AI infrastructure stocks Lumentum, Celestica, Seagate beat Nvidia 2025

Nvidia has been the biggest infrastructure winner of the artificial intelligence boom, soaring nearly 13x in value since the end of 2022, reaching a market capitalization of $4.6 trillion.

Nvidia’s rally continued into 2025, but investors who bet on other AI data center efforts made even more money over the past 12 months.

With the four largest technology companies expected to spend $380 billion on building data centers and infrastructure this year, and rising in the coming years, Wall Street is pouring money into various vendors looking to reap the benefits.

The value of manufacturers of memory, storage, fiber optic cables, central processors and other types of enterprise hardware has soared this year as the AI ​​boom grows.

Investors will likely scrutinize these five companies carefully in 2026, as there are high expectations built into their stock prices.

lumenBased in San Jose, California, the company manufactures switches, transceivers, and other optical laser-based components needed for fiber-optic cables. Customers are typically carriers and device manufacturers, such as: applepreviously used Lumentum parts for the FaceID sensor.

However, AI servers also require a lot of optical connectivity. All graphics processing units in the rack must be connected to all other GPUs. Future AI systems will scale out and require rack-to-rack optical connectivity. Eventually, entire data centers will need to be interconnected with fiber optic connections.

Lumentum’s stock price has risen 361% this year, giving the company a market capitalization of more than $27 billion. Revenue for the most recent quarter was $533 million, up 58% from the same period last year.

“Our growth is fueled by the demand for AI across laser chips and optical transceivers in data centers and the interconnected long-distance networks that link them,” Lumentum CEO Michael Hurlston said on an earnings call in November. He said 60% of the company’s revenue now comes from cloud and AI infrastructure.

Sales are expected to rise 58% in the fiscal year ending June, according to LSEG, but analysts expect them to slow from there to 32% and 15% growth over the next two years.

western digital It is one of the three major hard drive manufacturers. seagate And Toshiba. The 55-year-old company’s stock is up 296% this year.

In addition to computing power, AI companies require increasingly more space to store applications and other data. That means data centers need hard drives.

Western Digital makes solid-state hard drives that use chips to store data, but the company is best known for its hard drives that use spinning disks to store terabytes or more of data.

“Data is the fuel that powers AI, and HDDs provide the most reliable, scalable and cost-effective data storage solution,” CEO Irving Tan said during an October earnings call. He gave the example of a hospital deploying AI to analyze 7 billion images.

Revenue for the most recent quarter increased 27% to $2.82 billion. The company said increasing sales of storage for data centers will improve profitability as AI companies require larger and more expensive hard drives.

Sales are expected to increase by approximately 23% in fiscal 2026, but growth is expected to slow to 13% in 2027.

In February, Western Digital spun out its flash business. sandiskCurrently has a market capitalization of about $35 billion, more than half of Western Digital’s value.

micron is one of the three major memory manufacturers along with Samsung and SK Hynix, but the only one based in the United States.

Artificial intelligence servers require large amounts of memory to store and process large AI models. Nvidia chip or advanced micro device It has several gigabytes of state-of-the-art memory, called high-bandwidth memory. Chipmakers have monopolized all memory production capacity, leading to global shortages and soaring prices.

Micron beat Wall Street expectations for sales and profits in its quarterly report last week, and its stock is up 228% for the year.

Sumit Sadhana, Micron’s head of business, said the company’s memory chips are “more than sold out.” In December, it shut down its consumer memory and solid-state drive product lines to conserve supplies for AI.

Morgan Stanley analysts said in a December note that, other than Nvidia, Micron’s results showed the best revenue and profit upside in “the history of the U.S. semiconductor industry.”

Sales are expected to nearly double in the year ending August, but slow significantly to 24% in fiscal 2027 and less than 1% in 2028, LSEG said.

seagatewhich was founded nine years after Western Digital, is also benefiting from the surge in storage demand. The stock is up 228% this year.

Revenue for the company’s fiscal third quarter, which ended Oct. 3, increased 21% to $2.63 billion. The company said at the time that 80% of its sales were going to the data center market.

“There’s no question that AI is increasing the economic value of data and data storage and reshaping demand for hard drives,” CEO Dave Mosley said on a conference call with analysts.

Bank of America analysts wrote in November that Seagate was unlikely to stock additional hard drives because any additional shipments would be billed immediately. Analysts also noted that customers have made-to-order contracts with firm quantity and price commitments.

“Additional units will typically be purchased by hyperscale and other high-capacity customers,” the analysts wrote in recommending the stock to buy.

Seagate’s trajectory is similar to Western Digital’s. Analysts expect sales to rise 21% this fiscal year, and about 15% and 6% in the following two years, according to LSEG.

Celesticawas established in 1994. IBM The subsidiary manufactures switches that interconnect networks and manage the data and traffic flowing through them.

The stock is up 213% this year.

The company sells many of its switches to the largest AI buyers. Third-quarter sales increased 28% to $3.19 billion. Analysts expect sales growth to rise from 26% this year to 33% in 2026 and 34% in 2027, according to LSEG.

Celestica CEO Robert Myonis said on an October earnings call that hyperscalers recently approached the company to manufacture components that connect to water-cooled rack-scale computers for AI, and that mass production is expected to begin next year.

One boon for Celestica is the surge in demand for custom chips called ASICs. Although ASICs are less flexible than GPUs, they may be cheaper to operate for certain AI applications.

“Our largest and fastest growing market presence is within our AI data centers, supporting high-performance networking and custom ASIC AI/ML computing platforms,” Mionis said.

Goldman Sachs analysts wrote in a note Friday that Celestica supplies components such as: Google ASIC.

“The company should benefit from being the leading provider of Google TPU rack-level solutions in 2026,” the analysts wrote.

Attention: AI is expected to become more efficient by 2026