Google will pay SpaceX $920 million a month for compute, driven by unexpected AI demand. That number is so large it briefly reframes what infrastructure means. In the same news cycle, a whistleblower lawsuit alleges IBM covered up multiple data breaches in its cybersecurity subsidiaries during the mid-2010s. One story is about the scale of AI's appetite. The other is about what gets buried when scale becomes the only metric that matters. They are more connected than they appear.

Infrastructure Scale as Accountability Shield

The Google-SpaceX deal is remarkable not just for its size but for its candor. Google's statement cited unexpected demand for recently launched AI products as the driver. That is a company publicly admitting it underestimated its own infrastructure needs by the equivalent of a small nation's GDP, on a monthly basis. When systems operate at that scale, the incentives to disclose problems invert. The IBM whistleblower case is about breaches from a decade ago, but its logic is current: when you are critical infrastructure for governments and enterprises globally, admitting a breach is an existential reputational event. The cover-up is rational from inside the logic of scale.

Compute as Geopolitics, Breaches as Archaeology

The New Yorker's piece on how the Iran conflict is reshaping the global economy adds a third layer. Green energy and compute infrastructure are the two sectors that benefit most from geopolitical disruption, and Google just made a bet the size of a mid-size country's annual budget on exactly that thesis. The IBM story is the archaeology of what the last infrastructure buildout left behind: undisclosed breach, institutional cover, and a whistleblower who waited years before the legal moment was right. The question for AI infrastructure in 2026 is not whether similar archaeology exists. It is how long before someone starts digging. A 2025 paper in IEEE Security and Privacy by Cavusoglu et al. found that breach disclosure timelines are inversely correlated with company size, meaning the larger the organization, the longer the delay. Google and SpaceX are now very large organizations indeed.