AI Chips Stocks to Own in September 2026: 7 Names with Real Setup
This countdown spans AI accelerators, custom silicon, networking, foundry manufacturing and semiconductor equipment across seven infrastructure stocks.
AI chips remain one of the market's most powerful infrastructure themes, but the investment case has broadened well beyond simply owning a leading GPU designer. Training and inference workloads continue to require enormous amounts of compute, memory, networking and power, keeping semiconductor suppliers central to the data-center buildout. The market is also demanding more proof of durable earnings, because strong AI enthusiasm has pushed many chip and equipment valuations higher. That combination creates an opportunity set with substantial growth, but also a wider range of business models and risk profiles for investors to evaluate.
The structural drivers now reach across the AI stack. Hyperscaler capital spending, the shift from model training toward large-scale inference deployment, and the need for denser systems are supporting accelerators, high-bandwidth memory, advanced packaging, custom silicon and high-speed interconnects. Foundries and semiconductor-equipment companies benefit from the manufacturing intensity behind those systems, while networking specialists address the movement of data between processors. Broadcom's September 2026 update, which raised its AI chip revenue outlook to about $115 billion for fiscal 2027, further underscored the continuing scale of infrastructure demand.
This countdown covers seven US-listed companies with meaningful exposure to AI chips, from equipment and manufacturing through merchant accelerators, custom silicon and networking. The list is presented in countdown order, beginning with #7 and moving to #1. Exposure to the theme comes first, while business fundamentals, growth, profitability, earnings execution and valuation help separate companies operating in closely related parts of the ecosystem.
The screen covers US-listed companies with market capitalizations above $500 million and uses primary-source financial data plus composite metrics for quality, profitability, growth and valuation. Rankings emphasize depth of exposure to the AI-chips theme first, then business fundamentals. Each company is reviewed through its products, relevant AI infrastructure role, margins, growth rates, earnings history and analyst consensus. This is a countdown: the best pick is reserved for #1 at the end, while the monthly refresh reflects changing fundamentals and market expectations rather than short-term price movement.
What they do. The company supplies materials-engineering equipment, services and software to semiconductor manufacturers. Its Semiconductor Systems portfolio covers deposition, etch, rapid thermal processing, metrology, inspection, packaging and ion implantation, while Applied Global Services provides spares, upgrades, services and factory-automation software. That combination gives Applied Materials exposure to both new wafer-fab equipment and the recurring support required to keep installed tools productive.
Why it fits. AI accelerators and data-center processors require advanced manufacturing steps, and Applied Materials supplies equipment across several of those steps rather than selling a single chip product. Its wafer-packaging, deposition, etch, inspection and metrology capabilities also connect directly to the industry's push toward more complex, densely integrated AI systems. This makes AMAT a picks-and-shovels exposure to AI-chip capacity, even though its revenue is tied to the broader semiconductor capital-spending cycle.
Numbers that matter. Applied Materials generated a 49.4% gross margin, a 33.74% operating margin and a 30.05% net margin. Revenue growth was 24.8% year over year, while earnings growth was 42.8%; trailing EPS was $11.59 and next-year estimated EPS was $18.4719. The trailing P/E was 39.39 and the forward P/E was 25.58, reflecting a meaningful valuation premium despite strong profitability and growth.
Recent momentum. Applied Materials has beaten estimates in all eight reported quarters in the supplied history. In the latest reported quarter, dated August 13, 2026, EPS was $3.50 versus an estimate of $3.38, a 3.6% surprise. Analysts recorded four Buy ratings and eight Holds, with a 4.49 consensus score and a $640.8857 average target, suggesting favorable expectations but a more mixed stance than the company's earnings record alone might imply.
What they do. Lam Research designs, manufactures and services semiconductor-processing equipment used in integrated-circuit fabrication. Its products include deposition systems such as ALTUS and Striker, etch platforms including Akara, Flex, Vantex and Kiyo, cleaning systems, copper-interconnect tools and customer-service, spares and upgrade offerings. The breadth of its processing and service portfolio gives Lam multiple revenue streams tied to chip-fabrication complexity and the installed base of semiconductor equipment.
Why it fits. AI chips require advanced logic and memory manufacturing, and Lam's deposition, etch and clean products address the nanoscale process control needed to produce those devices. Its tools are especially relevant to the industry's drive for more precise interconnects, selective etching and higher yields as AI systems become more densely integrated. Lam therefore offers equipment-level exposure to the AI buildout without depending on the success of one accelerator architecture.
Numbers that matter. Lam posted a 50.5% gross margin, a 37.39% operating margin and a 31.27% net margin. Revenue grew 30.0% year over year and earnings grew 34.8%; trailing EPS was $5.76, with next-year estimated EPS of $8.1804. The trailing P/E was 51.77 and the forward P/E was 32.26, making valuation a clear counterweight to the company's high returns and strong operating profitability.
Recent momentum. Lam has beaten estimates in each of the seven reported quarters in the supplied history. Its latest reported quarter, July 29, 2026, produced EPS of $1.82 versus an estimate of $1.69, a 7.7% surprise. The analyst breakdown was four Buys and nine Holds, with a 4.3333 consensus score and a $373.129 average target, indicating solid confidence but limited unanimity.
What they do. Marvell develops data-infrastructure semiconductor solutions spanning the data-center core to the network edge. Its portfolio includes Ethernet controllers, adapters, transceivers and switches; custom application-specific integrated circuits; storage and connectivity products; and interconnect technologies such as digital signal processors, silicon photonics, optical components, PCIe retimers and Compute Express Link switches. The company sells through direct customers and distributors into data centers, communications and other markets.
