The stock market has cooled off just a bit in the past few weeks after its torrid pace over the past two years, but it is still solidly in bull market territory. The S&P 500 is off its all-time high set in mid-February by about 6.4%. Meanwhile, artificial intelligence (AI) continues to be a driving force behind this market's movements, with recent advances in this technology looking like they could have game-changing applications for the companies implementing them.

If this bull market is to continue, AI will likely play a role. Let's look at three AI infrastructure stocks well worth considering in this bull market.

1. Nvidia

As long as spending on AI infrastructure continues unabated, Nvidia (NVDA -5.74%) remains the company best positioned to benefit. Its graphics processing units (GPUs) have become the backbone of AI infrastructure due to their strong processing power, which makes them great chips for training AI models and inference.

Meanwhile, the company's CUDA software platform, which was long ago created to allow developers to program its chips for tasks outside of their original purpose of speeding up graphics rendering, has created a wide moat for the company. Since then, Nvidia has only widened its moat through CUDA X, a set of libraries and microservices designed for AI. This has allowed Nvidia to take a dominant 90% market share in the GPU market.

Meanwhile, AI infrastructure spending only continues to ramp up. The big three cloud computing companies have budgeted over $250 billion in capital expenditures (capex) this year, with the vast majority of that going directly toward AI infrastructure. Meta Platforms plans to spend up to $65 billion on AI infrastructure, while a consortium of companies led by OpenAI and Softbank has pledged to spend $500 billion on building out AI data centers in the U.S. as part of Project Stargate.

While not all of this capex money will go toward Nvidia's GPUs, the company will get more than its fair share. At the same time, the stock is not expensive, trading at forward price-to-earnings (P/E) ratio of 26.5 times 2025 analyst estimates and a price/earnings-to-growth (PEG) ratio of 0.5. PEG ratios under 1 typically indicate a stock is undervalued.

Image source: Getty Images.

2. Broadcom

Another chipmaker set to benefit from increased AI infrastructure spending is Broadcom (AVGO -6.33%). While Nvidia is the leader in mass-market GPUs, Broadcom is carving out a nice niche as a maker of custom AI chips. Its application-specific integrated circuits, or ASICs, are designed for very specific tasks. As such, its custom AI chips outperform GPUs for those specific tasks and tend to have more efficient power consumption. However, they do not have the flexibility of GPUs.

Broadcom's first custom AI chip customer was cloud computing giant Alphabet, whose Trillium tensor processing unit (TPU) is designed specifically to work within Google Cloud's TensorFlow (a software library for AI and machine learning). Alphabet has said that using a combination of TPUs alongside GPUs has resulted in improved inference times and cost savings.

Since then, Broadcom has been adding other customers, which are believed to be Meta Platforms, TikTok owner ByteDance, OpenAI, and more recently, Apple. It has said its three oldest customers (Alphabet, Meta, and ByteDance) could each deploy 1 million AI chips by its fiscal 2027 (ending October 2027), representing a $60 billion to $90 billion opportunity. Its newer customers could add to those totals. It took about 15 months for Alphabet's chips to be designed and then deployed in its data centers, which was considered very fast.

Trading at 30 times forward P/E, Broadcom is well-positioned to benefit from increased AI infrastructure spending.

3. Taiwan Semiconductor Manufacturing

Another company set to continue benefiting from increased AI infrastructure spending is Taiwan Semiconductor Manufacturing (TSM -4.58%), or TSMC for short. The company is the leading semiconductor contract manufacturer in the world. Due to the cost to build out manufacturing plants (foundries), the need for them to be run near full capacity to be profitable, and the technological expertise needed to continually innovate and push chip sizes lower, most semiconductor companies today just design chips and let third parties manufacture them. TSMC counts both Nvidia and Broadcom among its top customers, along with its largest customer Apple.

TSMC has become the leader in making advanced chips, such as those used for AI, while other foundry competitors such as Intel and Samsung have struggled. As such, it has become an integral part of the semiconductor value chain. The company is continuously advancing its manufacturing technology and lowering the size of its chips, which makes them more powerful and energy-efficient. TSMC's market-leading position, meanwhile, has led to strong pricing power and expanding gross margins.

At the same time, the company is building new facilities to try to keep up with demand. The combination of adding capacity, pricing power, and improving margins bodes well for TSMC moving forward. At the same time, the stock is inexpensive, trading at a forward P/E of just above 19 times and a PEG of 0.67 times.