AI Infrastructure Boom Drives $3B AT&T-Corning Deal and Power Efficiency Plays

Key points

• AT&T and Corning announced a multiyear agreement exceeding $3 billion for optical fiber to support AI-driven network expansion.
• Corning states AI data centers require over 10 times more optical fiber than traditional facilities, driving demand for high-speed connectivity.
• Vicor Corp. is positioned to benefit from AI power demands, offering vertical power delivery technology that reduces heat and energy loss in processors.
• MaxLinear and Corning are highlighted as suppliers for AI data highways, with MaxLinear providing chips for 400G, 800G, and 1.6T connections.
• Everpure (formerly Pure Storage) is seen as a winner in AI storage, with its FlashBlade platform addressing the need for fast access to corporate data for AI applications.
• Analysts suggest that as AI shifts from training to inference, demand will broaden to include networking, storage, and power efficiency across the entire system.

Background

This shift follows Nvidia’s forecast of a $3–$4 trillion AI market by 2030 and PwC’s projection of $31.6 trillion in global data center capex by 2050. Recent legislative blocks on data center bills in the US Senate have also highlighted the growing political and economic significance of AI infrastructure.

How outlets are covering it

InvestorPlace emphasizes the physical demands of AI, such as power and connectivity, over chip-centric views. Yahoo Finance’s data highlights market volatility, with Western Digital and Seagate dropping over 10%, while Space Exploration Technologies rose 7.35%, reflecting broad market movements beyond just AI infrastructure.

Why it matters

As AI adoption expands, the bottleneck is no longer just computing power but the physical infrastructure to support it. Companies providing power efficiency, fiber optics, and storage are becoming critical to the AI economy, offering investment opportunities beyond traditional semiconductor stocks.

What to watch

Investors should monitor the adoption of 1.6T connections and the performance of power delivery technologies as AI inference workloads grow. The success of these infrastructure plays will depend on sustained revenue growth and the ability to manage execution risks in a rapidly evolving market.