Onsemi's $175m Ferric earnout and $500m AI target measure different things

Onsemi's $175m Ferric earnout and $500m AI target measure different things

The full $175m earnout depends on Ferric revenue reaching a redacted threshold in any four straight quarters by 31 December 2029. The $500m increase in onsemi's 2030 AI data centre target is a separate measure.

The purchase price and the $500m revenue target

Onsemi agreed on 2 November to buy Ferric, a maker of integrated voltage regulators. The agreed price is $600m in cash plus up to $175m. The cash is subject to adjustments. The $175m depends on Ferric's specified product and licence revenue.

On the evening call, chief executive Hassane El-Khoury said the deal lifts the 2030 AI data centre revenue target by $500m to more than $3bn. Asked if all of it came from the acquisition, he said: "This is tied to the acquisition."

The most the sellers can receive is $775m, and 22.6 percent of that is contingent. The 8-K adjusts the upfront $600m for Ferric's cash, debt, transaction costs and working capital. Unlike the $175m, the $600m does not depend on a revenue threshold, and neither payment is triggered by the $500m target.

The cash price

The $600m equals 14.4 percent of the $4.16bn of cash onsemi held on 2 October, per its third quarter financial release. The full $775m equals 18.6 percent. That is a scale comparison, not an estimate of cash out the door, because the $175m is contingent.

The pending Synaptics purchase is larger, at roughly $5.7bn in cash. A cash snapshot from 2 October cannot show onsemi's debt or liquidity once both deals are done. Finance chief Thad Trent said on the call that Ferric is funded from the balance sheet, and that once Synaptics closes, cutting debt is "our top priority".

What onsemi gets

Onsemi proposes to buy a way to regulate power closer to an AI processor. Ferric puts the controller, switches, inductors and capacitors of a voltage regulator into one package. Onsemi says this cuts board area by 50 to 70 percent against discrete regulators, and that the industry is moving regulation closer to the processor to reduce losses.

The 50 to 70 percent figure is about board area, not electricity saved. El-Khoury also claimed the best transient response in the industry and said Ferric's height beats the 2.5 millimetres he called typical. These are onsemi's claims, and the release names no customer.

CTOL's view: Ferric's technology gives onsemi a plausible way to compete at the Vcore stage if the deal closes. Customer adoption and investment returns are unproved.

What management did not promise

El-Khoury expects Ferric to raise onsemi's share in "the Vcore stage of the AI data center power tree", in an opportunity he put "well beyond 2030". Asked about revenue and design wins, he declined to guide. He said Ferric had been engaged "with all of the hyperscalers from a design and platform architecture perspective".

That is design-stage contact, not an order. Onsemi has published no order, forecast or assumption that links it to $500m of revenue.

The company projects "no meaningful near-term financial impact". El-Khoury also acknowledged rivals: "There are other innovative solutions out there," he said, including ones that are "not necessarily magnetic".

If the deal closes, onsemi gets about 50 employees, mainly in New York and Silicon Valley, and more than 40 patents. Neither figure shows customers have adopted the product.

How the earnout works

The earnout is one payout formula, not a ladder of milestones. Under the merger agreement, the sellers get the full $175m if specified revenue in any four consecutive calendar quarters between closing and 31 December 2029 reaches a threshold.

If that never happens, the payout scales with revenue above a floor in the four quarters ending 31 December 2029. At or below the floor it is zero. The filing redacts both dollar figures.

The definition is narrow. It covers sales of Ferric's specified products, including those inside modules for voltage core applications, and separate licences. It excludes onsemi's own products that incorporate Ferric technology.

On CTOL's reading, sales of those own products could count toward onsemi's AI data centre revenue while staying outside the earnout. So the two measures differ in scope and in date. The earnout tests defined Ferric revenue through 2029. The $500m is management's addition to a 2030 segment goal. Neither figure can be used to calculate the other.

Strong segment growth could coexist with no earnout if qualifying revenue stays at or below the floor. An earnout could also become payable without proving the 2030 segment target will be met.

The $500m against onsemi's current base

Trent called the raise 20 percent, which on CTOL's arithmetic puts the earlier goal at about $2.5bn. AI data centre revenue was $180m in the third quarter, about 10.5 percent of the $1,722.3m total. That is up 39 percent on the second quarter and 175 percent on a year earlier. Those results come from before the Ferric agreement and say nothing about its contribution.

Management guides to about $620m for 2026, with fourth quarter growth of mid teens percentage points, and a doubling in 2027 off a higher base. An exact doubling would give an illustrative $1.24bn. Reaching $3bn in 2030 would then mean cumulative growth of about 142 percent.

That is a sensitivity for the whole segment, not a forecast and not a Ferric number. The required rate falls if 2027 revenue is higher and rises if 2030 exceeds $3bn.

The $500m is targeted annual revenue in 2030, not earnings or realised sales. Onsemi has given no margin, investment need or ramp timing for Ferric, so the figure supports no payback or return calculation on the $600m.

Closing dates and reporting

Closing is planned for the fourth quarter, subject to customary conditions that include expiry of the Hart-Scott-Rodino waiting period. Either side may terminate if the deal is not done by 1 February 2027. Full payment follows onsemi's next regular financial statements after the threshold is reached.

If it is not reached, onsemi prepares an earnout statement for the sellers after the statements that follow 31 December 2029.

Revenue reports go to the sellers' representative within ten days of each 10-Q or 10-K, not to the market. That is the gap for investors: onsemi discloses total AI data centre revenue in public, while the Ferric figures that decide the payout stay private and the thresholds stay redacted.

Could organic growth explain later results?

One objection is that the target moves with onsemi's existing business. El-Khoury said the 2027 outlook is higher than it was 90 days ago, before the Ferric agreement. Organic growth could therefore explain some of the later results. But nothing on the record shows management put organic growth into its $500m raise.

Onsemi now reports AI data centre revenue as its own fourth market segment, which makes total growth visible. Segment growth alone cannot isolate a Ferric contribution. Management tied the $500m to the acquisition, but has disclosed neither Ferric revenue nor a forecast that bridges to $500m. Design engagement and claimed packaging gains suggest potential, not measured sales.

CTOL's view and the next evidence

The agreement provides $600m of upfront cash, subject to adjustments, and a possible $175m earnout. The commercial return on that money is unproved.

CTOL expects, with high confidence, that onsemi's fourth quarter report will not give both redacted thresholds and matching qualifying Ferric revenue. If it gives both, the call is wrong. That says nothing about whether Ferric succeeds.

Three outcomes, in order of likelihood:

• Most likely: the report updates the segment against the $620m 2026 guide and does not reveal the earnout calculation. Redacted thresholds and private reporting support this.
• Plausible: the report names a Ferric design win or product or licence revenue. That would show traction, not hyperscaler volume sales or closeness to the undisclosed thresholds.
• Unlikely on the current record: the report supplies the thresholds, qualifying revenue and a bridge to the $500m. That needs disclosures not yet made. The earnout measurement ends on 31 December 2029, and the segment goal concerns 2030.