AMD: The $460 Trigger Is Now 3% Away.

On Friday AMD gapped up to $510.27, traded as high as $515.62, and then crashed intraday to close at $476.15 — a 7.5% reversal in one session from the open. That is not how a stock behaves when investors are comfortable with the setup. It is how a stock behaves when a big binary catalyst is forty-eight hours away, and nobody wants to hold the wrong position through it. Tuesday's Q2 print is that catalyst. The question for me is not whether AMD beats or misses — it is whether the price action puts the entry I defined within range.

The Call

HOLD. The stock is at $476.15, 3.4% above my ACCUMULATE line at $460. The setup that earned a BUY from every memo since July 22 is still intact. But the price is not yet at the level where the risk-reward tilts clearly toward adding. If the Q2 print drops the stock through that threshold, the memo-generated thesis still says to fill the order.

Why Now

This issue is the opposite of a pre-earnings guess. I don't know whether Q2 data-center revenue is above or below the whisper number, and neither does anyone else who isn't on the Q2 earnings call prep deck. What I know is this:

1. The gap-and-reversal on Friday was a signal about positioning, not fundamentals. AMD opened up on buying that could have been algo short-covering or Core Scientific deal follow-through, and sold off all day on heavy volume (26.6M shares vs. the 10-day average of ~35M). The fact that the low of $475.76 held near the prior support from Wednesday's rout tells me there is a bid — but not enough of one to hold a pre-weekend position. That is nervous money.

2. The $460 trigger is now close enough to execute on, with a catalyst to do it. Two weeks ago I said ADD below $450 (July 26 memo, confirmed July 31 issue). Three days ago I raised it to $460. The stock has closed four consecutive sessions between $476 and $485. A 5-6% post-earnings move in either direction is entirely plausible for AMD. If the move is down, my trigger fires and I accumulate. If the move is up, I wait — the thesis hasn't changed, but the price has.

3. Q2 earnings are Tuesday, August 4. The first hard data since the AMD-Anthropic and Core Scientific announcements. This quarter matters not because the past is exciting — it matters because management's forward commentary is the first time they will address both deals on the record. Revenue and EPS are backward looking. The guidance and ramp commentary are what the market is pricing.

The Evidence

Re-upping the figures that carry the argument, with today's mark:

The price move since both deals: $521.95 → $429.56 (selloff) → $485.39 (Core Scientific bounce) → $476.15 (Friday reversal). The market has moved the stock 19% peak-to-trough and then oscillated in a 10% band around the middle of that range, waiting for Tuesday. The deals themselves have not been called into question. The only question is what the stock is worth once the cards are face-up.

From the July 25 and July 31 memos, the key assumptions that still hold: - ROCm-Claude optimization remains a multi-month project — no quick fix, but a genuine intent to narrow the CUDA gap. - TSMC capacity is the binding constraint on MI450 — this was true before the deals and is still true. Accelerating that is what the Core Scientific partnership buys. - The "up to" language matters. Both the 2 GW Anthropic and 2.5 GW Core Scientific figures are ceilings, not firm commitments. That ceiling risk is priced into the stock at each level of the volatile range.

Levels & Triggers

Action table:

Confirming the thesis: - Q2 data-center revenue growth and any explicit mention of MI450 customer commitments beyond Anthropic. - Management commentary on 2027 capex and whether the $5B Anthropic equity investment changes the capital allocation plan. - ROCm adoption metrics (even directional) on the call.

Kill condition: If Q2 data-center revenue is flat or down sequentially while MI300X is supposedly ramping, and management provides no forward quantification of the Anthropic/Core Scientific deployment timeline, the thesis that "AMD has a real AI GPU story" shifts from proven to speculative. That triggers a TRIM at $520+ and a reassessment at any level below.

Horizon

12–18 months for the full Anthropic/Core Scientific revenue realization. This week for whether the setup survives the first numbers test.

The Bear Case

The biggest risk is not that Q2 is bad — it is that Q2 is fine and the market still dumps it because the forward guidance is cautious. AMD is not managing the deal announcements; the deals manage AMD. If Su or her team uses conservative language on the August 4 call about "execution timelines" and "phased deployments," a market that wants to hear certainty will punish the stock even if the numbers are solid. At a ~150x trailing multiple (per the July memos), the stock has no tolerance for ambiguity, even justified ambiguity. The second risk is NVDA. If Blackwell's yield ramp improves and B200 availability accelerates in Q3, AMD's window to capture share before MI450 arrives narrows. The Anthropic and Core Scientific deals are important precisely because they buy AMD time. Time it doesn't have means the thesis doesn't have conviction.

What I'm Watching

Tuesday, August 4, after-market: AMD Q2 FY2026 earnings and call. I am listening for three things and nothing else: (1) how management frames the first Anthropic GW for H1 2027 — firm commitment or aspirational, (2) whether the Core Scientific deal gets quantified as backlog or remains verbal, and (3) gross margin on data-center products. The first two tell me if the big numbers are real. The third tells me if AMD is making money on what is already shipping. If you only listen for one number, make it gross margin.

Axelrod Research is independent equity analysis, for information only. This is not investment advice.

Sources

• AMD Q2 FY26 Earnings (IR) — scheduled Aug 4, 2026
• AMD-Anthropic Partnership Press Release — Jul 22, 2026
• AMD-Core Scientific Infrastructure Partnership — Jul 28, 2026
• SEC EDGAR — AMD Quarterly Filings — accessed Jul 31, 2026
• Price data: Polygon/Massive daily aggregates, Aug 1, 2026 close