AMD reported a roughly 57% year‑over‑year rise in data‑centre revenue and its shares jumped about 15% after results, a sharp number that traders and analysts say signals a broader CPU comeback. The shift reflects AI workloads moving from pure model math toward agent-style tasks that need orchestration, data movement and the kind of general‑purpose server processors made by AMD and Intel, Yahoo Finance reported. That has pushed data‑centre operators to buy more CPUs and system components alongside GPUs, rebalancing server designs and procurement. Nvidia’s results on May 20 are the next market milestone analysts say will test whether this rack-level investment cycle holds up.

AMD and Intel are no longer just incidental beneficiaries of the AI boom. After AMD’s latest quarter, market participants are describing what they call a "CPU comeback," as demand for general‑purpose server CPUs rebounds while AI projects move beyond pure training math to more complex, agent-style workloads, according to Yahoo Finance.

AMD’s quarterly report stood out. Yahoo Finance noted that the company beat Wall Street expectations, posted a roughly 57% increase in data‑centre revenue year over year, and saw its stock rise about 15% on the news. Analysts tied the gain directly to demand for both AI accelerators and server processors. Intel, meanwhile, reported a better‑than‑expected outlook and improving data‑centre demand that also produced a post‑earnings move in its stock, underscoring that the AI trade is broadening beyond a GPU‑only story, Yahoo Finance said.

Why CPUs are back

The technical reason is simple. As AI projects evolve from one‑off model training to applications that act on data, coordinate services and manage workflows, architecture matters at the rack and system level, not only at the chip level. That means CPUs, memory, storage and networking become part of the investment thesis. Data‑centre operators are buying more chips that manage workflows rather than only GPUs that perform heavy matrix computation, Yahoo Finance reported.

Market watchers highlighted the change. Wedbush was quoted as saying, "CPUs stole the headlines," while Bernstein upgraded AMD to Outperform and raised its price target because the company has exposure to both CPUs and GPUs, Yahoo Finance reported. The framing shifts the conversation from a single‑chip GPU narrative to a rack‑level story in which servers, interconnects and system components carry value.

That rebalancing matters for procurement and for companies that make the components surrounding processors. If operators plan larger racks filled with balanced CPU and GPU capacity, they will order more memory, storage and networking gear that keeps models operational. The effect stretches beyond Nvidia and accelerator makers to traditional server CPU vendors and their supply chains, Yahoo Finance said.

Market ripples and a note of caution

The market response has been dramatic at times. One tally attributed to GuruFocus counted roughly US$200 billion of market value added across chipmakers in a single session tied to AI headlines and OpenAI‑related flow. Yahoo Finance reported a separate calculation that put the post‑March rebound among the 10 biggest chip‑related winners at about US$3 trillion in added market value, with companies such as Nvidia, Broadcom, TSMC, Micron, Intel and AMD near the top.

Mastheads

Exchange‑traded funds that track large‑cap semiconductors have likewise outpaced broader tech this year. One report found the iShares Semiconductor ETF, SOXX, up about 31% year to date and VanEck’s SMH up roughly 39% year to date. Those trackers also showed 14‑day relative strength indices in the high 70s to 80, a technical band that some market technicians describe as overbought.

FinancialContent flagged that historical comparisons show ETFs breaching extreme RSI thresholds have sometimes preceded negative forward returns. That note sits alongside more optimistic takes. One report cited analysts at Citi who still see room for margin, revenue and inventory improvement across chipmakers because of AI demand. That same account estimated AI chip sales represent about a quarter of semiconductor sales, leaving space for further upside if the cycle deepens.

Not all new investment themes are evenly supported by evidence. Commentators and some institutional desks are pointing to analog and industrial‑focused semiconductor names as potential rotation beneficiaries.

The pitch is that analog chips handle real‑world signal conversion and power management for autos, industrial IoT and sensor‑heavy systems, which could deliver steady demand even if the AI narrative cools. That view was developed in a single‑source piece from Strykr.ai and isn't corroborated elsewhere in the set of coverage.

The bundle of reporting also shows uneven sourcing. Several headline figures are single‑sourced in the coverage. The roughly 57% year‑on‑year jump in AMD’s data‑centre revenue and the US$3 trillion rebound figure appear only in the Yahoo Finance earnings recap. The one‑day, US$200 billion session gain and an OpenAI employee‑share sale valuation noted in a related account are reported only in GuruFocus. Readers and investors should note which claims rest on multiple independent reports and which derive from a single newsletter or market note.

Still, the broad pattern is clear across multiple accounts. AI is pushing some centres of hardware demand beyond accelerators to include the processors and systems that put models into production. That shift is already affecting buying decisions and is drawing fresh attention to names that had not been the focus of the early GPU mania, Yahoo Finance said.

For investors, that presents two competing frames. One is a fundamentals‑driven argument that AI creates new, durable demand across memory, processors and systems, and that some semiconductor makers have room to improve margins and inventory turns. The other is a momentum story that has pushed prices to extreme technical readings and that has historically been followed by pullbacks at times.

Which frame proves dominant will depend on forthcoming company results and on whether capital spending at data centres becomes more balanced at the rack level than it was during the first GPU gold rush, analysts and market trackers said.

Related Articles

Nvidia's May 20 earnings are the next clear test: analysts expect the report to show whether the GPU leader's results confirm a broader rack-level AI investment cycle that would underpin a rotation into CPUs and analog chip names.

This article was created with AI assistance.