Why it fits. Marvell is one of the more direct beneficiaries of the AI infrastructure shift toward custom silicon and faster movement of data. Its custom ASICs, Ethernet products, interconnects, optical technologies, PCIe retimers and ultra-accelerator-link switches address the networking and scale-up requirements surrounding AI accelerators. That exposure makes MRVL a complementary AI-chip investment: its opportunity is tied not just to compute, but to the connections that let large clusters operate efficiently.
Numbers that matter. Marvell reported a 52.2% gross margin, a 16.68% operating margin and a 27.93% net margin. Revenue growth reached 36.5% year over year and earnings growth reached 50.0%; trailing EPS was $3.02 and next-year estimated EPS was $3.3913. The trailing P/E was 78.18 and the forward P/E was 56.18, considerably richer than the equipment names in this list and dependent on continued execution against its AI infrastructure opportunity.
Recent momentum. Marvell's supplied earnings history shows six beats in eight quarters. The latest report, August 27, 2026, was a miss: EPS of $0.62 compared with an estimate of $0.65, a negative 4.6% surprise. Analysts still listed nine Buys and five Holds, producing a 4.4865 consensus score and a $284.6409 average target, but the recent miss highlights the execution risk behind its premium valuation.
Get AI research on any stock
Instant reports, daily intelligence, and an AI analyst in your pocket.
What they do. Broadcom supplies semiconductor devices and infrastructure software internationally. Its semiconductor portfolio includes custom silicon, Ethernet switching and routing, network-interface controllers, optical components, PCIe switches and server-storage connectivity, while its software business includes private-cloud, cybersecurity, enterprise and mainframe offerings. This combination gives Broadcom a diversified revenue model, with semiconductor solutions tied to data-center connectivity and infrastructure software adding a separate earnings engine.
Why it fits. Broadcom has unusually deep exposure to the custom-silicon and networking layers of AI infrastructure. Its custom solutions, Ethernet switching and routing, optical connectivity, NIC controllers and PCIe products help connect and scale accelerator clusters, while its private AI software portfolio extends into the surrounding infrastructure stack. The September 2026 outlook increase to about $115 billion of AI chip revenue for fiscal 2027 reinforced the importance of this business to the broader theme.
Numbers that matter. Broadcom's margins were among the strongest in the group: 75.5% gross, 54.31% operating and 42.94% net. Revenue grew 85.5% year over year and earnings grew 215.3%; trailing EPS was $7.86 and next-year estimated EPS was $19.3839. The trailing P/E was 46.05, but the forward P/E was 19.27, suggesting that the valuation depends heavily on the continuation of its rapid earnings expansion.
Recent momentum. Broadcom has beaten estimates in all eight reported quarters in the supplied history. On September 2, 2026, it reported EPS of $3.03 versus an estimate of $2.83, a 7.1% surprise. Analysts listed seven Buys and three Holds, for a 4.7174 consensus score and a $531.8468 average target, while the latest earnings result provided a fresh confirmation of strong operating momentum.
What they do.AMD develops processors, GPUs, AI accelerators, adaptive SoCs, FPGAs, data-processing units and networking products across its Data Center, Client and Gaming, and Embedded segments. Its portfolio includes Instinct accelerators, EPYC server processors, Radeon graphics, Pensando and Alveo products, along with infrastructure for hyperscale providers. That breadth gives AMD several routes into compute and acceleration, while also leaving the company exposed to client, gaming and embedded demand.
Why it fits.AMD is a broad merchant-silicon participant in the AI buildout, with Instinct accelerators aimed at data-center workloads and EPYC processors supporting the host-compute layer. Its AI network-interface cards, DPUs, FPGAs and adaptive SoCs extend the exposure beyond a single GPU product category. That combination gives AMD direct participation in the accelerator market while also positioning it for the surrounding server and networking infrastructure required by hyperscale AI deployments.
Numbers that matter.AMD posted a 55.7% gross margin, a 17.25% operating margin and a 15.58% net margin. Revenue rose 50.1% year over year and earnings increased 159.5%; trailing EPS was $3.92, compared with next-year estimated EPS of $15.5105. The trailing P/E was 131.67 and the forward P/E was 33.22, showing how much of the valuation rests on the expected conversion of rapid growth into future earnings.
Recent momentum.AMD has beaten estimates in five of the seven reported quarters in the supplied history. Its latest reported quarter, August 4, 2026, produced EPS of $1.66 versus an estimate of $1.35, a 23.0% surprise. Analysts recorded four Buys and 13 Holds, with a 4.16 consensus score and a $615.068 average target, reflecting constructive expectations alongside a relatively cautious distribution of ratings.
Pick #2Premium members only
Premium members see this pick's full breakdown — investment thesis, key financial metrics, recent earnings execution, and analyst consensus.
Premium members get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
This monthly screen begins with US-listed companies valued above $500 million and identifies businesses with a meaningful connection to AI-chip infrastructure. The ranking prioritizes depth of thematic exposure: direct accelerators and data-center compute, custom silicon, networking, interconnect, foundry manufacturing and semiconductor equipment receive the closest review. Business fundamentals then refine the order, using composite quality grades, profitability, revenue and earnings growth, valuation multiples, earnings surprises and analyst consensus. Market capitalization and financial figures are refreshed with each edition, while the countdown format runs from #7 to #1. The result is a thematic ranking, not a guarantee of performance or a substitute for individual risk analysis.
▌The Daily Briefing · Free
A new stock idea, every evening.
One stock worth watching each weekday, plus the analysis behind it. Free, in your inbox.
