Peter Smith
⬤ Major tech companies are pushing to build massive gigawatt-scale data centers at breakneck speed—some in as little as one to two years from groundbreaking. The first of these facilities should be operational by 2026, representing one of the fastest industrial buildouts ever seen. Companies like Anthropic-Amazon, xAI, Microsoft, OpenAI, Meta, and Amazon are all racing toward 1-gigawatt capacity with aggressive timelines.
⬤ This rapid expansion is sparking policy debates about potential AI infrastructure taxes and regulations. Lawmakers want stricter oversight of these energy-hungry facilities, but industry leaders worry that new rules could slow construction, mess up funding plans, and push talent to other regions. For companies in the middle of this massive buildout, any delays could seriously hurt their competitive position.
⬤ The timelines are incredibly compressed. Facilities like Anthropic-Amazon's New Carlisle site and xAI's Colossus 2 are targeting 1-2 year builds, while OpenAI's Stargate Abilene and Microsoft's sites are looking at three years. These tight schedules reflect how desperately companies need more computing power for next-gen AI models and reliable long-term energy sources. Even Meta's Prometheus project—one of the slower builds—shows just how massive this infrastructure wave has become.
⬤ The industry is now defined by a scramble for power capacity, land, and specialized engineers. If these companies actually deliver working gigawatt facilities by 2026, they'll completely reset expectations for global computing infrastructure and shake up the AI competitive landscape. Construction speed itself has become a strategic weapon—whoever builds fastest wins.
Peter Smith
