Samsung backs Nvidia AI chip rival in $230 million funding round as GPU alternatives boom - CNBC

Skyler Monroe (0:10)

Hey everyone, welcome back to The AI Hardware Show. I'm your host Skyler Monroe, and this is the show where we dig into the silicon, the servers and the semiconductor breakthroughs that are powering the AI revolution. Big thanks to our sponsors today. Ada, helping businesses integrate AI seamlessly into their workflows. Ago Consulting, that's AIgo, your go-to for silicon development consulting from AI chips, to full-sock designs, and Zen Semiconductor, building the processors and AI fabric behind modern data centers with ventures like their Sierra RISC-V CPUs and Loom AI fabric, plus Ada for deploying AI at scale in any business. Alright, we have a packed episode today. Samsung is throwing serious money behind an Nvidia rival, AMD is making some loud noise in the data center, and Apple and Nvidia are apparently burying the hatchet for something big in 2029

Let's get into it. Okay, first up, Samsung has backed an Nvidia AI chip rival in a $230 million funding round, and this is part of a broader wave of investment into GPU alternatives. Now when you see Samsung writing a check that size, you need to pay attention. Because Samsung doesn't just throw money around casually, they are one of the most vertically integrated semiconductor companies on the planet. They make the chips, they make the merry, they make the displays, they make the phones.

So, when they decide to back someone going after Nvidia's turf, that is a strategic signal, not just a financial one. The story here fits into a much larger narrative that we've been watching build for the last couple of years. Nvidia has an extraordinary grip on the AI accelerator market. We're talking somewhere in the neighborhood of 70 to 80 percent market share in day-centered GPUs. That's a monopoly-level position, and it creates two very predictable reactions from the rest of the industry. First, customers start panicking about supply and pricing power. Second, investors start hunting for the next thing that could challenge it. And right now, both of those reactions are happening simultaneously, which is why you're seeing these massive funding rounds pop up. So what does a GPU alternative actually look like? Because when people hear GPU alternative, they sometimes imagine just a slightly different GPU.

But that's not really what's happening here. A lot of these challengers are purpose-built AI accelerators, sometimes called AI ASICs or neural processing units. Instead of being general purpose parallel processors like a GPU, they're optimized specifically for the kinds of math that dominates AI workloads. Matrix multiplications, Tensor operations, that kind of thing. Think of it like the difference between a Swiss army knife and a chef's knife. A GPU is the Swiss army knife incredibly versatile. A purpose-built AI chip is the chef's knife. It does one thing, but it does it faster and more efficiently than anything else in the drawer.

Samsung's backing here is particularly interesting because it could signal their intent to get more deeply embedded in the AI chip supply chain beyond just being a memory and foundry provider. Right now, Samsung supplies HBM, that's high-bandwidth memory, to AI chip makers, and they have their foundry business. But backing a chip design company gives them a stake in the actual compute layer, not just the supporting infrastructure around it. And with $230 million on the table, this startup has serious runway to actually compete. The GPU alternatives boom is real, it's funded, and Samsung just made sure everyone knows they're paying attention. This one's worth watching closely. Moving on to story 2, AMD has announced new data center GPUs, and the headline from EE Times is calling them warning shots. I love that framing because that's exactly what they are. AMD isn't just releasing products here, they're sending a message to the entire data center market that they are serious, they are competitive, and they are not going away. Now let's give some context here, because AMD's journey in the data center GPU space has been a bit of a roller coaster for years. Nvidia had such a commanding lead with their CUDA software ecosystem that it was almost impossible for AMD to get meaningful traction. Even when their hardware was genuinely competitive on paper. Software motes are real, and Nvidia built one of the deepest in the history of computing. CUDA isn't just a programming language, it's years of libraries, tools, frameworks, and developer muscle memory. Switching away from it is genuinely painful. But here's what's changed. The AI workload explosion has created a new class of customers, hyperscalers and cloud providers who are motivated to reduce their dependence on any single vendor. When you're spending billions of dollars a year on AI compute, you care a lot about having negotiating leverage and supply chain diversity. That makes AMD's raw some software platform, their CUDA alternative, worth investing in seriously, even if the migration is bumpy. So what are these new GPUs bringing to the table? AMD has been steadily improving their Instinct line, and the latest announcements are targeting the exact workloads that matter most right now, large language model training and inference. They're pushing on memory bandwidth, interconnect speeds and compute density.