- AI spending is moving into another phase as hyperscalers expand data-center capacity, deploy larger AI clusters, and prepare for increasingly demanding workloads. These three semiconductor stocks sit at different points in that infrastructure buildout.
The artificial intelligence trade is no longer limited to companies developing chatbots and foundation models. The next phase of spending is increasingly tied to the physical infrastructure required to train and run those systems.
That creates several potential beneficiaries across the semiconductor supply chain. NVIDIA Corp. (NASDAQ: NVDA) supplies accelerated computing, Broadcom Inc. (NASDAQ: AVGO) provides custom AI accelerators and networking technology, while Taiwan Semiconductor Manufacturing Co. (NYSE: TSM) manufactures many of the advanced chips designed by leading AI companies.
The opportunity is being supported by the sheer scale of planned infrastructure investment. TrendForce estimates that the combined 2026 capital expenditure of the world's nine largest cloud service providers could reach roughly $830 billion, up 79% from 2025. The estimate includes Amazon, Alphabet, Meta Platforms, Microsoft and Oracle.
For investors looking beyond the most obvious AI names, the important question is which companies can continue supplying the hardware required as that spending cycle expands.
1. NVIDIA (NASDAQ: NVDA): The Core AI Compute Supplier
NVIDIA remains one of the most direct ways to gain exposure to continued spending on AI computing infrastructure. The company reported $96.2 billion in fiscal second-quarter 2027 revenue, up 106% from a year earlier. Data Center revenue reached $89 billion, an increase of 117%, as demand for its Blackwell Ultra infrastructure accelerated. NVIDIA's latest financial results also showed strong demand from hyperscalers, enterprises, AI-native companies and sovereign customers.
NVIDIA's position is expanding beyond individual graphics processors. Its platforms increasingly combine GPUs, CPUs, networking, storage and software, giving customers integrated systems for large AI workloads.
The company also announced that AWS plans to deploy 2 million additional NVIDIA GPUs across its infrastructure, highlighting the scale of future compute requirements. The AWS-NVIDIA expansion is one example of how hyperscaler spending can translate into demand for AI hardware.
NVIDIA's challenge is equally clear: expectations are extremely high. Its future growth depends on continued capital spending by cloud providers and AI developers, while competition, supply constraints, export restrictions, and the economics of AI workloads remain important risks.
2. Broadcom (NASDAQ: AVGO): The Custom AI and Networking Play
Broadcom offers a different exposure to the AI infrastructure cycle because its opportunity extends beyond general-purpose accelerators. The company reported $16.7 billion in AI semiconductor revenue during its fiscal third quarter, up 221% from the previous year and 54% from the prior quarter.
Broadcom said demand for its custom AI accelerators and networking products remained strong and expects AI semiconductor revenue to reach $21.7 billion in the fourth quarter. That makes Broadcom's latest earnings report particularly relevant to the next stage of AI spending.
Custom accelerators are becoming increasingly important as major cloud companies design chips tailored to their own workloads. At the same time, larger AI clusters require increasingly sophisticated networking to move data between processors. Broadcom is therefore exposed to two major infrastructure requirements: custom compute and high-speed connectivity.
The company is also participating in larger AI infrastructure financing arrangements. In June, Broadcom announced an AI infrastructure platform with Apollo and Blackstone designed to support more than 20 gigawatts of global AI deployments using Broadcom XPUs and networking solutions.
That does not guarantee future revenue or stock performance, but it illustrates how the AI spending cycle is spreading beyond GPU purchases into the broader architecture surrounding large-scale computing.
3. Taiwan Semiconductor Manufacturing (NYSE: TSM): The Manufacturing Bottleneck
Taiwan Semiconductor Manufacturing Co. represents another part of the AI supply chain: advanced semiconductor manufacturing. TSMC produces chips for many of the world's leading semiconductor designers and has been increasing capacity to accommodate growing demand for advanced computing.
The company's August revenue provides a current indication of that demand. TSMC reported NT$514.8 billion in monthly revenue for August, up 53.3% from the same month a year earlier. Through August, 2026 revenue had increased 39.3% year over year.
TSMC management has also repeatedly highlighted the strength of AI-related demand. During its second-quarter earnings call, the company said AI demand remained "extremely robust" and raised its full-year 2026 revenue-growth outlook to slightly above 40% in US dollar terms.
The company is simultaneously ramping its 2-nanometer technology and expanding capacity. Its latest investor materials show how advanced manufacturing capacity is becoming a critical component of the AI infrastructure expansion.
The main risk is that semiconductor manufacturing requires enormous capital investment and long lead times. TSMC also faces geopolitical exposure because most of its production remains concentrated in Taiwan.
The Next AI Spending Cycle Is Broader Than GPUs
The AI infrastructure market is increasingly becoming a multi-layer supply chain rather than a single-chip story. NVIDIA (NASDAQ: NVDA) provides accelerated computing, Broadcom (NASDAQ: AVGO) is benefiting from custom silicon and networking, while TSMC (NYSE: TSM) provides the advanced manufacturing capacity needed to produce many of those chips.
There are also risks to the broader spending cycle. Morgan Stanley has noted that power availability, data-center permitting and labor shortages could constrain AI infrastructure expansion even if demand for computing remains strong. Meanwhile, rising capital requirements mean investors are increasingly focused on whether AI infrastructure spending can produce adequate returns.
For these three semiconductor companies, the next stage of the AI cycle will therefore depend not only on demand for artificial intelligence but also on how quickly data centers are built, how much computing capacity customers require, and whether the economics of AI continue to justify the enormous infrastructure investment.