FEATURED

• Goldman Sachs raised its AMD price target from $450 to $640 — one of the largest single-analyst upgrades in the current chip cycle
• Citi added AMD to its upside catalyst watch list the same morning
• Nvidia’s Kyber NVL144 rack-scale system has been pushed to 2028; the fallback design, NVL72x2, was scrapped entirely
• AMD’s Data Center segment hit a record $5.8 billion in Q1 2026 revenue, up 57% year-over-year
• Management guided Q2 2026 revenue to ~$11.2 billion — roughly $700 million above prior Street estimates
• AMD is up approximately 142% year-to-date through early July 2026
• The July 22-23 Advancing AI event in San Francisco is the next major catalyst before Q2 earnings on August 4

Here’s what happened this morning. Goldman Sachs analyst James Schneider raised his AMD price target from $450 to $640 — one of the most aggressive upward revisions from a major Wall Street bank in the current chip cycle. Citi added AMD to its upside catalyst watch list on the same day. Both moves hit before 10 a.m., and the stock responded instantly.

But the analyst calls are almost the least interesting part of today.

The part investors should actually be reading is buried in the Nvidia news. Nvidia’s Kyber NVL144 rack-scale system was reported delayed until 2028 due to manufacturing problems with its PCB midplane — a specialized multi-layer circuit board that holds the whole architecture together. The fallback design, NVL72x2, was scrapped after cloud providers pushed back. For anyone trying to understand why AMD could compound from here, that sentence is doing a lot of work.

And then there’s Turing.

Japanese autonomous driving startup Turing announced that AMD Ventures joined its investor group, and that roughly 10% of its AI training workloads — previously handled entirely by Nvidia hardware — have now migrated to AMD accelerators. That migration matters beyond its size. It signals something the market keeps underpricing: AMD is winning real enterprise AI customers, not just landing cloud hyperscaler contracts that never quite materialize.

Schneider flagged AMD’s July 22-23 Advancing AI event in San Francisco as a potential positive moment where the company is expected to outline the durability of server CPU demand and share more details on Data Center GPU customer adoption. That event now sits directly ahead of Q2 earnings, confirmed for August 4 after market close. Two consecutive data points for the market to chew on.

The financial foundation underneath all of this is real. AMD’s Data Center segment posted a record $5.8 billion in Q1 2026 revenue, up 57% year-over-year, surpassing 50% of total company revenue for the first time. Management guided Q2 2026 revenue to approximately $11.2 billion — representing 46% year-over-year growth — a midpoint that came in roughly $700 million above the prior Street estimate.

The reason analyst price target upgrades keep landing below where the stock already trades is not that analysts are bearish. AMD is up approximately 142% year-to-date through early July 2026, and the sharp rally has pushed the forward price-to-earnings multiple into the mid-to-high 70s. The consensus is playing catch-up. The stock has been moving faster than the models.

Three things are now converging ahead of that August earnings report. The MI450 GPU ramp and Helios rack-scale platform, expected to ship more visibly in H2 2026, are what institutional investors are most focused on. Layer in a competitive delay from Nvidia, a growing customer win list outside of hyperscalers, and a macro tailwind from a weaker-than-expected June jobs reading — and the July 22 event starts to look like a preview, not a peak.

The risk is straightforward. AMD trades at a trailing P/E near 190 and export controls on its MI308 chips to China cost roughly $440 million in net inventory charges last year. A stock priced for flawless execution has no margin for error if August guidance disappoints.

What’s interesting is that the question investors keep asking — has AMD already priced in the upside? — may be the wrong question entirely. The better question is whether the MI450 ramp actually closes the gap with Nvidia in rack-scale infrastructure. Goldman’s $640 target implies shares gain roughly 20% from current levels. Cantor’s bull case is $700. Those aren’t the numbers of a thesis that’s finished.

The July 22 event is the first real test.