Jabil Raises AI Revenue to $13.6 Billion and Confirms Third Hyperscaler in Its Portfolio

EMS reported Q3 with revenue of $8.8 billion (+12% y/y) and raised annual AI guidance by $500 million. CEO Mike Dastoor confirmed a third contract with a hyperscaler and a multi-gigawatt opportunity with Adani.
Jabil announced its fiscal Q3 2026 results on Wednesday (17) with revenue of $8.8 billion, a 12% increase over the same period last year, and raised its annual revenue guidance related to artificial intelligence for the second time this year. The contract manufacturer now expects $13.6 billion in AI-related revenue for fiscal 2026, $500 million above the March projection and approximately a 50% growth from the $9 billion reported in 2025. The stock surged about 14% in pre-market trading in New York.
Adjusted earnings per share stood at $3.16 for the quarter, compared to a GAAP earnings per share of $2.59. "We delivered a very strong third quarter, with results above our expectations in revenue, core operating margin, core EPS, and free cash flow," said Mike Dastoor, CEO of Jabil. Regarding AI demand, he added: "The demand for AI infrastructure remains extremely strong, and our annual revenue projection related to AI is now significantly higher."
The Third Hyperscaler
A key highlight from the earnings call was the confirmation of what Dastoor referred to as the "third hyperscaler" in the portfolio, without naming the company. The contract encompasses the data center infrastructure space, overlapping with the second client already announced in previous cycles. Jabil estimates initial revenue of "several hundred million dollars" in fiscal 2027 and the potential to exceed $1 billion starting in 2028, as capacity ramps up. In an industry where three buyers (Microsoft, Meta, and AWS) account for most of the global capex in AI servers, acquiring the third client practically means capturing the majority of the available book.
The effect on the balance sheet is structural. In 2025, AI accounted for approximately 30% of Jabil's revenue. With the new projection, in 2026, that share exceeds 38%, in a group whose historical base is manufacturing for automotive, healthcare, and consumer goods. The complete annual guidance rose to $35 billion in revenue, with a core operating margin of 5.8% and diluted core EPS of $12.70.
The Indian Front with Adani
Dastoor also confirmed that the partnership with Adani Enterprises, announced in May, aims for multi-gigawatt manufacturing of AI infrastructure in India, with significant revenue expected only in fiscal 2028. The CEO stated that the main bottleneck today is building physical capacity: "We are limited by the speed of building out, not by demand." This move aligns Jabil with New Delhi's efforts to attract electronics manufacturing under the PLI incentive policy and simultaneously addresses pressure from American hyperscalers to diversify production away from China.
For readers following the IT services industry, this signal is significant. Jabil shares a top-tier EMS oligopoly with Foxconn, Flex, and Sanmina and was the first in the group to separate the "AI-related revenue" line in its official reporting. In the U.S., the immediate takeaway is validation of the margin thesis for AI capex players. In Taiwan, where Foxconn has already announced plans to manufacture Nvidia GB200 servers in Wisconsin, Jabil's move increases pressure for exclusivity of contracts with hyperscalers.
The Interpretation Beyond the Earnings Call
In India, the agreement with Adani is the third major project announced in AI manufacturing in the last twelve months, following Tata's factory for Apple components and Foxconn's expansion in Bengaluru. The local base of hardware engineering talent remains shallow, and Jabil will need to import senior engineers from Penang and Mexico during the ramp-up. For Brazil, where Jabil has operations in Manaus since 2002 focused on networks and healthcare, the focus on AI does not bring a significant transfer of capacity in the short term. The latest local expansion of the company, in 2024, was toward a line of home appliances.
What the results clearly indicate is that AI capex is not slowing down, contrary to what analysts projected in December 2025 when Nvidia reported its first margin slowdown. Instead, it is spreading to the tier of manufacturing suppliers. The next reading will come in July when Foxconn and Flex report, and investors will compare how many hyperscalers each carries in its portfolio.