Earnings Call Highlights: Fabrinet (FN) | Q1 FY2027 — Revenue Up 51%, Fastest in a Decade; a New Building Starts Before $10B of Capacity Is Even Finished

1 day ago

9 min read

1. Three Key Takeaways

First, data center already drives about 70% of the growth, before the new programs count. Data center revenue was $777M, up 78% YoY, or 53% of total revenue. Of the company's $496M YoY increase, $340M came from this segment. CEO Seamus Grady said in Q&A that OCS and NPO are "not even in the numbers yet," and that multi-rail is only just getting started.

Second, this contract manufacturer's capacity plan can't keep up with customer forecasts. Building 11 went from about $2.1B of revenue capacity to about $3B, and long-term total capacity moved from $14B to $16B. That is two straight quarters of upward revisions: $11.5B two quarters ago, $12.5–14B last quarter, $16B now.

Third, the cost of this growth is already in the cash flow. Operating cash flow was just $21M against $159M of capex, for free cash flow of −$137M, and cash and short-term investments fell $119M in the quarter. That money buys future capacity, so whether the new programs reach the numbers on schedule in the second half is the real test of this bet.

2. Revenue and Financials

The market usually reads Fabrinet on a non-GAAP basis, but the gap between GAAP and non-GAAP was wide this quarter. The earnings card shows GAAP; both are side by side below:

Metric	GAAP	YoY	non-GAAP	YoY
Revenue	$1,474.5M	+50.8%	$1,474.5M	+50.8%
Gross margin	11.8%	−0.1pp	12.1%	−0.2pp
Operating income	$146.0M	+55.0%	$158.5M	+53.0%
Operating margin	9.9%	—	10.8%	+0.2pp
Net income	$103.5M	+7.9%	$157.4M	+49.5%
Diluted EPS	$2.86	+7.5%	$4.34	+48.6%

Revenue rose 12.1% QoQ and came in 3.5% above the $1.425B high end of guidance; non-GAAP EPS of $4.34 also beat the $4.25 high end. This was the 14th straight quarter of growth and the 7th straight quarter of accelerating YoY growth.

GAAP net income grew only 7.9% because of two non-cash items. One is an unrealized loss of about $21.5M on the minority stake in Raytek. The other is a provision of about $19.9M for the Pillar Two global minimum top-up tax in Thailand. Both are excluded from non-GAAP. Raytek is valuation noise; last quarter it was a $56.7M gain. Pillar Two deserves a different reading. It has now appeared two quarters in a row, is worth about $0.55 per share, and CFO Csaba Sverha says the offsetting credits are "progressing" but not yet implemented. Until they are, the 4% non-GAAP effective tax rate is an optimistic baseline.

Gross margin was 12.1%, down 10bp QoQ and 20bp YoY. The sequential decline came from annual merit increases in Q1, the YoY decline from FX. Operating expenses held at 1.3% of revenue, so non-GAAP operating margin rose 20bp YoY. The leverage is still there.

▲ Fabrinet FY2027 Q1 core financials on a GAAP basis (revenue, gross margin, operating margin, net income, EPS, YoY, segment revenue, next-quarter guidance). (Image source: Simple Tech Trend original; data from Fabrinet Q1 FY2027 earnings release and SEC filings)

Segment Breakdown

Data Center: $777M, YoY +78%, QoQ +16%, 53% of revenue. Transceivers, DCI and HPC all grew.
Communications Infrastructure: $444M, YoY +35%, QoQ +8%, 30% of revenue. Growth was spread across telecom systems, components and satellite communications.
Automotive, Industrial and Other: $253M, YoY +20%, QoQ +8%, 17% of revenue. Lasers drove the sequential increase, as the CFO said Fabrinet is making headway with customers that have historically favored in-house manufacturing.

Cash Flow Breakdown

Cash and short-term investments ended the quarter at $757M, with about $75M of borrowings. Of the $159M in capex, $77M bought the second Santa Clara campus. The remaining $82M or so roughly matches the CFO's "about 5% of revenue" normal run rate. He also said FY2027 capex will rise further as Building 11 and other capacity additions layer in.

The reason for borrowing is simple: debt in Thailand costs less than breaking the cash invested in the U.S. The board added $103M to the buyback authorization, bringing the remainder to $200M. The CFO was clear on priorities: growth investment first, surplus cash to the 10b5-1 plan second, opportunistic repurchases third.

