The summer months tend to bring volatility, thin trading, and a rotation away from momentum stocks. But some AI names have more than momentum behind them.
The more durable AI trade sits elsewhere: businesses so structurally embedded in the AI supply chain that market slowdowns don't dislodge their competitive positions. They operate with moat-like advantages—monopoly-scale manufacturing, near-irreplaceable chip architecture, and analog signal dominance built over decades—and their recent earnings suggest the AI demand cycle hasn't taken a vacation either.
Advanced Micro Devices Positioned for Infrastructure Inflection
For most of the AI buildout cycle, Advanced Micro Devices (NASDAQ: AMD) played second fiddle to NVIDIA (NASDAQ: NVDA) in the GPU narrative. Prior to the company’s Q1 2026 earnings report, investors were pricing AMD as a PC-era chipmaker with an AI story.
The company’s Q1 2026 earnings report confirmed that it’s time to consider giving it first chair status. That means reclassifying it as an AI infrastructure company with a PC business attached.
There were several highlights in the report. Total revenue reached nearly $10.3 billion, up 38% year-over-year and well above consensus estimates of $9.9 billion. Q2 guidance of approximately $11.2 billion represents a roughly 46% year-over-year acceleration.
Where revenue growth came from is also part of the story. The data center segment generated $5.8 billion, a 57% year-over-year increase that was driven by EPYC server CPUs and Instinct GPU shipments. That was the first time EPYC and Instinct were the primary revenue and earnings drivers.
CEO Lisa Su described agentic AI as "a clear inflection in our growth trajectory.” The rise of AI agents and inference workloads is driving unexpected demand for server CPUs, not just GPUs. In that environment, AMD's EPYC architecture is the principal beneficiary.
The company forecasts the server CPU market to grow at 35% annually through 2030, reaching more than $120 billion, up from its prior 18% estimate. The company has a multi-year partnership with Meta Platforms (NASDAQ: META) to deploy up to 6 gigawatts of GPU capacity using custom MI450-based units.
That said, AMD's moat isn't at NVIDIA’s scale yet. But it's built on the only x86 architecture competing at the data center AI workload level, with an integrated CPU-GPU roadmap that hyperscalers need for supply chain diversification.
Taiwan Semiconductor Manufacturing Is the Monopoly Everyone Depends On
There is no company more central to the AI supply chain than Taiwan Semiconductor Manufacturing Company (NYSE: TSM).
The company has a moat in the semiconductor industry, one that is close to its position in advanced-node manufacturing.
Many of the most important AI accelerators and advanced processors from NVIDIA GPU, AMD, and Apple (NASDAQ: AAPL) rely on TSMC’s leading-edge manufacturing.
The company is not a customer-facing brand. It's the foundational layer beneath every name that is.
The company’s Q1 2026 results confirmed the demand picture.
Furthermore, management guided Q2 revenue of $39 to $40.2 billion, implying another 32% YOY gain. It also raised its full-year 2026 revenue growth forecast to above 30% in U.S. dollar terms. Management is targeting $52 to $56 billion in capital expenditure to expand capacity. That figure would be implausible without firm customer commitments underpinning it.
TSMC’s dominance is not hyperbole. Its lead in 3nm and the upcoming 2nm node is measured in years, not quarters. No hyperscaler can build the AI infrastructure it's committed to without TSMC fabs. That structural dependency, combined with AI chip revenue projected to grow more than 50% annually through 2029, makes TSMC a clear, long-term compounder in the sector, and the one least likely to take a summer vacation from growth.
Analog Devices Is the Quiet Quasi-Monopoly
Analog Devices Inc. (NASDAQ: ADI) doesn't make GPUs or the chips that run large language models. What it makes is the analog and mixed-signal infrastructure that sits between the physical world and digital systems. This includes the sensors, power management, data converters, and signal chain components embedded in data centers, industrial robots, defense systems, autonomous vehicles, and medical equipment.
The company’s products have average useful lives of 15 to 20 years in its most profitable segments. And customers don't switch lightly. That's a moat built on a specific, durable engineering need.
The AI connection at Analog Devices is indirect, but essential. Every GPU rack needs power management. Every AI inference server needs thermal regulation and signal conditioning. Every robotic system in an AI-enabled factory runs Analog Devices signal chain and power technology.
The company's agreement to acquire Empower Semiconductor, which targets advanced power delivery for AI accelerators, is explicitly intended to deepen that position. Management stated that flexible manufacturing investments now give it the capacity to service up to $20 billion in annual revenue, more than double the current run rate.
If the AI trade turns volatile this summer, ADI's 15-to-20-year product cycles and diversified end-market exposure make it far more resilient than a pure-play GPU name. The moat here was built long before AI became a market theme.