Silex Microsystems Buys Onsemi's Pennsylvania Fab for $40 Million, Opening a US MEMS Manufacturing Base to Chase Mag 7 Demand
M&A call held July 7, 2026, following signing of asset purchase agreement
Silex Microsystems, the Swedish pure-play MEMS foundry that listed earlier this year, has signed a binding agreement to acquire an 8-inch semiconductor fab from Onsemi in Mountain Top, Pennsylvania, marking the company's first manufacturing footprint outside Sweden. The deal, structured as an asset purchase rather than an acquisition of the legal entity, covers the real property, production infrastructure, manufacturing equipment and roughly 130 existing employees. Total consideration is $40 million, split between $10 million at signing and $30 million at closing, which is expected at year-end 2027 pending CFIUS approval, a process the company's lawyers estimate will take three to six months.
A Financially Engineered Transition Period
The structure of the deal is unusual and, on balance, favorable to Silex. Onsemi will continue operating the fab to run down its legacy IC production over the next 18 months, while Silex simultaneously qualifies and ramps MEMS products inside the same facility ahead of taking full ownership. CEO Edvard Kalvesten framed this as the central logic of the transaction: "The advantage with this long closing time is that they have the time to finalize their production, while we, in parallel, can ramp up our development." Rather than paying for idle capacity, Silex effectively gets a running start on qualification while Onsemi absorbs the cost of winding down its own operations.
Why Geopolitics, Not Economics, Forced the Move
The most important disclosure on the call was not the price tag but the customer pressure behind it. Silex generates 55% of revenue from US dollar-based customers, and management was explicit that its largest accounts, including several of the Mag 7 hyperscalers, have been pushing for domestic capacity for some time. Kalvesten said the company had retained these customers so far only because Silex is the largest MEMS foundry with capabilities unavailable elsewhere, but conceded that "long term, this was a necessity." He added that "the most important is really that you are where your customers are... because of all the geopolitical reasons we have nowadays, which can change in both directions." That last phrase is worth flagging: management is explicitly hedging against a bifurcating semiconductor supply chain, not simply chasing lower costs. Analysts pressed on how tangible that customer demand actually is. Kalvesten declined to disclose committed volumes but described "very positive response" from both existing and new customers once the transaction became public, and confirmed that at least one large existing customer is expected to shift a program from Sweden to the new US site as it outgrows current capacity, a datapoint that suggests the deal is not purely speculative capacity-building.
Capital Intensity and a Long Runway to Breakeven
Investors should size the financial commitment carefully. Including the $40 million purchase price, Silex expects to spend approximately SEK 1 billion in capital expenditure across 2026 and 2027, followed by roughly SEK 200 million annually through 2030 to build out MEMS-specific tooling inside the acquired shell. Management targets EBIT breakeven only in 2029 or 2030, and by 2034 expects the US fab to mirror the Swedish operation's 2025 financial profile: revenue of approximately SEK 1.4 billion with an EBIT margin around 27%, or 29% on an adjusted basis. That is an eight-year runway from signing to margin parity with the existing business, a timeline management declined to compress when pressed by SEB's Erik Lindholm-Rojestal on interim loss guidance. Kalvesten's response, "we don't guide on that... you have to do your own calculations," leaves a meaningful modeling gap for analysts to fill independently. Funding will come primarily from existing cash, including proceeds from the SEK 1 billion IPO raise, with an undrawn leasing or debt facility available if needed. CFO Maria Engstrom noted the company is evaluating potential CHIPS Act or state-level grants but has nothing to announce yet.
Why an Old IC Fab Rather Than Building from Scratch
The rationale for buying a legacy fab rather than constructing new capacity centers on cost efficiency. Kalvesten explained that a typical MEMS fab is composed of roughly 80% standard semiconductor equipment and only 20% MEMS-specific tooling, meaning the Onsemi facility hands Silex the bulk of the infrastructure at a fraction of build-from-scratch cost, with MEMS-specific equipment layered in over the next several years. The site's existing cleanroom is 3,000 square meters but can expand to 12,000 square meters of previously used cleanroom space, versus roughly 4,000 square meters at the Swedish facility today. In other words, management believes the US site has structurally more room to scale than the home fab, "even much bigger than the Sweden fab long term, if we do the necessary investments." Management was careful to frame this as additive rather than cannibalistic. Kalvesten said only one or two programs would migrate from Sweden to the US, with the Swedish fab continuing to receive parallel investment, including plans to build the industry's first 12-inch pure-play MEMS line there around 2030.
Execution Risk Concentrated in Going Multi-Site
When asked directly about key risks, Kalvesten pointed less to cost overruns; he said discussions with equipment suppliers give no indication of delays or budget slippage despite tight conditions in the broader semiconductor equipment market; and more to organizational complexity. "It's always a risk that a new location, a new fab is challenging for a company like Silex being only in one site. Suddenly we have two sites to take care of," he said, while framing the second continent as predominantly opportunity rather than threat given the small size of the company at roughly 500 employees. Analysts from Nordea and ABG pushed for granularity on utilization thresholds and margin comparability between the two fabs, but management repeatedly declined to translate its Swedish rule-of-thumb metrics, previously around 40% utilization to reach 80% gross margin, directly onto the US facility, citing the different equipment mix. That reluctance to commit to specific interim financial bridges, paired with firm long-term 2034 targets, leaves a wide, unmodeled middle period that investors will need to track closely through the 2026-2028 ramp.