AMD AI Event 2026: New Chips, 400% EPS Growth Incoming
AMD's Advancing AI 2026 event hits San Francisco today. Wall Street sees 400% YoY EPS growth for Q2. Here's what traders need to watch before August 4.
Key Points
- AMD hosts its Advancing AI 2026 event in San Francisco today while Wall Street forecasts Q2 EPS of $1.35 — a 400% year-over-year gain versus $0.27 in the prior-year period — when it formally reports August 4.
- Bank of America flagged strong AI server demand ahead of the event and suggested AMD's next earnings report could beat consensus and land with raised guidance, making today's product demonstrations a forward-looking signal for the August print.
- AMD slid 1.8% in pre-market Wednesday alongside the broader chip selloff, but the San Francisco event gives the company a rare chance to set its own narrative before Intel's Thursday report resets sector sentiment.
AMD slides into its Advancing AI 2026 conference in San Francisco today down 1.8% in pre-market trading, bleeding alongside Intel's 2.7% decline and Nvidia's 0.9% drop in a chip sector that has turned defensive ahead of tonight's Alphabet print and Thursday's Intel earnings. The timing is either terrible or perfect, depending on what AMD's product team shows the room — because with Q2 formal earnings due August 4 and Wall Street already modeling a 400% year-over-year EPS gain, today's event is the last major product signal before a print that could define AMD's second half.
The 400% EPS Setup
The numbers AMD is walking toward on August 4 are striking even by the elevated standards of this earnings season. Analysts expect Q2 EPS of $1.35 versus $0.27 in the year-ago quarter — a 400% year-over-year improvement driven almost entirely by AI accelerator revenue as hyperscalers diversify their GPU purchasing away from Nvidia's near-monopoly position. Revenue consensus is tracking at levels that would represent AMD's strongest top-line growth since its EPYC server CPU line began displacing Intel in data center racks two years ago.
Bank of America's pre-event commentary this week flagged strong AI server demand as a structural tailwind and indicated that AMD's August print could beat current consensus and arrive with guidance raised above street estimates. That is a meaningful call in a market where IBM's infrastructure miss last week demonstrated exactly how quickly elevated expectations become execution traps. Goldman Sachs analyst James Schneider carries a Buy rating with a $640 price target on AMD, specifically citing rising CPU demand as a driver distinct from the GPU narrative — which matters because it means AMD's upside case has two independent engines rather than one.
The macro environment cuts both ways for AMD today. The 10-Year Treasury yield sitting at 4.6% and SOFR at 3.57% keep valuation pressure on high-multiple growth names across the board. AMD's forward multiple has compressed meaningfully from its 2025 peak but remains elevated relative to historical norms — meaning the stock needs the revenue growth to keep arriving on schedule. Any softness in today's product demonstration cadence, or management commentary suggesting AI server order timing is shifting, would hit the stock disproportionately given where semis are trading heading into a data-dense 48 hours.
The Nvidia Shadow and the Differentiation Play
AMD cannot escape the Nvidia comparison, but today's event is an opportunity to sharpen the differentiation story in front of customers and investors simultaneously. Nvidia's Q1 FY2027 results — reported May 20 — showed non-GAAP EPS of $1.87 on $81.61 billion in revenue, up 85.2% year-over-year, with Q2 guidance targeting an unprecedented $91 billion. Nvidia trades at approximately 43x forward earnings and 21.5x EV/Sales. AMD, even after its own re-rating, offers a materially lower entry multiple for what bulls argue is comparable AI revenue exposure with CPU upside layered on top.
The competitive reality is more nuanced than the bull case suggests. Nvidia's CUDA software ecosystem remains the dominant moat — not the hardware itself — and AMD's ROCm platform has made genuine progress but has not closed the gap at the scale that would push hyperscalers to shift meaningfully above their current AMD allocation percentages. Today's Advancing AI event in San Francisco is where AMD has a chance to demonstrate whether its software story has matured enough to win incremental share, or whether the growth is primarily coming from capacity constraints in Nvidia's supply chain rather than genuine platform preference. Those are very different businesses at different risk profiles.
The semiconductor sector's broader context sharpens the stakes. TSMC reported a sharp profit beat last week but fell 3.64% the following Friday as the U.S. chip selloff spread to Asian markets. SoftBank dropped 9.2%, Tokyo Electron lost 9%, and Advantest slid 9.4% in a coordinated pullback that coincided with Chinese startup Moonshot AI unveiling a model reportedly rivaling Anthropic and OpenAI's top offerings. The Moonshot development matters for AMD specifically because it raises the question of whether frontier AI model training — AMD's highest-value accelerator use case — could face structural demand moderation if highly efficient models require less raw compute per training run. That risk has not been priced into the $1.35 Q2 EPS consensus, and it is a question AMD management should expect on August 4.
What Today's Event Needs to Deliver
Intel's Thursday print after the close is the immediate competitive reference point AMD cannot ignore. Intel is down 33% for the month — a number that reflects both execution concerns and the broader chip rout — with analysts expecting Q2 EPS of $0.22 on $14.4 billion in revenue, an 11.6% year-over-year gain. If Intel delivers a credible Terafab data center update and demonstrates that the 18A node is gaining external foundry customers, it rehabilitates the entire alternative-to-Nvidia narrative that AMD also benefits from. If Intel disappoints again, AMD looks relatively stronger but the sector multiple compresses regardless.
For AMD specifically, traders should watch three things out of the San Francisco event today. First, any new AI accelerator product announcements with concrete performance benchmarks against Nvidia's H200 and B200 — specific numbers, not directional claims, because the market has learned to discount directional claims. Second, customer reference wins from named hyperscalers or enterprise accounts, which would validate Bank of America's demand commentary with real-world data points. Third, any language from CEO Lisa Su about supply chain capacity and delivery timelines — AMD's AI accelerator backlog has historically been capacity-constrained, and confirmation that supply is scaling to meet demand is the operational proof the August guidance raise needs.
Goldman Sachs analyst James Schneider rates AMD a Buy with a $640 price target, but the stock needs today's event to build the product credibility that bridges from current price levels to that target over the next two quarters. The August 4 earnings date is six weeks out — long enough for the San Francisco narrative to either compound into a pre-earnings run or fade if the demonstration falls short of the product cycle expectations Bank of America's commentary has helped set. Watch AMD's tape closely during and after the San Francisco event: a move above the pre-market opening level on strong session volume would signal that institutional buyers used today's chip selloff as an entry point rather than confirmation of a trend break. A failure to recover through the session puts the $580 technical support level in focus before Thursday's Intel print resets the chip sector narrative entirely.
The Weekly Investor
Daily market analysis for active traders. Free.
