The move to AI is, like most big technology waves, progressing incrementally. The first step is the broad introduction of a product that pivots the market, like Microsoft’s ChatGPT-based Copilot. Ironically, due to Microsoft execution issues, OpenAI and NVIDIA are becoming the biggest initial beneficiaries. The second phase that we’re in now is a rapid build-up of services capabilities to provide this recent technology to companies and individuals. This will be followed by an even larger feeding phase where end user companies change their hiring priorities to focus on a by-then-depleted pool of talent. This is generally done in conjunction with a major change in trade school, college, university and internal training to fill what will likely be massive shortages of AI talent across the tech ecosystem.
AMD’s purchase of Silo AI is a clear example of Phase 2. The company has an impressive 120 AI experts out of its three hundred total employees and has already been working with AMD to build out its new AI service and software competency.
Let’s talk about this acquisition and what it means for AMD this week.
Silo AI
Silo AI isn’t an AI creator, it is an AI implementor, which is consistent with Phase 2 of a technology roll out. It has already completed over two hundred successful implementations. Since the majority of initial AI implementations are now seen as failures, this is a significant advantage.
In a world that desperately needs capable people with AI experience to assure AI implementations are done correctly and meet their design goals and/or exceed expectations, Silo AI is unique in its capabilities and accomplishments.
Teams of this size and capability rarely emerge this early, but NVIDIA and others have been working on AI for decades, so while most were caught unprepared, these folks were ready. I can’t imagine what a team like this would be worth, but what is extremely clear is that whatever AMD paid for Silo AI, it was a hell of a bargain.
What This Does for AMD
AI isn’t really about the hardware. It’s about the solution the hardware is part of. To build better hardware and craft the software that will make it work, you need deep expertise not only of AI but what end-user customers want to do with it. NVIDIA highlights this at every event, and its two decades-long AI effort has made that company the current king of AI.
To even the playing field, let alone keep up, a challenging company like AMD needs to rapidly gain deep insights into what the market needs and specifically what its competitors are missing. An entity like Silo AI can provide this kind of information in a form that can be understood by the hardware and software engineers building their parts of an AI solution, and with that deep understanding of both, it can help enterprise buyers become even more successful than Silo AI could be by itself.
Wrapping Up:
This is one of those rare occasions where the sum of the whole is greater than the sum of the parts. Silo AI will massively accelerate AMD’s AI efforts while giving it an implementation resource that is both effective and rare, and eventually turn AMD into a far stronger challenger to NVIDIA. This will also be good for NVIDIA because companies can lose focus if they lead too long without challenge.
And given that the current demand for successful AI implementations exceeds the combined capabilities of both NVIDIA and AMD, this shouldn’t hurt NVIDIA at all. It should focus NVIDIA more tightly on its own customer efforts, as well, with the benefit being far more successful implementations from both firms.
In the end, we are seeing the birth of a more powerful AI resource at the exact time the world desperately needs it.