Nvidia (NVDA +1.63%) crafted an incredible business model over the past two decades. In 2006, the company unveiled its CUDA developer platform, which included a programming interface, compiler, driver, runtime environment, and toolkit. Nvidia gave developers a customized ability to program its GPUs to accelerate applications, giving Nvidia's chips an advantage over the competition.
By the late 2010s, as deep learning became more popular, CUDA already was the go-to standard for GPU acceleration. And since most deep-learning tools were specifically designed for CUDA, this made it hard for users to switch to other GPUs, effectively locking them into using Nvidia's products.
It's this dynamic that has helped Nvidia become a $3 trillion business. But as we've seen in past cycles, competition for GPUs will slowly heat up. Over time, another company could end up even bigger than Nvidia.
NASDAQ: AMD
Key Data Points
This chipmaker is catching up with Nvidia
Advanced Micro Devices (AMD +1.63%), colloquially referred to as AMD, is an afterthought in the race for artificial intelligence (AI) GPU dominance. According to most estimates, Nvidia has at least an 80% market share for AI GPUs, though some estimates reach 90% or higher. Suffice it to say, when it comes to supplying the AI industry with critical components like GPUs, Nvidia is indisputably king.
But let's take a look back at previous chip cycles to see how things have played out over not just months or years, but full decades at a time. In 2006, Intel had a dominant lead in graphics chips, with a 40% market share. AMD took second place with just over 25%, and Nvidia came in at just under 20%.
One year later, Intel still held the lead position with a roughly 39% market share, but AMD and Nvidia swapped places. Nvidia now commanded an almost 30% market share, with AMD slipping under 20%.
Here's another example. In 2021, Intel held a 64% market share for data center chips. Nvidia came in at 27%, with AMD at just 9%. Two years later, Intel had fallen to just 26%, with Nvidia soaring to 66% and AMD still languishing at 8%.
The point here isn't that these shifts will translate cleanly to today's AI GPU environment, but that throughout history, chip leads have traded hands many times, even over fairly short periods of time.
Which companies will eventually catch up with Nvidia? The future remains unknown, of course, but AMD is investing heavily to compete on AI over the long term. Last quarter, the company released its MI325X AI accelerator chip, which will compete directly with Nvidia's H200 GPUs. AMD also recently unveiled its next-generation MI350 chips, which should compete with Nvidia's next-generation Blackwell chips by the middle of 2025.
AMD's CEO Lisa Su has said that she'd like the company to become the "end-to-end" AI leader over the next decade. "This is the beginning, not the end, of the AI race," she told investors late last year. But can AMD actually catch up to Nvidia?
NVDA Revenue (TTM) data by YCharts
Will AMD dominate AI GPUs by 2030?
Nvidia's dominance in the AI GPU market is not necessarily generated by better chips. The company arguably has superior chips versus the competition right now. But over time, the combined research and development budgets of the world's competing chipmakers will likely narrow this lead, and eventually topple it.
What makes Nvidia's products so special is that, as AMD's CEO hopes will one day be true for her company, Nvidia really is an "end-to-end" provider, controlling both the hardware and software components, leading to exceptional lock-in with developers and businesses.
But AMD is clearly making inroads with customers with its latest generation of chips. Microsoft and Meta Platforms recently signed on as customers for its current generation of MI300 AI GPUs. According to AMD's own estimates, the total addressable market for AI chips will reach $400 billion by 2027. Considering both AMD and Nvidia combine for just over $30 billion in AI data center chip sales today, there should be plenty of space for AMD to grow considerably, even if Nvidia remains on top for years to come.
While possible, it'll be difficult for AMD to overtake Nvidia over the next five years. But that doesn't rule out AMD as a promising AI GPU investment.