3. Capacity Breakdown: $5.8B, $10B, $13B, $16B

Seamus re-ran the whole roadmap on this call:

Stage	Added capacity	Cumulative annual revenue capacity	Timing
End of FY2026	—	~$5.8B	June 2026
Building 10 (2M sq ft)	+~$3.5B	—	Completes January 2027; initial revenue already
Nava One	+~$400M	—	Up and running
Second Santa Clara campus	+~$300M	~$10B	By January 2027
Building 11 (1.5M sq ft)	+~$3B	~$13B	Breaks ground January 2027; ~18 months; end of FY2028
Building 12 (if built)	+~$3B	~$16B	Undecided

Three numbers are worth pulling apart.

Building 11's footprint is only 25% larger, but its capacity is 43% higher. Last quarter's plan was about 1.2M sq ft and at most $2.1B; now it is 1.5M sq ft and $3B. There are two reasons: this building needs little office or warehouse space, so more of it is production floor, and products are getting smaller and denser, which keeps lifting revenue per square foot. That works out to about $1,750 per sq ft for Building 10 and about $2,000 for Building 11.

Existing capacity is already leaning on the new plants. Q1 annualized revenue of about $5.9B already exceeds the $5.8B of capacity at the end of FY2026, and the Q2 guidance midpoint annualizes to about $6.5B. Accelerating construction and qualification on parts of Building 10 was not simply building ahead of demand.

Management says it can't wait for Building 11, and the math supports that. Q2 annualizes to about $6.5B, leaving 54% of room to $10B. A rough STT estimate: at 10% sequential growth, Fabrinet would reach $10B in about four to five quarters, or early 2028, ahead of Building 11 coming online in mid-2028. This is only an extrapolation of the current slope, not company guidance, but it explains why Fabrinet is already looking for land and existing factory space. For last quarter's full roadmap, see Fabrinet Q4 FY2026 Earnings Highlights.

4. Marvell Buys Lumentum: Two Customers Become One

On the morning of the call, Marvell announced an all-stock deal to acquire Lumentum at 4.5 Marvell shares per Lumentum share, for an implied enterprise value of about $115B, expected to close in 12 to 18 months. Analysts asked about it right away.

Seamus first said he had not yet spoken with either company, then pointed to history: Lumentum buying NeoPhotonics, Infinera buying Coriant, Nokia buying Infinera. Each time there was concern for Fabrinet, and each time the business grew afterward. Nokia becoming a 10% customer for the first time last quarter is the latest example.

STT reads this as a positive too, but the old playbook doesn't fully apply.

Near-term risk is limited. Fabrinet's 10% customers in FY2026 were Cisco, NVIDIA, Nokia and Amazon. Neither Lumentum nor Marvell is on that list, and both say they will operate independently until the deal closes.
The long-term question is how much stays in-house. The merger documents repeatedly highlight Lumentum's specialized manufacturing and Marvell's high-volume production. How much of the back end the combined company keeps and how much it outsources is the real variable for Fabrinet. Past cases were mostly optical companies buying optical companies. This one pairs silicon with lasers, at a different scale and level of vertical integration.

We broke down the upstream logic of this deal in Marvell Wants InP Capacity. For Fabrinet the same deal raises another question: once the laser capacity is locked up, who does the back-end packaging and assembly?

5. Technology and Business Highlights: New Programs Just Getting Started

NPO and CPO. Seamus used the word "confident" for NPO: several NPO programs should start contributing revenue in the second half of this fiscal year. There are CPO programs too, but those are "a little bit further out." The wording clearly puts NPO ahead of CPO, which fits Celestica's first rack-scale CPO program ramping in late 2027. For where the NPO competition actually sits, see OCP Global Summit: NPO Lock-in Sits in the Board.

OCS. Capacity is being installed. Asked whether OCS and NPO could each be multi-hundred-million or even billion-dollar businesses, Seamus called that "a fair way to think about it" and said each could be a multi-hundred-million-dollar opportunity, but that Fabrinet hasn't "really started to ship anything yet." The same day, Lumentum's earnings release said OCS will become a multi-billion-dollar annual revenue stream and announced a second major NPO/CPO win backed by more than $5B of purchase commitments. Fabrinet did not link those orders to itself, so the two should not be equated.

