Nvidia Put $2 Billion Into Coherent. Is Optical Bandwidth the Next AI Bottleneck?

Nvidia has invested $2 billion in Coherent, an optical technology company, signaling a strategic bet that data transmission infrastructure may become the limiting factor for AI compute scaling. The investment underscores a shift in focus from raw processing power to the interconnects and bandwidth that move data between chips and systems - a challenge that becomes more acute as AI models grow larger and require faster communication between processors.

The move suggests chipmakers are identifying optical networking as a potential solution to bottlenecks in data center architecture. By improving how efficiently information flows through data centers, optical systems could unlock gains in both performance and power efficiency that traditional electrical interconnects struggle to provide at scale.

• Nvidia's $2 billion Coherent stake bets on optical technology as critical infrastructure for next-generation AI systems, not just processors themselves.
• Data transmission bandwidth and latency may represent the next constraint on AI scaling, after years of focusing primarily on chip compute improvements.
• Optical networking could improve data center efficiency and scalability, but widespread adoption remains dependent on technological maturation and cost dynamics.