
Last 4 earnings calls · through Q4 FY2026
The bull case is that AI data center demand keeps outrunning the capacity Jabil can bring online, so growth depends on execution rather than on customers changing plans.
Jabil sells integration across compute, power, cooling and networking, which lets it win more of a rack than a maker of any single component would.
Management has raised its revenue outlook at each of the last three quarterly updates, which suggests it sets targets it expects to beat.
New capacity is tied to booked programs and costs little capital, so revenue growth converts into cash that funds buybacks.
The bear case is that a handful of hyperscale buyers decide when and where AI capacity gets built, and Jabil cannot schedule its factories without them.
Memory and high-density circuit boards are being allocated towards the largest buyers, which limits what Jabil can ship in its smaller segments.
Every new plant carries startup expense and low initial yields, so the profit from this growth arrives later than the sales do.
Connected Living is still being pruned and healthcare slipped in the last quarter, so the non-AI parts of the portfolio do not balance the AI build.
In short: AI demand drives rapid growth, but the outlook rests on customer ramps and supply Jabil cannot control.
Management kept a third hyperscaler out of its numbers a year ago; it now counts the program as won and expects the relationship to widen across the data center, as the second customer's did.
Management expects roughly $7.7 billion of added AI revenue on a much larger base, which is the main reason it guides total company revenue up about 24%.
A market management described as one it stayed conservative on is now expected to grow, with defense and aerospace added, making Regulated Industries a second contributor instead of a drag.
Management says facilities do not perform from the day they open, so much of the guided growth has to come from capacity that is not yet running at rate.
Jabil keeps exiting consumer programs that miss its return requirements, so that business subtracts from growth even as warehouse automation and robotics expand.
In short: Margins and cash flow are improving, but new capacity adds startup costs and inventory ahead of demand.
A year ago management would not say which year 6% would arrive; it now builds 6.1% into the fiscal 2027 outlook and attributes the gain to better mix and higher capacity utilization.
Cash generation of that size funded the capacity additions and the buyback at the same time, and management guides to about $1.6 billion next year on the same asset-light model.
Buybacks mean earnings per share rises faster than operating profit, which is why guided earnings growth runs well ahead of the guided revenue increase.
Management expects margins to be lighter in the first half because yields, training and expenses come before volume, so the full-year margin depends on a second half that has not been delivered yet.
On large rack assembly programs Jabil passes through a lot of material, so gross margin moves little with volume and operating margin has to come from mix and overhead leverage instead.
In short: Integration wins hyperscale work, but a few large AI customers now drive much of revenue.
Not competing with the silicon and architecture a customer has chosen widens the programs Jabil can win and keeps its place even when the customer changes the design.
Registered sites, cleared staff and a qualified quality system take years to put in place, so defense and aerospace adds a regulated revenue stream outside the AI spending cycle.
Management expects six Intelligent Infrastructure customers above $1 billion next year, so a change in one hyperscaler's build plan moves a large part of the total.
Jabil builds in Mexico, India and Vietnam for US customers, so a change in tariff treatment can move where a program is manufactured and what it costs to deliver.