of AMD to reduce the cost of the annual investment made through NVIDIA.
NVIDIA not only dominates the gaming graphics card market, but also dominates the gaming graphics card market. GPUs to train AIwith an 80% share thanks to the H100 GPU, with more than 3 million units sold throughout 2023. Although AMD’s presence in this area is not so well known, that does not mean that it don’t bet on it.
A clear example can be found in the two GPUs launched by AMD in the middle of last year, GPUs aimed at becoming a real alternative to NVIDIA in large data centers through the Instinct MI300X and Instinct MI300A APUs, which is why Microsoft is studying the possibility of a partnership. with AMD to create its own GPUs, as we mentioned above.
According to various estimates, the market for GPUs focused on training artificial intelligence by 2027 will be approximately 400 billion dollars. Intel, which currently does not offer any similar solutions for data centers, must get its act together if it does not want to be completely excluded from this lucrative market that NVIDIA monopolizes with 80% market share.
