Between a Sector Rout and a 6-Gigawatt Deal: AMD’s AI Narrative Hangs in the Balance

The semiconductor sector is caught in a peculiar storm: stellar earnings are being punished instead of rewarded, and the market’s mood has turned defensive almost overnight. AMD shares have not been spared, sliding 2.17 percent on Wednesday to €439.95 and shedding 7.45 percent over the past week. The stock now sits roughly 14 percent below its 52-week high, as investors rotate capital into safer corners of the market amid fresh geopolitical jitters and rising oil prices.

The trigger for the broader sell-off was Samsung’s eye-popping 1,800 percent profit surge from Seoul — a result that, paradoxically, sent its stock down nearly seven percent on Tuesday. The pattern of “selling the news” has spread across the chip landscape, hitting AMD even as the company makes aggressive operational moves. The annualised volatility of AMD shares now stands at around 77 percent, a figure that underscores just how skittish the market has become.

A Landmark Meta Partnership and the Push for Local AI

Yet beneath the surface-level turmoil, AMD is quietly building the infrastructure for the next phase of artificial intelligence. The company has finalised a sweeping deal with Meta to supply up to six gigawatts of AI computing power. The first systems from the MI450 family are scheduled to go live in the second half of 2026, cementing AMD’s status as the primary alternative to Nvidia in the data centre.

At the same time, AMD is broadening its addressable market beyond the hyperscaler crowd. The launch of the new Ryzen AI Halo processors this week brings powerful on-device AI capabilities to local machines, reducing reliance on the deep-pocketed budgets of cloud giants. It is a strategic hedge that dovetails neatly with the industry’s growing interest in “agentic” AI — autonomous agents that can execute multi-step tasks rather than simply processing text.

Goldman Sachs sees this trend as a key reason to turn bullish on AMD, raising its price target while maintaining a Buy rating. The bank argues that demand for both high-performance CPUs and GPUs specialised for agent-based workloads is accelerating. Two data points support that thesis: the Japanese startup Turing has shifted 10 percent of its AI training from Nvidia to AMD chips, and Meta is reportedly lining up as an early customer for the forthcoming EPYC Venice processors. When the world’s biggest tech spenders begin diversifying their hardware supply, AMD’s architecture gains credibility that extends well beyond a single quarter.

Should investors sell immediately? Or is it worth buying AMD?

The Bull and Bear Cases Collide Ahead of July’s Showcase

All eyes are now on the “Advancing AI” event scheduled for late July, a gathering that will serve as the clearest test yet of AMD’s ability to convert narrative into revenue. Analysts at Citi and Goldman Sachs expect the company to unveil new hyperscaler partnerships — precisely the sort of validation needed to justify the current valuation. Without fresh customer names, the stock could struggle to gain traction, especially with a trailing price-to-earnings ratio that has stretched as high as 207.

On the bullish side, AMD’s data-centre revenue grew more than 50 percent year over year in the first quarter of 2026, and management has guided for approximately 70 percent growth in the second quarter. The MI300X accelerator continues to win acceptance among cloud operators, and Nvidia’s next-generation “Kyber” system has reportedly slipped to 2028, giving AMD an extended window to capture share. The consensus price target among analysts is a modest €445.53, though a separate survey puts the average at €444.34 — a narrow range that suggests the market is waiting for a decisive catalyst.

The bear case, however, focuses on the fragility of that growth. Some critics warn that the semiconductor cycle may be nearing a local peak, and that the “sell the news” reaction to Samsung’s results signals deepening ROI concerns around massive AI infrastructure spending. More troubling still is the threat of fragmentation: the AI startup DeepSeek is reportedly developing its own inference chips, which could erode AMD’s addressable market even if overall demand remains robust. And Meta itself is said to be trying to offload unused computing capacity, a move that would temper the runaway growth in AMD’s data-centre segment.

Technical Signals and the Short Side

The chart offers little comfort for the near term. The relative strength index sits at a neutral 48.8, and the share price is barely holding above its 50-day moving average. Prominent investors, including Michael Burry, have already taken short positions against the chip sector, betting that the euphoria around AI hardware has peaked.

The “Advancing AI” event at the end of July will be the first major inflection point. If AMD can demonstrate that its agentic-AI thesis is translating into concrete hyperscaler contracts, the stock could reclaim its upward trajectory. If not, the next hard test arrives in early August with the official quarterly report — a moment that will reveal whether the promise of autonomous agents is actually generating the margin-rich revenue the market has already priced in.