The list is growing longer.
Cisco is the latest to join the xAI party. This is becoming so large the numbers are garnering attention. All of this is poised to help xAI with its AI endeavors.
What are we talking about?
In this situation, it is infrastructure. The idea of AI being some processors hooked up together is a gross simplification. There is a lot more to it.
The AI Infrastructure Partnership (AIP) includes such noted names as Blackrock, Nvidia, Microsoft, and Global Infrastructure Partners. It is an alliance to build the infrastructure behind AI’s next era.
It is a partnership that is looking at $30 billion, potentially reaching the $100 billion level with debt financing. Each time a new partner signs on, the alliance gets stronger.
xAI’s partnerships have expanded significantly, with 14 technology collaborators. NVIDIA supplies critical graphic processing units (GPUs) powering xAI’s AI models, including Grok and the Colossus supercomputer. Other partners have not been fully disclosed, but likely include cloud computing or hardware providers enhancing xAI’s infrastructure.
Once again, as mentioned in the past, we revert back to infrastructure. Hundreds of billions of dollars is required to build out all the levels of AI. We are not only looking at things such as GPUs (Nvidia) but also networking (Cisco), data integration (Palantir) and energy (GE, NextEra).
Elon Musk recently tweeted his latest pondering: what would it take to create 1 terrawatt of energy. It seems he is looking to pursue this in a way that contracts the time to deployment. After all, these data centers are hungry for energy.
xAI is going to dominate the AI world. Grok3 is already one of the top models, this from a company that is not quite in the game 3 years. The gap that was closed with the likes of ChatGPT is stunning.
By the end of the year, we will have an idea where things stand with the language models. If the speculation of some is correct, the compute that xAI will have at its disposal will usurp the rest of the industry.
Then we have embedded AI. Here again, xAi's collaboration with Tesla means we could be looking at it leading here also. This provides something that Google, Meta, and OpenAI do not have access to.
The benefit of embedded AI is that it will send the amount of data skyrocketing. Of course, this only furthers the need for infrastructure and expanding every layer of the AI stack.
What we are looking at is a race. This is more than xAI and its partners against the other LLM companies. It is also a hedge against China. If this alliance brings $100 billion to the table, that will be a lot of US based infrastructure from which to compete with the Chinese companies.
It is also why getting energy companies involved is crucial.
This is today's supercomputer race. Not long ago, governments were competing to see who could build the largest supercomputer. I believe there are a handful that moved into the exascale range.
The cluster systems that companies like xAI are building will surpass the other giants in size. When we are looking at hundreds of thousands of GPUs networked together, that makes those others look like low end smartphones. Naturally, they serve different purposes but it does show the compute power being implemented.
In this game, size does matter.
