Generac Holdings has agreed to a long-term deal to supply Amazon with up to $8 billion worth of backup generators for the cloud and AI giant's data centers, according to Bloomberg's Michelle Ma1. As part of the arrangement, Generac granted Amazon a warrant for up to a 2.6% stake in the generator maker.
The deal is among the largest single infrastructure-supply commitments publicly tied to AI-driven data-center buildouts. Business Insider characterized the Amazon-Generac agreement as evidence that the AI trade may be shifting away from chips and into power2.
Data centers require uninterrupted electricity, and backup generation capacity has become a binding constraint as hyperscalers race to expand compute footprints for AI training and inference workloads. The equity warrant structure gives Amazon a direct financial interest in Generac's performance over the life of the supply agreement.
ANALYSIS The warrant mechanism mirrors the kind of strategic-supplier arrangements more commonly seen in aerospace or energy, where a large buyer locks in long-term capacity and aligns incentives through an equity stake. Structuring the deal this way ties Generac's upside to sustained delivery against what could be a multi-year, multi-site rollout.
The agreement lands in a week already marked by large-scale AI infrastructure financing. A consortium of 10 banks is extending a $22 billion loan to Crux AI, the Blackstone- and Alphabet-backed cloud venture, to fund TPU purchases[1]. ◆ Taken together, the Generac supply contract and the Crux AI debt facility illustrate the capital now flowing not just into chips but into the physical plant, power, and logistics layers that support large-scale AI compute.
The "up to $8 billion" framing indicates the contract value is a ceiling rather than a firm purchase order, and the 2.6% warrant cap sets the upper bound of Amazon's potential equity position in Generac. ◆ The ceiling structure suggests deliveries will be staged against Amazon's data-center construction timeline, with actual spend determined by the pace and geography of new facility deployments.