Multi-rail. This is a newer class of optical line system that treats several parallel fiber pairs as the unit of growth. It calls for heavy fiber management, splicing and precision assembly, which is Fabrinet's strength. Capacity is being installed here too, and the program is just getting going.

New transceiver customers. Hyperscale direct and merchant programs are both ramping. Asked how many new OEMs there are, Seamus said only "more than one," and declined to answer when pressed on whether it was more than two.

CPO won't eat pluggables. Asked by a Needham analyst about the market's cannibalization worry, Seamus called the either/or view overly simplistic and cited USFP, the ultra-dense pluggable roadmap from Cisco and NVIDIA. For the specs, see USFP Explained. Of course, a contract manufacturer that builds all three form factors has no need to pick a side, so that view should be weighed accordingly. On coherent-lite, the company said only that it fits its capabilities and has announced nothing specific.

6. Management Outlook: Conservative Guidance, Stronger Language

Q2 FY2027 guidance (quarter ending December 25, 2026):

Revenue of $1.60B–$1.65B, with the midpoint up about 43% YoY and about 10% QoQ
Non-GAAP EPS of $4.77–$4.92; GAAP EPS of $3.81–$3.96
All three segments expected to grow YoY and QoQ, led by data center

One number from the transcript needs correcting. Seamus twice said the Q2 midpoint would be up $151M year over year, but $1.625B minus Q1's $1.4745B is about $151M, so that is the sequential increase. Working back from the 43% growth rate, the YoY increase is about $490M.

The other thing to watch is the acceleration streak. The Q1 guidance midpoint also implied +43%, and the actual was +51%, 5.3% above the midpoint. For an 8th straight quarter of acceleration, Q2 revenue has to exceed about $1.71B, again about 5.3% above the midpoint. In other words, the company needs to repeat a beat of the same size.

The language can be read in three tiers:

Highest certainty: "strong confidence" in the Q2 outlook, with the CFO also citing improving visibility.
Medium: Asked whether FY2027 growth can top FY2026's 36%, Seamus said he is as positive as 90 days ago, "if not a little bit more." First-half growth is already implied at about 47%, and he gave no second-half number.
Strongest, and most in need of a discount: customers show demand "as far as the eye can see." That is a forecast, not an order. Seamus added his own hedge: the model is asset-light with low fixed costs, so even a "doomsday scenario" wouldn't leave much added fixed cost. Still, one self-funded building after another is very real cash going out the door.

On gross margin, the CFO expects no further FX headwind in the second half of FY2027 and improvement over time as Fabrinet adds more value, but noted that fixed costs are only about 5%, which limits leverage at the gross margin level. Operating margin is where the leverage mainly shows up.

7. Supply Chain and Customer Clues

Components are still tight. An analyst inferred from the beat that components were getting easier to find. Seamus pushed back: supply is tight across the board. Fabrinet has simply been getting its "fair share and, in some cases, unfair share," and could always ship more with more parts. That matches the upstream bottleneck in last week's review, and it is a reminder that however fast the buildings go up, the near-term ceiling may still be the laser or chip upstream.

Customers are coming in earlier. Seamus said some engagements start at the concept stage, helping customers work out how to build products that are manufacturable with predictable yield. That is a contract manufacturer's source of pricing power as photonics and electronics converge, and it underlies the CFO's view that gross margin will improve over time. It still has to show up in the reported numbers.

The Santa Clara signal. The second campus more than doubles Silicon Valley capacity, and a handful of customers are already in discussions. This is the entry point for new product introduction, so a longer queue here usually leads Thailand production by one to two years.

8. Summary

Three things to watch next quarter:

Whether NPO, OCS and multi-rail start reaching revenue. Management has pointed to the second half of this fiscal year, which makes it the easiest promise to verify.
Free cash flow and gross margin. With Building 10 finishing and Building 11 starting, cash will keep going out. Watch whether non-GAAP gross margin recovers from 12.1% as management implied, and when operating cash flow catches up with earnings.
Whether the acceleration streak reaches 8 quarters, and whether the Pillar Two credits land. The first takes about $1.71B of revenue; the second decides how long the 4% tax rate holds.

In one line: Fabrinet has gone from worrying about whether there is enough demand to worrying about whether there is enough land. The real risk isn't next quarter's guidance. It is whether those forecasts stretching as far as the eye can see turn into the orders that one building after another requires.

This article is for technology and industry trend analysis only and is not investment advice.