Is AMD The Better Bet Versus Nvidia?

Nvidia (NASDAQ:NVDA) has been the clear front runner in the generative artificial intelligence race, with its stock surging 180% this year propelling it to become the world’s most valuable company with a market cap of almost $3.5 trillion. However, trading at a lofty valuation of nearly 48x consensus FY2025 earnings, future gains could be harder to come by for Nvidia stock. In contrast, AMD presents a more compelling opportunity. Trading at a more reasonable 28x forward earnings, AMD stock (NASDAQ:AMD) is poised to benefit from the long-term growth of AI while offering better value. While the AI revolution is possibly just getting started, investors will need to pick the right winners to continue profiting from this trend. Specifically, we think it might be time to reconsider Nvidia stock and look closely at AMD. And it’s not just AMD’s more attractive valuation – trends like a growing focus on cost-effectiveness by end customers and shifts in the model training landscape could also work in AMD’s favor.

AMD stock has fared well over the last 4-year period, rising from levels of about $90 in the beginning of  2021, to highs of over $200 earlier this year. However, the gains have been far from consistent, with annual returns being considerably more volatile than the S&P 500. Returns for the stock were 57% in 2021, -55% in 2022, and 128% in 2023. In contrast, the Trefis High Quality (HQ) Portfolio, with a collection of 30 stocks, is considerably less volatile. And it has outperformed the S&P 500 each year over the same period. Why is that? As a group, HQ Portfolio stocks provided better returns with less risk versus the benchmark index; less of a roller-coaster ride as evident in HQ Portfolio performance metrics.

Lower AI Training May Hurt Nvidia 

Companies have devoted immense resources to building AI models and these models have been getting steadily larger in terms of the number of parameters, a key measure of model size and complexity. Training these massive models is more of a one-time affair that requires considerable computing power and Nvidia has been the biggest beneficiary of this, as its GPUs are regarded as the fastest and most efficient for these tasks. This is very evident from Nvidia’s revenue growth trajectory. Sales are on track to grow from a mere $27 billion in FY’23 to almost $130 billion in FY’25. However, as models grow larger, there are signs that their capabilities and accuracy may not improve in proportion to the incremental investments. This could cause customers to rethink their spending plans.

There’s another factor that could cause AI-related training to decelerate. The availability of high-quality data for training large language models could become a bottleneck. Much of the high-quality text and other content that’s available readily on the Internet has likely already been internalized by these general-purpose AI models, limiting easy opportunities for further model expansion. This could push companies away from brute-force scaling.  There’s a possibility that companies will shift toward smaller, specialized models tailored to specific organizations or tasks, and this could impact demand for Nvidia’s high-powered GPUs. This could mean that the big boom Nvidia saw over the last three years was front-loaded.

Shift To Inference Benefits AMD

Given the above constraints, the AI market at large could gradually shift from the process of model training to inferencing, which is the process of generating outputs and answers from trained models. The process of inferencing is less computationally intensive, and it’s very much possible that these talks can potentially be handled reasonably well by AI processors from the likes of AMD. While Nvidia has indicated that inference accounts for roughly 40% of its data center revenue, the company could face mounting competition.  MLCommons, a testing organization, indicated that AMD’s MI300X is very competitive with Nvidia’s H100 GPU on AI inference benchmarks. This competitiveness is translating into real business wins.  There have been reports that AMD has been getting some big orders for the MI300X from hyperscalers for AI inference workloads. For example, IBM has indicated that AMD’s MI300X AI silicon will be available on IBM Cloud services during the first half of 2025. Can Nvidia Stock Lose 50%? 

Customer Focus On Costs Can Hurt Nvidia Margins

Costs are also becoming a concern for end customers of AI chips. The economics of the AI business remain weak, with heavy investments in GPU chips yielding minimal revenue. During the initial wave of AI, enterprises and big tech companies scrambled to invest in AI due to the fear of missing out, rather than focusing on returns. This led to a surge in pricing power for Nvidia, with its net margins coming in at over 50% in recent quarters. However, companies and their investors will eventually look for returns on their investments meaning that they could become more judicious about AI costs going forward. This trend could benefit AMD and potentially newer entrants such as Intel, as customers look for more affordable alternatives to build out data centers quickly.  With Nvidia’s GPUs costing upward of $25,000 each, customers are actively seeking more cost-effective alternatives. For instance, Oracle recently chose AMD’s accelerated computing chips to power its latest supercluster for high-intensity AI workloads, after testing showed that AMD’s GPUs delivered low latency and strong performance at a competitive price.

AMD’s Lower Valuation

As we mentioned before, AMD also has a more attractive valuation. The stock trades at just about 28x projected FY’25 earnings, compared to Nvidia’s lofty 48x multiple. This makes AMD a more reasonable bet, especially as earnings expand, driven by its growing CPU market share and the long-term growth potential of the AI market. While AMD’s overall growth has been slower than Nvidia’s, the company is performing well where it matters. Over Q3, earnings and revenues came in ahead of expectations, driven by strong demand for AI chips. Notably, the data center business – which is the company’s most closely watched segment – posted $3.5 billion in sales, up 118% compared to last year. This could make AMD a more favorable play in the current AI market.

[1] Returns as of 11/27/2024
[2] Cumulative total returns since the end of 2016

Invest with Trefis Market-Beating Portfolios
See all Trefis Price Estimates