Jabil is no longer interesting because it assembles other companies' boards; it is interesting because the mix inside a still-thin manufacturing franchise is shifting toward artificial intelligence infrastructure faster than the multiple has fully decided whether that shift is structural. The investment case turns on whether Intelligent Infrastructure, the segment that now carries cloud, networking, and data-center work, can keep compounding volume while the rest of the portfolio stops leaking enough profit to cancel the gain. That is a mix argument, not a cycle argument, and the two are easy to confuse when the same plants are running both kinds of work.
The June print was the first clean look at that mix after the Hanley Energy Group close, and the mechanism is more important than the beat. Management raised the full-year AI-related revenue outlook to $13.6 billion after winning a third hyperscale customer and watching networking ramps in India run hotter than the March plan. That is not a one-program spike. It is the same land-and-expand motion that turned the second hyperscaler from a single capability into a broader data-center conversation, and it is why the company now treats power, cooling, and rack integration as the product rather than as add-ons around contract assembly.
The tension is that the equity still behaves like a cycle stock even as the income statement starts to look less cyclical. Inventory days sat at 84 in the May quarter. That print sat well above the internal target band, because Intelligent Infrastructure shipments slipped into the fourth quarter. Five customers still account for roughly a third of sales, and book equity is thin after years of buybacks. The bear case that survives the beat is simple: this is still a high-turnover assembler whose peak earnings power is being capitalized as if the AI buildout were a permanent take-or-pay contract.
The next resolution date is the September investor briefing, when management has said it intends to put a full fiscal 2027 framework around the early claim that AI-related growth can stay near this year's rate on a much larger base. Until that number is public, the market is pricing a margin path above six percent without having seen the working-capital unwind or the third hyperscaler's first meaningful revenue. A briefing that merely repeats June language without a numbered plan leaves the current multiple hanging on hope rather than on a forecast an outsider can test.