By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panel's discussion on AMD's valuation and growth prospects is mixed, with concerns about execution risks, demand normalization, and potential capex slowdowns, but also optimism about the MI450's performance and market share gains in AI accelerators.
Risk: The single biggest risk flagged is the potential underwhelming utilization of AI capex and real usage, which could prevent AMD from translating the MI450/Helios ramp into meaningful revenue or margin uplift.
Opportunity: The single biggest opportunity flagged is AMD's potential to capture a significant portion of the AI accelerator market share from Nvidia, driven by the MI450's performance and the demand for infrastructure efficiency from hyperscalers.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
On June 3, semiconductor giant Broadcom posted extremely strong operating results for its fiscal second quarter, but management's sales guidance came in lighter than Wall Street's expectations. This sparked a sell-off of practically every semiconductor stock, but those with high exposure to the artificial intelligence (AI) revolution were particularly hard hit.
Advanced Micro Devices(NASDAQ: AMD) is one of the world's top suppliers of graphics processing units (GPUs) for data centers, which are the main chips used to handle artificial intelligence training and inference workloads, and its stock is down 10% from last week's all-time high as I write this.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
However, AMD is still sitting on an eye-popping 12-month return of 300%. So could the recent dip be a buying opportunity, or have investors missed the boat?
AMD is preparing to ship its best AI chips so far
AMD entered the data center GPU race in 2023 with the MI300X. It was designed to compete with Nvidia's H100, which was the leading-edge product at the time. AMD has since launched several new generations, including the MI350 and MI400 series, each more powerful than the last, and they have attracted major customers such as OpenAI, Oracle, and Meta Platforms.
But later this year, the company will start shipping commercial volumes of its MI450 series processors, which are its most powerful yet. These GPUs can be configured in AMD's new Helios data center rack, which includes highly specialized networking components and software to extract maximum processing speeds. In fact, the company says the MI450 series can deliver up to 36 times more performance than its previous-generation GPUs, thanks to the fully integrated Helios architecture.
AMD CEO Lisa Su says several major customers are inquiring about large-scale deployments for the MI450 platform. So far, OpenAI and Meta have each signed deals to deploy 6 gigawatts of computing capacity using AMD's chips in the coming years, and they will start with the MI450.
Data center revenue growth is expected to accelerate
AMD's data center segment generated $5.8 billion in revenue during its fiscal 2026 first quarter (which ended March 28). That was a 57% increase from the year-ago period, and made up more than half of the company's total revenue of $10.3 billion.
However, Su believes AMD's data center revenue could grow by more than 80% per year starting in 2027 as MI450 shipments ramp up. In other words, this segment is about to become an even bigger part of AMD's business, which explains why its stock sank last week after Broadcom's data center guidance hinted at a slight potential demand slowdown.
AMD stock is a difficult buy because of its sky-high valuation
Demand for AI data center chips exceeds supply, creating an imbalance that gives AMD significant pricing power, boosting its profit margins. As a result, the company's adjusted (non-GAAP) earnings soared by 43% to $1.37 per share in the first quarter.
However, based on its trailing 12-month adjusted earnings of $4.58 per share, its stock is trading at a price-to-earnings (P/E) ratio of 108.7, making it nearly 3 times as expensive as Nvidia, which has a P/E of 36.1. A premium of that magnitude is hard to justify for AMD, considering that not only is Nvidia the leader in the data center hardware market, but its sales are also growing much faster than AMD's -- hence it's still extending its leadership position.
According to Wall Street's consensus forecast (provided by Yahoo! Finance), AMD will grow its earnings to $13.08 per share in 2027, giving its stock a forward P/E ratio of 38.1. While that seems like a more reasonable valuation, it's still higher than Nvidia's current trailing P/E, so investors might simply be pricing AMD's expected growth too aggressively at the moment.
For that reason, I don't think it would be a good idea to pile into AMD stock at this point. Interested investors might do better to wait for a deeper pullback.
Should you buy stock in Advanced Micro Devices right now?
Before you buy stock inAdvanced Micro Devices, consider this:
The Motley FoolStock Advisoranalyst team just identified what they believe are the10 best stocksfor investors to buy now… andAdvanced Micro Deviceswasn’t one of them. The 10 stocks that made the cut are built for long-term growth and could produce monster returns in the coming years.
Consider whenNetflixmade this list on December 17, 2004... if you invested $1,000 at the time of our recommendation,you’d have $439,038!* Or when Nvidiamade this list on April 15, 2005... if you invested $1,000 at the time of our recommendation,you’d have $1,277,804!*
That performance is why people listen. With a track record ofbeating the S&P 500 by nearly 5x,Stock Advisoroffers a distinct advantage. Don't miss the latest top 10 list, available withStock Advisor, and join an investing community built for the long haul.
Anthony Di Pizio has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Advanced Micro Devices, Broadcom, Meta Platforms, Nvidia, and Oracle. The Motley Fool has a disclosure policy.
Four leading AI models discuss this article
"AMD's valuation already prices in aggressive AI-driven revenue growth that may not materialize if demand cools, Nvidia retains dominance, and MI450/Helios ramp faces execution and pricing risks."
Despite AI tailwinds, AMD trades at a rich multiple and faces meaningful risks the article glosses over. The MI450/Helios ramp hinges on execution, supply discipline, and durable customer pull; pricing power depends on continued chip scarcity, but any demand normalization could compress margins. Nvidia remains a dominant force, and AMD’s upside hinges on market-share gains rather than pure volume. The 2027 ~80% data-center growth assumes aggressive ramp timing and utilization that may prove optimistic. With a forward P/E near 38x, the valuation embeds aggressive growth that could disappoint if AI capex slows or supply/demand cools.
