Flex Ltd. announced it will spin off its Cloud and Power Infrastructure segment into an independent publicly traded company, separating roughly $4.2 billion in annual revenue from its core electronics manufacturing services business. The segment, which builds rack-scale infrastructure and power distribution systems for hyperscalers, currently operates at 11% EBITDA margins—above the 8.7% blended rate for Flex's other divisions.

The decision arrives eighteen months after Flex acquired Anord Mardix for $318 million, consolidating its position in modular data center infrastructure. That acquisition added $420 million in annual revenue and gave Flex direct contracts with three of the four largest U.S. cloud providers. Management has spent the intervening quarters attempting to convince public equity investors that cloud infrastructure deserves a different multiple than automotive electronics or industrial sensors. The market declined. Flex trades at 0.43x trailing revenue, below peers like Jabil at 0.51x and Sanmina at 0.48x. The spin is an admission that the conglomerate discount is real and durable.

The separated entity will enter public markets with established contracts across hyperscale deployments in Northern Virginia, Phoenix, and Dublin. Data center construction timelines run 22 to 28 months from design freeze to energization, which means the spin-off inherits a revenue backlog extending into late 2026. Power infrastructure for AI clusters—where a single rack can draw 120 kilowatts versus 8 to 12 kilowatts for traditional compute—commands premium pricing and faster refresh cycles. The unit's gross margins have expanded 240 basis points year-over-year as AI-related orders have grown from 14% to 31% of segment revenue.

Allocators should note three implications. First, the spin creates a pure-play public equity vehicle for data center infrastructure at a moment when hyperscaler capex guidance for 2025 sits north of $230 billion. Second, Flex's remaining business becomes a cleaner automotive and industrial play, likely re-rating closer to Sanmina's valuation within two quarters. Third, the separation eliminates transfer pricing opacity that has masked true margins in both divisions—expect the spun entity to report adjusted EBITDA closer to 13.5% once corporate allocations disappear.

The transaction is expected to close in the second half of 2025, subject to board approval and an SEC registration statement. Flex has not yet named the spin-off's CEO, though three internal candidates have run verticals within the segment for more than four years. The company will host a separation structure call in April, where management will detail the capital structure, any debt allocation, and whether the new entity will carry a dividend. Watch for the S-1 filing, which should land in May and will clarify customer concentration—if any single hyperscaler represents more than 35% of revenue, the equity will trade at a discount until the second customer diversifies the book.

The spin does not solve Flex's core problem, which is that contract manufacturing is a scale and execution business where differentiation is measured in basis points. But it does let two different investor bases own two different bets. The cloud infrastructure entity will trade on capex visibility and margin expansion in AI workloads. What remains of Flex will trade on automotive electrification and industrial automation, where the company has $11.8 billion in active programs. Both businesses are better off without the other's balance sheet and strategic ambiguity.