The AI infrastructure race: I’ve seen this movie before

In 2022, when the automotive sector was captivated by autonomous vehicles, I was co-founder and chief technology officer at Luminar, developing lidar technology for self-driving cars. At an automotive conference, an executive remarked that I wasn't just a spectator but an integral part of the transformation. That statement resonated with me, highlighting a broader trend: when technology transforms an industry, the companies that initially built the infrastructure may not be the ones that shape the new one.

In autonomy, tech companies like Tesla, Waymo, and Zoox, rather than traditional automakers, led the charge. Today, I see a similar pattern emerging in AI infrastructure. Technology companies are confronting infrastructure built for a different era, rather than waiting for traditional power and telecommunications providers to catch up.

Like in the autonomy race, the technology firms are taking the reins, reimagining the entire system around software, AI, and compute. The stakes are higher this time, with hyperscalers like Google, Microsoft, Amazon, and Meta racing to build data centers to power the AI economy. Power is a significant constraint, as AI infrastructure is incredibly energy-intensive, and the grid cannot keep up.

Technology companies are investing directly in power generation, private optical networks, advanced cooling, and the technologies needed to connect AI campuses. They're becoming infrastructure companies themselves, building private optical networks and purchasing networking equipment directly from manufacturers. This shift mirrors the autonomy race, where tech companies were quicker and more agile than traditional automakers.

While utilities and telecommunications providers will remain crucial, the AI industry's rapid pace requires them to adapt quickly. We cannot afford to wait for traditional industries to catch up; instead, we must recognize the new players and their urgency in shaping the future of AI infrastructure.