U.S. semiconductor equities surged on Wednesday, fueled by a bullish outlook from Advanced Micro Devices (AMD), which reinforced investor confidence that artificial intelligence infrastructure spending is set for another growth phase—this time increasingly driven by central processing units (CPUs) alongside graphics chips.

The rally reflected growing market conviction that AI development is moving beyond the early GPU-dominated training era into broader “inference” deployment, where models are actively used in real-world applications.

AMD led the charge, jumping nearly 18% in premarket trading and positioning itself close to a record high if gains hold through regular trading hours. The surge followed the company’s updated long-term expectations for its server CPU market, which now projects annual growth exceeding 35% through 2030—significantly higher than its earlier 18% forecast.

Rival chipmakers followed AMD higher. Intel rose about 6%, while Arm Holdings climbed 11%. Qualcomm gained roughly 4%, reflecting broad-based optimism across semiconductor design firms tied to AI computing demand.

Memory and data infrastructure names also joined the rally. Micron Technology surged 6.4%, while Marvell Technology added 1.7%.

AI Shift Toward “Inference” Expands CPU Opportunity

A key driver behind the rally is the industry-wide shift toward agentic AI systems—applications designed to perform autonomous tasks without constant human prompting. This transition is broadening demand beyond GPUs, which have dominated AI training workloads, into CPUs that handle inference and real-time processing.

AMD emphasized this pivot in its latest outlook, noting that inference workloads are opening new opportunities for its server processor business.

As Matt Britzman, senior equity analyst at Hargreaves Lansdown, explained:

“AMD story is no longer just about having a GPU pipeline to challenge Nvidia... It's increasingly about a broader compute opportunity, with CPUs and GPUs both playing a role as AI workloads become more demanding.”

Strong Performance Fuels Investor Enthusiasm

AMD’s stock momentum has already been strong throughout the year. The company’s shares have risen about 66% year-to-date, significantly outpacing the roughly 5% gain recorded by market leader Nvidia.

The rally has also pushed AMD’s valuation higher, with the stock trading at nearly 40 times forward earnings—well above its five-year average and almost double Nvidia’s valuation multiple—despite Nvidia’s dominant position in the AI GPU market.

Broader optimism in the semiconductor space was further reinforced earlier in the day when Samsung Electronics briefly became only the second Asian company to reach a $1 trillion valuation, underscoring the scale of the global AI-driven chip rally.

Super Micro Surges Despite Regulatory Concerns

AI server manufacturer Super Micro Computer also posted a sharp gain, rising nearly 19% after issuing a stronger-than-expected fourth-quarter revenue and profit forecast.

The company’s upbeat outlook helped ease investor concerns tied to an ongoing U.S. Department of Justice case involving alleged illegal chip shipments to China. Despite those legal uncertainties, demand for its AI-optimized servers continues to accelerate.

Chief Executive Charles Liang said demand remains robust across data center and cloud software segments, with production expanding across Taiwan, Malaysia, and the Netherlands.

Analysts at J.P. Morgan said the improved outlook reflects stronger margins and a more diversified customer base, though they cautioned that governance-related concerns remain a long-term overhang.

Market Narrative: AI Infrastructure Still in Expansion Phase

Overall, Wednesday’s rally underscored a central theme in global tech markets: AI infrastructure spending is still early in its lifecycle, and the demand curve is broadening rather than slowing.

With CPUs, GPUs, memory chips, and AI server makers all participating in the surge, investors appear to be positioning for a multi-layered expansion of computing demand rather than a single-product cycle centered on GPUs alone.