As artificial intelligence scales, the physical limits of copper wiring are forcing a multi-billion dollar pivot to optical interconnects, creating a new class of winners in the data center supply chain.
The insatiable data demands of artificial intelligence are exposing the limits of traditional copper interconnects, forcing hyperscalers like Microsoft and Amazon to channel a significant portion of their planned $725 billion in 2026 capital expenditures toward optical networking solutions. This shift is creating a distinct investment opportunity in the companies that provide the underlying light-speed infrastructure.
"The AI buildout is creating supply shortages that favor the underlying infrastructure providers over the more overvalued computing stocks," according to research from industry analysts. "The market is beginning to recognize that data transport is becoming as critical as data processing."
The move is fueling a rally in optical specialists. Ciena (NYSE:CIEN) has seen its stock jump nearly 29 percent in 30 days following its re-inclusion in the S&P 500, while Nokia (NYSE:NOK) shares rose after an analyst upgrade highlighted surging demand for its optical transport and IP routing hardware. Other key suppliers include Lumentum (NASDAQ:LITE), which is collaborating with Nvidia on its silicon-photonics ecosystem, and Credo Technology Group (NASDAQ:CRDO), whose Active Electrical Cables (AECs) are critical for high-speed connections.
For investors, this represents a "picks and shovels" play on AI that extends beyond just chips. While companies like Ciena trade at a high 330x P/E ratio, the forecasted $7 trillion in AI infrastructure spending by 2030 suggests a long runway for growth. The core issue is that as AI models become more complex, the speed and bandwidth of data transfer between servers become a primary bottleneck, a problem that only optical solutions can solve at scale.
The race to build out artificial intelligence capabilities has entered a new phase, one defined not just by the power of semiconductor chips but by the speed of the networks connecting them. Hyperscale data center operators, including the likes of Meta Platforms, Alphabet, Microsoft, and Amazon, are collectively raising their AI capital expenditure budgets to an estimated $725 billion for 2026. A growing portion of this historic spend is being directed to solve a fundamental physics problem: the inadequacy of copper wiring for massive AI clusters. This is creating a powerful tailwind for a specialized group of optical networking and component suppliers.
The Optical Interconnect Players
At the center of this transition are companies providing the fiber-optic technologies essential for high-speed data transmission. Ciena (NYSE:CIEN), a provider of high-capacity optical connectivity, has returned to the S&P 500 index, drawing renewed investor attention. The company's order backlog is increasingly linked to AI infrastructure projects, though its shares trade at a steep 330 times price-to-earnings, reflecting high expectations.
Similarly, Nokia's (NYSE:NOK) network infrastructure unit is seeing heightened demand for optical transport and IP routing equipment. An April analyst upgrade, which sent trading volume surging to 119 percent above its three-month average, specifically cited the data center buildout as a key driver for future revenue.
The ecosystem extends to component makers essential for the technology to work. Lumentum Holdings (NASDAQ:LITE) manufactures high-speed transceivers and lasers, positioning it as a crucial supplier to hyperscalers. Its collaboration with Nvidia to develop the Spectrum-X Photonics networking switches underscores its role in the next generation of data center architecture. Meanwhile, Credo Technology Group (NASDAQ:CRDO) is gaining traction with its Active Electrical Cables (AECs), a key enabler for high-performance connectivity within AI-driven networking deployments.
Broader Infrastructure Demands
The AI-driven infrastructure boom is not limited to just optical components. The massive power consumption and cooling requirements of these new data centers are creating opportunities in other adjacent sectors. Johnson Controls (NYSE:JCI) is seeing a surge in demand for its data center chiller platforms, with its backlog growing 20 percent year-over-year to $18 billion.
Connectivity between data centers is another critical area. Cogent Communications (NASDAQ:CCOI), which operates 186 data centers, is experiencing a 73.7 percent year-over-year surge in its wavelength business, an optical transport platform. The stock trades at just 1.2 times sales, significantly below its historical average of 4.4 times, suggesting a potential recovery play if inter-data-center bandwidth becomes scarce.
Perhaps the most fundamental requirement is power. With global grids already under strain, companies like NuScale Power (NYSE:SMR) are developing small modular reactors (SMRs) as a dedicated power source for data centers. While meaningful revenue is not expected until around 2029, the long-term potential for nuclear-powered data centers is attracting significant attention.
The $7 Trillion Question
The core driver for all this activity is the sheer scale of capital being deployed. Research firm McKinsey & Co. has estimated that global AI-powered data center infrastructure capex will reach around $7 trillion by 2030. This enormous spending is creating crunches across the supply chain, from high-performance chips to the optical cables that connect them. For investors looking to participate in the AI buildout, the companies providing the essential, and increasingly scarce, infrastructure for this technological revolution present a compelling, if less hyped, alternative to the chipmakers themselves.
This article is for informational purposes only and does not constitute investment advice.
