Analyst Price Targets (as of August 4, 2026)
- Benchmark – Buy – $685
- Baird – Outperform – $1,250
- UBS – Buy – $730
- BofA – Buy – $620
- Jefferies – Buy – $640
- Wedbush – Outperform – $600
- Wells Fargo – Overweight – $615
- RBC – Sector Perform – $540
- 24/7 Wall St. Price Target – $563.98
AMD is trading up roughly 4% this morning, somewhere around $508, and the move has nothing to do with the numbers coming after the close. The market already knows what those numbers look like. Management told investors on May 5: Q2 revenue of $11.2 billion, plus or minus $300 million, non-GAAP gross margin around 56%. The Street layered consensus around $11.32 billion and $1.61 in adjusted EPS.
So what’s the market actually buying this morning?
The answer is the Helios story, and whether tonight’s earnings call gives it teeth.
The Gigawatt Commitments Are Real. The Revenue Is Not Yet.
Here is the thing that gets glossed over in most AMD previews. The company has assembled 14 gigawatts of announced AI compute commitments across three of the most capital-intensive AI labs on the planet. OpenAI and Meta each committed to 6 gigawatts of AMD Instinct GPUs across multiple chip generations, with initial 1-gigawatt tranches scheduled to begin in the second half of 2026. Anthropic added another 2 gigawatts in a deal announced July 22, with the first gigawatt beginning deployment in the first half of 2027, and AMD committing up to $5 billion in a strategic equity investment in the Claude developer.
That is a staggering number on paper.
None of it is Q2 revenue.
The Helios rack-scale system, which bundles 72 MI455X GPUs with EPYC Venice CPUs, Pensando networking, and 31 terabytes of shared HBM memory per rack, is in full production and is expected to start shipping in the coming months, according to CEO Lisa Su at the Advancing AI 2026 event on July 23. AMD executives said they expect Helios deployments at massive scale toward the end of 2026, accelerating throughout 2027. The Anthropic commitment doesn’t begin delivering at scale until 2027. The OpenAI and Meta first gigawatt tranches are also second-half 2026 events. Analysts expect that early rack-scale systems may not be fully recognized as revenue until installation and customer acceptance, which could push some upside into later quarters.
So you have a situation where the deals are signed, the hardware is in production, and the stock has rallied more than 100% year to date, but the revenue line for the quarter being reported tonight reflects almost none of it. What matters tonight is not what happened in Q2. It’s what management says Q3 and Q4 look like as Helios starts shipping at volume.
The Numbers AMD Has to Deliver
Q2 actuals to watch:
- Revenue consensus: ~$11.32 billion (company guidance midpoint: $11.2B), up ~47% year over year
- Adjusted EPS consensus: $1.61 per share (vs. $0.48 in Q2 2025, representing about 235% growth year over year)
- Data Center segment: analysts expect roughly $6.5 billion, including approximately $4 billion from server CPUs and $2.5 billion from AI accelerators
- Non-GAAP gross margin: guided at 56%
For context, Q1 2026 was the inflection quarter. Revenue hit $10.25 billion, up about 36% year over year, with non-GAAP EPS of $1.37. Data Center posted $5.8 billion, up 57% from a year ago, and AMD reached a record ~46% share of x86 server CPU revenue based on Mercury Research data reported by multiple outlets. That was the quarter that sent the stock up nearly 19% in a single session.
To match that reaction tonight, AMD probably needs Data Center to clear the $6.5 billion line cleanly, and Q3 guidance above $12 billion would signal the Helios ramp is on schedule rather than slipping.
The Gross Margin Question Is the Real Test
Slight tangent, but it matters. AMD management has signaled that a rack-scale ramp can create near-term margin pressure. Early production runs on integrated rack-scale systems tend to be margin-dilutive as yields improve and the cost stack gets optimized.
CEO Lisa Su also highlighted at recent AMD commentary that inferencing and agentic AI workloads are increasing CPU compute needs, which can shift the CPU-to-GPU mix in certain deployments. More EPYC content per Helios rack is a margin offset, and it’s exactly the kind of mix shift that can make the gross margin line look cleaner than initial estimates suggested.
Watch the 56% non-GAAP gross margin target closely. A miss there sends a different signal than a revenue beat.