However, the bull case argues that MI450+Helios could unlock meaningful pricing power and durable enterprise deals, and that AI compute demand remains in early innings, offering upside surprises. Still, this is contingent on flawless execution and no material pullback in AI capex.
"AMD's current valuation is a growth-stage premium justified by the upcoming MI450 cycle, not an overextended bubble."
The article’s focus on a 108x P/E ratio is a rearview-mirror metric that ignores the massive operating leverage inherent in AMD's transition to the MI450. While Broadcom’s guidance caused a sector-wide jitter, AMD’s valuation reflects a 'catch-up' premium rather than a structural flaw. If the MI450 delivers on its 36x performance claim, AMD isn't just selling chips; it’s selling infrastructure efficiency that hyperscalers like Meta and Oracle are desperate for to lower their TCO (Total Cost of Ownership). The forward P/E of 38x for 2027 is actually quite conservative if they capture even 15% of the AI accelerator market share from Nvidia.
If the hyperscalers successfully develop their own custom silicon (ASICs), AMD’s GPU margins will face a brutal commoditization trap that no amount of performance gains can offset.
"AMD's 2027 earnings forecast assumes sustained pricing power and market share gains that Broadcom's guidance just signaled may not materialize at the assumed pace."
AMD's 108.7x trailing P/E is indefensible on current earnings, but the article conflates two separate problems: valuation and demand. The MI450 ramp with 36x performance uplift and 6GW commitments from OpenAI/Meta are real. The issue is whether $13.08 2027 EPS is achievable or fantasy. If AMD captures even 20% of incremental AI capex growth (vs. Nvidia's 80%+), that number compresses by 30-40%. The Broadcom miss signals potential capex moderation—not collapse, but slower than the article's 80%+ growth assumption. AMD is pricing in flawless execution and sustained supply-demand imbalance through 2027.
The article ignores that Nvidia's software moat (CUDA ecosystem) and architectural lead on inference (where margins are thinner) may widen, not narrow, despite AMD's hardware specs. Specs don't win markets; ecosystem lock-in does.
"MI450 customer commitments and 80%+ data-center growth from 2027 justify a valuation premium that the market is currently underpricing."
The article flags AMD's 108x trailing P/E and 38x forward multiple as excessive versus Nvidia, yet underplays the MI450/Helios ramp starting late 2025. With OpenAI and Meta already committing to 6 GW deployments and management guiding >80% data-center growth from 2027, AMD could close the gap faster than the 57% Q1 print suggests. Broadcom's softer guide may reflect near-term digestion rather than structural demand weakness. Pricing power in a still supply-constrained AI market supports margin expansion that the current multiple does not fully discount.
Nvidia continues extending its lead with faster sales growth and ecosystem lock-in; any delay in MI450 volume or lower-than-expected attach rates for Helios networking would leave AMD's premium valuation exposed to rapid de-rating.
"The MI450 upside requires proven, durable utilization from 6 GW commitments; without that, the 2027 EPS and forward multiple risk a meaningful re-rating."
Notable risk: the 6 GW commitments from OpenAI/Meta and the implied >80% data-center growth hinge on sustained, flawless AI capex and real usage. If utilization underwhelms, or hyperscalers defer build-out or switch to own ASICs, AMD may not translate the MI450/Helios ramp into meaningful revenue or margin uplift. Until those commitments prove up with durable attach rates, the 2027 EPS target and forward multiple look like a sentiment-driven bet rather than a cash-flow plan.
"AMD's success depends on the Helios networking stack's ability to break Nvidia's NVLink ecosystem lock-in, not just raw compute performance."
Claude is right that the software moat is the true bottleneck, but everyone is ignoring the 'Networking' variable. The Helios ramp isn't just about GPU performance; it's about AMD’s attempt to displace Broadcom and Nvidia in the data-center fabric. If AMD can’t bundle their networking stack with the MI450, they remain a commodity supplier. The 6GW commitment is meaningless if they can't solve the interconnect bottleneck that keeps hyperscalers tethered to Nvidia's proprietary NVLink ecosystem.
"Custom silicon adoption by hyperscalers poses a structural threat to AMD's 2027 EPS that the panel has underweighted."
Gemini's networking angle is underexplored but overstated. Helios matters, but AMD's real bottleneck isn't displacing Broadcom—it's whether hyperscalers actually standardize on open interconnects or double down on proprietary stacks. The 6GW commitments don't specify networking architecture. More critically: nobody has addressed that Meta and OpenAI's ASIC roadmaps (Trainium, Grok silicon) directly threaten GPU attach rates. If hyperscalers shift 30% of workloads to custom silicon by 2027, AMD's EPS target collapses regardless of MI450 performance.
"6GW commitments already bake in ASIC share that compresses AMD's MI450 attach rates below current modeling."
Claude flags ASIC roadmaps correctly but misses the timing: Meta and OpenAI's 6GW pledges already embed hybrid GPU-plus-custom-silicon plans through 2027. If 25-30% of those deployments shift to Trainium or equivalent by ramp, the >80% data-center growth and $13 EPS target both require upward revision of attach rates that current guidance never disclosed. This makes the forward multiple even more exposed than the networking bottleneck Gemini highlighted.
The panel's discussion on AMD's valuation and growth prospects is mixed, with concerns about execution risks, demand normalization, and potential capex slowdowns, but also optimism about the MI450's performance and market share gains in AI accelerators.
The single biggest opportunity flagged is AMD's potential to capture a significant portion of the AI accelerator market share from Nvidia, driven by the MI450's performance and the demand for infrastructure efficiency from hyperscalers.
The single biggest risk flagged is the potential underwhelming utilization of AI capex and real usage, which could prevent AMD from translating the MI450/Helios ramp into meaningful revenue or margin uplift.