AMD Versus Nvidia: Which AI Stock Can Gain More in the Short Term?

On Monday, Nvidia (NASDAQ: NVDA) stock began trading with its new 10-for-1 stock-split adjustment. While it caused a brief -2.3% drop, NVDA shares rallied and continue to trade sideways at present $120 per share. NVDA stock is now “cheaper” than the company’s long-standing competitor at that price point.

Advanced Micro Devices (NASDAQ:AMD) is trading at $163 per share, 75% above its 52-week low point of $93.11. Year-to-date, AMD is only up 19%, compared to NVDA, which is 151%.

Given that Nvidia’s stock split now acts as a foil against the “missing the boat” mentality, is Nvidia likely to gain further ground, or does AMD hold surprises to catch up?

Shifting Market Domains for NVDA and AMD

Both companies compete in the PC market as their original bread and butter. AMD, in particular, dominates the console gaming market, with the PS5’s GPU hailing from AMD’s RDNA 2 architecture. Likewise, Microsoft’s Xbox consoles use AMD’s CPU and GPU solutions.

This points to AMD’s prowess in the System-on-Chip (SoC) domain, further advanced by the latest accelerated processing units (APUs) in the Ryzen 8000G series. Comparable to the performance of gaming consoles, these single-package chips make a compelling case for cheap mini PCs over buying new consoles.

However, Nvidia outgrew that sector of consumer desktops and laptops by shifting to data centers and high-performance computing (HPC). At the same time, Nvidia retains a lion’s market share in GPUs at 80% vs AMD’s 19%, per Jon Peddie data as of Q4 2023.

Yet, there was a noticeable trend of AMD increasing its market share year-over-year by 7% while Nvidia’s decreased by 2%. Nonetheless, Nvidia more than compensated with massive data center revenue growth of 427% YoY to $22.6 billion.

For comparison, AMD reported for the same Q1 ‘24 period total revenue of $5.47 billion, which is a YoY uptick of only 2%. And of that total revenue, AMD could boast only $2.3 billion in the data center segment.

Can AMD Take the Bite out of the Nvidia-Generated AI Pie?

Clearly, Nvidia’s agility in pushing the H100 chips to train early AI language models was a critical first-mover gambit that paid off substantially.

AMD now has to compete with Nvidia’s dominant 94% market share in the data center segment, per Mercury Research. Nvidia accomplished this feat by pushing its own Ageia PhysX framework for simulations, alongside numerous libraries, CUDA Toolkit, HPC SDK, Modulus, and IndeX.

In other words, Nvidia cracked the formula for seamless development of AI-powered apps in conjunction with the offering of AI chips. On the other hand, AMD’s purportedly more cost-effective MI325X accelerator chips only address the hardware side of the equation. AMD’s Vitis AI framework will then have to rely on the adoption rate of AMD’s new chips.

As of the last earnings call, AMD CEO Lisa Su expects modest growth from data center GPU revenue, from $3.5 billion guidance in January to over $4 billion by the end of 2024. In the meantime, Microsoft (NASDAQ:MSFT) CEO Satya Nadella praised the AMD MI300X series, for its Azure cloud infrastructure, noting that “it offers the best price-performance on GPT-4 for inference”.

On the other hand, Elon Musk prioritized Nvidia chips for his X and xAI companies.

Moreover, it will take a while before AMD catches up with Nvidia’s in-sync releases of GPUs with proprietary CUDA Toolkit versions. As noted, these toolkits have thus far locked in many AI developers and researchers.

For those familiar with the history of Nvidia/AMD’s struggle in the PC gaming market, this mirrors the deployment of Nvidia’s real-time ray tracing (RTX (NYSE:RTX)), deep learning super sampling (DLSS), G-Sync, and other standards to lock in gamers. This approach landed Nvidia in a position to dominate gaming and data center segments.

Which Company Has Better Underlying Tech?

Furthering its unification strategy with APUs, AMD will deploy Embedded+ architecture, which combines Versai AI chips with Ryzen CPUs. Such embedded chips are widely used in automation and robotics. AMD’s next-gen MI325X will feature 3 nm packaging with HBM3E memory, while AMD’s new Zen 5 CPUs (codenamed “Turin”) are rumored to feature up to 192 cores and 384 threads.

Nvidia’s next GPUs, the Blackwell series, will also feature a 3 nm node process owing to TSMC’s cutting-edge foundries. Per Commercial Times’ reporting, Nvidia’s H200 will use 4 nm while B100 adopts the 3 nm process.

Based on these trends, Nvidia and AMD appear neck and neck in the tech arena. This mirrors the two companies’ history, where their products are comparable, but AMD caters to the mid-tier sector with competitive pricing.

This time, Nvidia managed to grab and secure a bigger AI pie with its coordinated CUDA/GPU offering.

Does Nvidia’s Overboughtness Matter?

Nvidia’s price-to-earnings (P/E) ratio was 102 in 2024 (fiscal), estimated to lower to 47 next year. AMD’s P/E was 84 last year, and it is estimated to grow to 64 next year. With P/E ratios for both companies suggesting strong investor confidence, Nvidia’s decreased P/E points to price correction and less risky earnings growth.

Analysts believe that both AMD and NVDA stocks are good picks for AI investment thesis despite AMD’s plunge in May following lower guidance than expected. Nasdaq’s forecasting data places AMD shares at $191 average price target vs current $163. Nvidia’s average price target is $187.76 vs current $120 per share.

This gives Nvidia stock a 56% upside potential compared to AMD’s 17%. In other words, Nvidia’s stock split was the right move at the right time, further boosting investor confidence.