Investors tend to love latching on to new trends and finding relationships between companies where success from one can mean equal (or greater) success for another. For instance, there was anticipation headed into Advanced Micro Devices' (AMD 3.04%) second-quarter 2023 earnings release that Nvidia's (NVDA 3.46%) blockbuster success in AI training would be reflected in AMD as well. That didn't end up happening.

The fact is, these things can take time. Nvidia's pioneering work is not going to be easily replicable. AMD's most recent update on its work with AI chips certainly wasn't bad, but it didn't set the world on fire either. Is AMD stock still a buy after the news?

AMD's outlook on AI chips was just OK

As expected, AMD had a messy Q2 2023. Revenue and adjusted earnings per share (EPS) were a respective $5.4 billion (down 18% year over year) and $0.58 (down 45% year over year). Weakness was especially seen in AMD's "client group" -- chips for PCs and laptops -- with sales ($998 million) sinking 54% from the same period in 2022 as excess inventory in consumer electronics continues to weigh on AMD.  

But even the "data center" segment wasn't immune. Sales of $1.3 billion represented an 11% year-over-year decline. Like Intel (INTC 1.28%), AMD's chips for the data center still primarily skew toward traditional servers, which rely more heavily on CPUs (central processing units). This differs from Nvidia (NVDA 3.46%) and its focus on accelerated computing (via its GPUs, or graphics processing units), which are powering the current trend toward generative AI training servers for data centers. 

This isn't to say AMD and Intel have zero exposure to accelerated computing. On the contrary, both companies unveiled new chips for AI to capitalize on the hype of the moment (generative AI, embodied by the ChatGPT service). However, Nvidia was the pioneer of GPU computing and thus has an impressive lead that will not be closed quickly. 

Where Nvidia forecasted an unheard-of $4 billion quarter-to-quarter revenue rise thanks to AI (from Q1 to Q2 of the current fiscal year, a more than 50% sequential increase), AMD is forecasting something far more pedestrian.

Metric Q4 2022 Q1 2023 Q2 2023 Q3 2023*
AMD revenue $5.6 billion $5.4 billion $5.4 billion $5.7 billion
Quarter-over-quarter change 0.7% (4.5%) 0% 6.5%
Year-over-year change 16% (9.1%) (18%) 2.5%

Data source: AMD. *Q3 2023 figures are the midpoint of analyst forecasts for the quarter.

AMD is no stranger to the "second fiddle" role

AMD clearly isn't getting the massive revenue liftoff that Nvidia is, but that isn't to say AI isn't having an early impact. Unfortunately for AMD, it's just well-diversified enough that other segments (its "embedded" chips, mostly from the Xilinx acquisition in early 2022, and "gaming") offset its smaller AI bets. 

But even as the "second fiddle" to Nvidia, AMD is doing well. CEO Lisa Su and the top team said data-center sales in the second half of 2023 are expected to be up 50% compared to first-half sales. That's quite significant. First-half data-center revenue was $2.6 billion, so that implies roughly $3.9 billion in sales in Q3 and Q4. Overall, Su and company think total AMD revenue will be up a high-digit percentage versus 2022's total revenue of $23.6 billion.

Additionally, Su said AMD expects that "the market for AI accelerators [will] reach over $150 billion by 2027" in data centers alone. Estimates indicate this market was in the $20 billion to $30 billion ballpark in 2022. Forecasts are all over the board right now, so suffice to say there's limited ability to accurately predict where this AI hype machine is headed.

For now, AMD is trying to catch Nvidia with its chip and AI software development, and that's an OK place to be. The top chip designer is still in the midst of a deep slump in consumer electronics, so trailing-12-month earnings aren't meaningful. However, based on initial 2024 estimates, Wall Street analysts think AMD's profitability will rally. Shares trade for about 26 times those earnings expectations. 

AMD PE Ratio Chart

Data by YCharts.

I'm happy to hold the AMD position I already own. For investors trying to figure out the best way to invest in the next five to 10 years of computing innovation, AMD isn't a bad route to go. Just mind the risk of chasing the market's high level of optimism right now and consider using a dollar-cost averaging plan.