The Competitive Position Nobody Is Pricing Correctly
Nvidia still controls the vast majority of the data center GPU market. AMD’s share is much smaller by most third-party estimates. That gap is not closing overnight.
But the structure of the recent AMD deals is worth thinking about differently than most commentary does. OpenAI received a warrant for up to 160 million shares vesting against deployment milestones tied to purchases. Meta received a performance-based warrant for up to 160 million shares tied to shipment milestones and additional performance terms. The Anthropic deal uses direct equity investment rather than warrants, a structurally different arrangement that signals confidence in the relationship rather than just the contract size.
This is not how companies sell chips. This is how companies buy durable market position against a dominant incumbent. AMD is essentially using its balance sheet to lock in multi-year AI lab demand at a scale that creates switching costs in both directions.
The ROCm software collaboration embedded in the Anthropic deal may be the most strategically important component. Anthropic will work with AMD on software and workload optimization on AMD’s ROCm stack, which has historically been the weakest link in AMD’s AI pitch relative to Nvidia’s CUDA ecosystem. If the engineering collaboration actually accelerates ROCm parity, the hardware story becomes significantly more defensible.
Forward Scenarios
Bull case: Data Center revenue beats $6.5 billion, Q3 guidance lands above $12.5 billion, and management provides specific language about Helios delivery windows tightening. Gross margin holds at or above 56%. The stock reclaims the $550+ range and analysts who have been sitting on the sidelines upgrade. Analyst consensus sits around the mid-$500s depending on the dataset, with the Benchmark target at $685 and the most aggressive call at $1,250 from Baird.
Base case: Revenue beats modestly, Data Center comes in around $6.3–6.5 billion, Q3 guide is in line with the $12.5 billion consensus, and margin commentary is stable but not exciting. Stock reacts mildly, bounces between $490 and $530, and waits for the first Helios delivery confirmations in the second half of 2026.
Bear case: Data Center misses $6.5 billion, Q3 guidance disappoints, or management softens language around the MI450-series ramp timing. Any hint of export-control headwinds into China, which has been a recurring overhang for the sector, could also pressure the stock. Support near the $460–470 range comes into play, where the stock was trading before the Advancing AI 2026 event.
Technical Context
AMD pulled back about 12% from its June high before finding footing near its 200-period moving average around $480. Shares have recovered toward the $505–510 range this morning. The 52-week range runs from $149.22 at the low to $584.73 at the high. The stock is essentially trading between its mid-year recovery level and the next meaningful resistance cluster near $530–540, which corresponds to where many analysts’ average price targets cluster. A strong Q3 guide tonight could pull the stock toward that zone quickly. A disappointment tests the $460 area.
What Investors Should Watch Tonight
- Q3 revenue guidance: Consensus is around $12.5 billion. Anything at or above that is constructive; softer language about Helios shipment timing would be a red flag.
- Data Center mix: How much of the $6B+ comes from GPU accelerators versus EPYC CPUs tells you how much of the contracted demand is already flowing through.
- Gross margin trajectory: Whether management frames a clear quarter-by-quarter margin step-up path as Helios scales matters as much as the Q2 number itself.
- Helios delivery language: Specific customer names and delivery windows, especially around large hyperscaler ramps and announced large-scale deployments, would be more valuable than any single revenue figure.
- Export control commentary: Any updated guidance on what AMD can and cannot ship to restricted markets affects the long-term demand picture.
Bottom Line
The pop this morning is not about what AMD earned in Q2. The market already knows the Q2 framework. What it’s actually pricing is the credibility of a second-half ramp that, on paper, involves 14 gigawatts of committed demand from three of the most important AI labs on earth, a rack-scale system moving from full production into initial shipping, and a software collaboration designed to close the one gap that has kept enterprise buyers anchored to Nvidia.
The question tonight is not whether AMD beats. The question is whether management can show, in specific and credible terms, that the gigawatt-scale deals signed over the past nine months are translating into recognized revenue on a schedule the market can believe in. A beat without forward clarity is worth less than guidance that genuinely raises the second-half ceiling.
The contracts are signed. The hardware is in production. Tonight AMD has to show the timeline is real.
For informational purposes only.