NVIDIA

By Jason Kirsch , Contributor. Edited by Sophia Acevedo , Associate Editor. The semiconductor sector has performed well this year overall. AI infrastructure spending has been a significant tailwind for growth this year. However, not every chip stock deserves a spot in your portfolio, so it's important to favor companies with strong fundamentals and durable AI-driven growth. Top recommendations include NVIDIA, the undisputed AI chip leader; TSMC, the global foundry backbone; Micron, capitalizing on insatiable memory demand; Broadcom, dominant in custom AI accelerators; and Lam Research, a key supplier of wafer fabrication equipment.

Despite recent stock surges, Intel may not be the as compelling of a choice. Investors should manage volatility through dollar-cost averaging and diversified portfolios across the AI supply chain, while closely monitoring valuations. The semiconductor sector is having a moment, even despite recent volatility . The benchmark Invesco PHLX Semiconductor ETF (SOXQ) is up roughly 100% year to date in 2026, easily lapping the broader tech sector. Behind the surge is a relentless wave of AI infrastructure spending from hyperscalers like Microsoft, Google and Meta, which are collectively expected to deploy nearly $700 billion in capital expenditures in 2026, with the majority earmarked for AI data centers.

That's a massive tailwind for chipmakers across the supply chain. But not every chip stock deserves a spot in your portfolio. With valuations stretched across much of the sector, selectivity matters more than ever. I’ve identified five semiconductor stocks with compelling fundamentals and durable AI-driven growth catalysts — and one that investors may want to sidestep despite the hype surrounding it. Picking semiconductor stocks requires more than riding momentum. For the buy list, the focus was on companies with structural advantages: dominant market positions, improving earnings trajectories and direct exposure to the parts of the AI supply chain where demand is structurally supply-constrained rather than cyclically elevated.

Revenue acceleration, analyst conviction and forward earnings estimates were all weighted heavily. AI spending is expected to stay strong in 2026, justifying premium valuations for the highest-quality companies. The stock to ignore was identified by the inverse criteria: where enthusiasm has outrun execution, where near-term profitability remains elusive despite a surging share price, and where the competitive position is more contested than the market currently prices in. Elevated valuations remain a common theme across much of the semiconductor industry, and distinguishing between AI beneficiaries and AI hype is the core task here.

NVIDIA remains the undisputed engine of the AI chip revolution. 9 billion — a 65% increase versus the prior year — with earnings growing at nearly the same pace. The company’s GPU architecture has become the default infrastructure for training and running AI models, cementing a competitive moat that rivals have struggled to breach. Despite the monster run, 48 analysts maintain a "Strong Buy" rating with a consensus price target implying roughly 46% upside from current levels. Nvidia estimates it can sell a trillion dollars ' worth of data center AI chips in 2026 and 2027 combined — a number that, remarkably, looks achievable given current demand signals.

The company has also identified a $200 billion AI CPU inference opportunity , expanding its total addressable market well beyond GPUs. With wafer-fabrication equipment spending accelerating and hyperscalers placing multi-year orders, Nvidia's revenue visibility is as strong as any company in tech. For investors with a multi-year horizon, NVDA remains the single clearest expression of the AI hardware megatrend. TSMC controls 72% of the global chip foundry market , making it the physical backbone of the AI supply chain. Every major AI chip — from NVIDIA’s GPUs to Broadcom’s custom ASICs — runs through TSMC’s fabs.

Its dominance in advanced nodes including 3nm volume production and 2nm ramp-up gives it a near-irreplaceable strategic position. In 2026, TSMC has also raised its full-year revenue growth guidance. TSMC’s competitive moat is structural, not cyclical. Building a world-class semiconductor foundry takes years and tens of billions in capital investment — a barrier that makes TSMC’s position self-reinforcing. The adoption of agentic AI is expected to drive another wave of demand, as AI agents processing tasks without human interaction require significantly more token generation and compute capacity.

Investors should monitor geopolitical risk given TSMC's Taiwan base, but for long-term holders, the company's combination of pricing power, technology leadership and AI-driven growth makes it one of the sector's most compelling holds. Micron is the memory play on AI infrastructure, and its numbers have turned extraordinary. Revenue spiked 196% , and non-GAAP earnings surged 682% year over year in its fiscal Q2 2026. The reason is structural: AI context lengths are increasing 30x per year while memory content per server has doubled in three years, creating an insatiable demand for DRAM and high-bandwidth memory (HBM) that Micron is uniquely positioned to supply.

The company has already sold out its HBM supply for 2026, with significant 2027 capacity committed under long-term agreements. Morgan Stanley analysts had named Micron their top semiconductor pick for early 2026, arguing the ongoing AI buildout has created the most severe DRAM and NAND shortage in three decades. That shortage is structurally driving prices higher. With most major analysts in "buy" mode and memory demand showing no sign of softening, MU offers a compelling combination of growth and value within the chip sector. Broadcom has emerged as the dominant supplier of custom AI accelerators (ASICs), holding 70–80% market share in a segment that counts Alphabet’s Google, Meta, ByteDance, OpenAI and Anthropic among its customers.

Unlike NVIDIA’s general-purpose GPUs, Broadcom's ASICs are purpose-built for each hyperscaler's specific AI workloads, offering performance and efficiency advantages at scale. 8 billion in the quarter. The AI accelerator market is projected to expand at 29% annually through 2033 , and Broadcom’s customer pipeline continues to grow. Beyond chips, Broadcom’s VMware acquisition adds a recurring software revenue stream in data center virtualization. That dual engine of custom AI chips and enterprise software makes AVGO one of the more diversified and durable names in the semiconductor space.

Lam Research makes the machines that make the chips — specifically, deposition and etch equipment used by foundries and memory manufacturers. It’s a pick-and-shovel play on the entire AI semiconductor build-out. Lam raised its 2026 wafer fabrication equipment (WFE) spending forecast to $140 billion , and third-party estimates from Counterpoint Research put the number even higher at $159 billion. Whether chips are made for AI training, inference or memory, Lam's equipment is almost always in the manufacturing chain. The beauty of Lam’s position is that it benefits regardless of which AI chip design wins.

Consensus estimates project a 26% revenue increase in FY2026 followed by a 31% jump in FY2027, and Bank of America projects the global semiconductor market could clock a 28% CAGR through 2030 , topping $2 trillion in annual revenue. That sustained growth cycle translates directly into sustained WFE demand. For investors who want AI chip exposure with less concentration risk in any single chip designer, LRCX offers an elegant upstream alternative. Intel’s stock has surged on AI optimism, powered by government investment, new foundry contracts and a narrative of turnaround.

But the underlying financials tell a more complicated story. 4 billion in sales. The foundry business remains deeply unprofitable, and while deals with Apple, Microsoft and SpaceX are promising, they have not yet yielded meaningful financial gains. Morningstar does not believe Intel has an economic moat. Morningstar has argued that its manufacturing setbacks have eliminated competitive advantages that chip design expertise alone cannot restore. Intel’s returns on invested capital have fallen in recent years, and Morningstar foresees no excess returns on capital in the years ahead.

Until the foundry business turns profitable and Intel demonstrates sustained execution, the risk/reward doesn't justify the current valuation. Even the best semiconductor stocks can be volatile, and 2026 has offered a clear reminder of that. For example, in early June, AMD and Intel shares dropped more than 10% in a single session following cautious AI chip guidance from Broadcom — despite the broader AI thesis remaining intact. For investors, the practical lesson is that single-session drawdowns in high-growth sectors are common.