AI Needs More Power and Faster Data, Not Just Chips

AI's need for energy is growing fast, like comparing it to many nuclear reactors. This shows a big problem with power supply.
By Newsroom | Technology, Business |

The relentless surge in artificial intelligence, while largely framed around computational power, is increasingly revealing other, more fundamental constraints. While giants like Alphabet Inc. and its peers are channeling vast resources into processing capacity, a different narrative is emerging: the critical need for robust physical infrastructure. This shift presents opportunities for companies addressing what is becoming the true bottleneck – namely, energy and data flow.

The discourse around AI’s limitations is moving beyond raw compute power to confront the physical realities of its expansion. Energy demands and the capacity to move data are now seen as the critical choke points, potentially eclipsing the focus on processing chips themselves. This evolving landscape is drawing attention to specific companies positioned to capitalize on these infrastructure challenges.

Beyond Compute: The Energy and Data Dilemma

The conventional wisdom places the heart of AI’s progress in processing power. However, recent analyses highlight a growing consensus that electricity is becoming the new bottleneck in AI growth. As computing capacity expands, so does its insatiable demand for power, with some comparisons drawing parallels to the scale of nuclear reactors. This underscores a critical underinvestment in energy infrastructure relative to the explosive growth in AI-driven computation.

Read More: Ex-Netflix Engineer Releases Free AI Cost-Saving Tool January 5 2026

Beyond energy, the ability to transmit data at the speeds required for advanced AI applications is proving equally vital. The development of next-generation optical networking technologies, particularly those supporting 800G and 1.6T pluggable transceivers, is crucial for enabling this high-speed data flow. Companies specializing in the components and assembly of these advanced modules are emerging as key players.

Key Companies in the New AI Infrastructure Narrative

While the spotlight often remains on semiconductor giants, several other firms are carving out significant roles in the AI ecosystem by addressing these emergent bottlenecks:

  • GlobalFoundries (NASDAQ: GFS): This company is taking a contrarian approach, arguing that the primary constraint is not compute but the underlying infrastructure.

  • Lumentum Holdings (NASDAQ: LITE): Specializing in electro-absorption modulated lasers (EMLs) and pump lasers, Lumentum is critical for the production of high-speed optical transceivers. The company anticipates substantial growth in its data center segment.

  • Coherent (NYSE: COHR): Coherent is focused on building complete transceiver modules, supporting speeds from 800G to 1.6T, and utilizing both EML and silicon-photonics platforms.

  • Fabrinet (NYSE: FN): This firm's facilities in Thailand are instrumental in the precision optical packaging required to turn components from companies like Lumentum and Coherent into functional modules for 800G and 1.6T applications.

A Crowded Field of AI Investment Vehicles

The broad interest in artificial intelligence has led to a proliferation of investment options, including numerous exchange-traded funds (ETFs) specifically focused on AI and related technologies. These range from broad AI innovation and technology funds to those targeting robotics, automation, and even specific aspects like AI semiconductors and power infrastructure. The sheer volume of these funds indicates a widespread, though perhaps undifferentiated, belief in the long-term trajectory of AI.

Read More: HP OmniBook 3 16: Good Battery, Bad Screen for Paris Buyers

Background: The Evolving AI Landscape

The initial phases of AI development were characterized by advancements in algorithms and increasing computational power, largely driven by innovations in chip design. Companies like Nvidia have been at the forefront of this wave. However, the exponential increase in AI model complexity and deployment has exposed the limitations of this singular focus. The sheer scale of operations now demands a critical re-evaluation of the physical underpinnings, including energy grids, cooling systems, and high-speed networking, which are essential for sustaining and scaling these powerful technologies. This re-evaluation is likely to reshape investment priorities within the technology sector.

Frequently Asked Questions

Q: Why is AI growth slowing down?
AI growth is slowing because it needs a lot more electricity and faster ways to move data. These are becoming bigger problems than just having powerful computer chips.
Q: Which companies are important for AI's future energy and data needs?
Companies like GlobalFoundries are looking at infrastructure. Lumentum, Coherent, and Fabrinet are making parts for faster data transfer needed by AI.
Q: What is the new problem for AI growth?
The new problem is that AI uses so much electricity, and sending data quickly enough is hard. This means power plants and fast internet cables are very important now.
Q: How is AI infrastructure changing?
Before, people thought only about computer chips for AI. Now, people see that we need more power and better ways to send data. This changes where companies and investors are focusing their attention.