Well, NVIDIA has dominated the AI markets since the boom back in 2022, and while the company has managed to get past competition pretty easily, it won't be wrong to say that AMD is significantly ramping up its AI lineup, under the leadership of CEO Lisa Su. And, by the looks of it, the competition is going to get a lot tougher for Jensen moving ahead.
Team Red Didn't See How Big The AI Hype Would Turn Out Back In 2022, But Now, Things Are Expected to Change Drastically
If someone still doesn't know why NVIDIA managed to front-run AI, well, Team Green was one of the first to introduce market-ready solutions for AI chips, while competitors like AMD and Intel were still stuck with CPUs and GPUs. Following the advent of ChatGPT, AMD realized that AI was turning out to be massive, which is why, back in 2023, CEO Lisa Su announced that the company would shift towards AI completely. But by then, NVIDIA had already flooded the market with its Ampere and Hopper solutions, along with a robust software ecosystem, so for AMD, they were entering a one-sided market.
Back in 2023, pivoting away from NVIDIA wasn't an easy job, since Big Tech had contracts with Team Green, which basically made it necessary for them to rely on the firm's AI chips, instead of opting for alternatives. Moreover, with a "proprietary" ecosystem created by NVIDIA's CUDA and other services, customers couldn't opt for the likes of AMD, and under Jensen, a market was created that couldn't rely on anyone else. Well, this led to Team Red facing difficulties with customer adoption, despite announcing solutions like the Instinct MI300X AI accelerators.
AMD's first major AI solution was the Instinct MI300 series, which was claimed to rival NVIDIA's popular H100. It featured twice as memory as NVIDIA's counterpart, on-par memory bandwidth, and more importantly, the CDNA 3 architecture, which promises massive performance gains. Internal results did show that the MI300X poses a viable threat to NVIDIA, especially in inferencing applications, but the key problem here wasn't beating a single entity; it was basically AMD vs a whole ecosystem.
AMD vs NVIDIA's AI Ecosystem; How Competing Against Jensen Isn't Only About Introducing a Powerful Chip
NVIDIA's ecosystem lock-in currently holds back AMD from becoming as mainstream as Team Green. NVIDIA has tight control over developer tools and AI frameworks, positioning the firm as the 'default place' for your AI needs, while AMD emerged as a challenger brand. Since building up AI infrastructure isn't only about acquiring millions of AI chips, it requires adoption into a whole system, or what we call, into Jensen's arena. For Big Tech, switching to AMD was a more expensive venture relative to using NVIDIA's chips, despite Team Red offering higher capabilities.
While many might not agree, investor/media hype often drives company decisions, and in this case, NVIDIA saw a really sweet spot. Big Tech mentioned NVIDIA's equipment in their respective PR statements, which not only boosted investor confidence but also gave companies a massive edge since NVIDIA's chips were seen as the go-to solution. So, in a scenario, if a company like OpenAI or Microsoft announced buying AMD's equipment, the usual retailer would perceive it as a sub-par development, so Team Red did have an image issue as well.
In this case, AMD was left with few choices: either ramp up its solutions or abandon the race entirely, just like Intel did. But Lisa Su chose to keep on competing, which is why AMD is arriving at a position where it is seen on par with NVIDIA in computing solutions. With that, Big Tech is responding accordingly since they need to follow the hype train to keepup with the giant CapEX.
Securing Big Tech Interest: Microsoft, OpenAI, Meta Now Under AMD's Boat
Well, AMD started from ground zero at a point when NVIDIA reached trillions in valuation. After introducing capable products, the company began pitching them to Big Tech by offering competitive prices, higher availability, and robust ecosystems. AMD's Instinct MI300 series is known to be generally 20%-30% cheaper than NVIDIA's counterparts, and offering a much better price-to-performance ratio, which is why firms like Microsoft and OpenAI are known to have employed Team Red's tech stack.
But Big Tech's investments in AMD aren't up to par with the spotlight NVIDIA sees, and that's probably expected, considering that for now, AMD has a lot to do to compete with its arch-rival. If you ask me how this post came into thought, it was probably after AMD's 'Advancing AI' event, since when Lisa Su took up the stage, the developments she announced weren't ordinary at all, especially when you stack them up in 'computing terms'.
The more optimistic announcements are the next-gen Instinct MI400 series, which is set to offer cutting-edge HBM4 technology and 50% higher memory compared to existing solutions. There's a lot to what MI400 will offer, but for a quick summary, it will give Rubin a hard time competing in the market. More importantly, AMD has diverted its attention towards rack-scale solutions, as the firm will introduce its top-end 'Helios' AI server rack as well, which will feature EPYC Venice CPUs onboard, and is claimed to rival against Rubin NVL144 offerings.
I dove into next-gen announcements because they show that on the computing side, AMD is ready with its weapons to combat NVIDIA. But when it comes to cracking into a monopolized market, superior performance won't do much alone. The market needs to ensure that AMD is ready to offer an ecosystem that rivals NVIDIA entirely, from hardware to software. However, another general perception that needs to be ruled out is that AMD doesn't need to replace Team Green entirely, but rather co-exist, similar to what it does with consumer GPUs.
AMD Q2 Earnings & Year-End: A Do Or Die Situation For Lisa Su & Her Team?
AMD is set to report Q2 earnings tomorrow, and that would likely tell us where the company is heading in the future. Analysts expect an increase in YoY revenue, driven by adoption from OpenAI and others, along with demand across multiple segments. In the AI space alone, AMD is expected to show positivity, but when it comes to competing with NVIDIA, you want to be at your absolute best.
