They expected Q4 down 3%, it printed up 8%, and the CFO disclosed the 10-point swing unprompted during Q&A.
Thesis: AEIS is generating $235M of operating cash on $1.8B of revenue with 240bps of structural margin expansion from China closure and volume leverage, and they've built capacity for $3.5B in the ground ahead of a high-teens growth guide. The Q4 beat came from a 10-point customer forecast revision in one quarter, which makes it fragile, but the sub-2nm design wins converting and the data center guide excluding second-wave customers give asymmetric upside. Own it for the execution, size it for the forecast risk.
Verdict: LONG — Conviction: MEDIUM
Catalyst: Second-half 2026 semiconductor acceleration. Customers told them to expect further increases for H2 and Kelley links it into a healthy 2027. Semi system power slots begin ramping in 2026. Thailand first products qualify in Q4.
Key Risk: Double-buffer inventory reversal. AE is building strategic inventory while customers hold more safety stock off the same forecast. If H2 2026 semi forecasts slip, both sides destock simultaneously off a 10-point swing that Paul says they didn't predict.
The Tell: Paul Oldham volunteered unprompted: 'We expected Q4 to be a down quarter based on everything our customers were telling us. At the end of Q3, we expected to be down 2% or 3%. We actually ended up growing 8%. That's a 10% swing.' Nobody asked about the beat's fragility. The CFO just told you the outperformance came from a forecast revision he didn't see coming, one quarter before guiding semi revenue as the primary sequential driver.
Friction Level: MODERATE_FRICTION — Both sides agree cash flow is real and margins are expanding. Disagreement is whether the Q4 beat reflects structural share gain at sub-2nm or a one-quarter forecast swing that reverses, and whether the channel double-buffer is insurance or a setup for double ordering.
2025Q4 · Feb 10, 2026
3 of 5 readings agreed3 LONG · 2 HOLDRead the full call
Forecast Swung 10 Points in One Quarter, Then They Raised
AEIS is generating $235M of operating cash on $1.8B of revenue with 240bps of structural margin expansion from China closure and volume leverage, and they've built capacity for $3.5B in the ground ahead of a high-teens growth guide. The Q4 beat came from a 10-point customer forecast revision in one quarter, which makes it fragile, but the sub-2nm design wins converting and the data center guide excluding second-wave customers give asymmetric upside. Own it for the execution, size it for the forecast risk.
Paul Oldham volunteered unprompted: 'We expected Q4 to be a down quarter based on everything our customers were telling us. At the end of Q3, we expected to be down 2% or 3%. We actually ended up growing 8%. That's a 10% swing.' Nobody asked about the beat's fragility. The CFO just told you the outperformance came from a forecast revision he didn't see coming, one quarter before guiding semi revenue as the primary sequential driver.
Three consecutive quarters of beating guidance. Q4 revenue hit high end at $489M, EPS beat at $1.94, and they raised data center growth from 25-30% to above 30% for 2026 while absorbing sub-100bps tariff headwinds.
Gross margin expanded 240bps in 2025 to 38.7%, driven by China factory closure and volume leverage, not accounting adjustments. CFO guides above 40% in 2026 and reaffirms 43% long-term target with Thailand fully baked in.
Capacity built for $3.5B in revenue ($2.5B existing footprint plus $1B Thailand) against $1.8B 2025 revenue. Management explicitly states factory floor space and equipment are not constraints; the constraint is bill of materials.
CapEx at $107M or 6% of revenue in 2025, holding near that level in 2026, funded by record $235M operating cash flow. Management is building capacity with cash, not debt.
AE building strategic inventory (turns guided down a tenth or two from 2.9x, 125 days on balance sheet) while simultaneously confirming customers will hold more safety stock because 'we all have memories of what happened back in 21.' Both sides buffering off the same forecast.
CFO states 'clearly well over 40% gross margin excluding tariffs' but guides Q1 flat at 39.5-40% and concedes tariffs persist. Tariffs called a headwind three separate times. Management frames as one-time, numbers show recurring.
-18
CAUTIOUS"Kelley says 'we don't know exactly how much we're gonna grow' and 'we'll be happy once the dust settles' while calling the forecast conservative. CFO guides EPS flat on higher revenue. Confidence on direction, refusal on magnitude."
Full transcript with specific segment revenue, margin bridge, capacity numbers, and inventory metrics. Management gave direct answers on capacity, 800V ASPs, and channel inventory. The only gap is visibility into semi growth magnitude.
Second-half 2026 semiconductor acceleration. Customers told them to expect further increases for H2 and Kelley links it into a healthy 2027. Semi system power slots begin ramping in 2026. Thailand first products qualify in Q4.
Double-buffer inventory reversal. AE is building strategic inventory while customers hold more safety stock off the same forecast. If H2 2026 semi forecasts slip, both sides destock simultaneously off a 10-point swing that Paul says they didn't predict.
“Fourth-quarter revenue of $489 million increased 6% sequentially and 18% year over year.”
Figures are extracted from the transcripts and checked in code: every quote appears word for word. Descriptive only: none of this feeds the verdict.
Beat-and-raise execution across three consecutive quarters. Gross margin expanded 240bps in 2025 with line of sight to 40% in 2026 and 43% long-term, structural drivers are China factory closure and volume leverage, not accounting. Data center revenue up 101% YoY to $178M in Q4 with guide raised to above 30% and zero second-wave contribution baked in. Semi inflecting ahead of schedule: Q4 came in 10 points above customer signals from end of Q3, Q1 and Q2 demand moved up after, and H2 customers forecast further increases. Everest, EVOS, and NavX converting sub-2nm design wins into double-digit millions of revenue in pilot with qualification pull-through. Capacity built for $3.5B against $1.8B base, funded by $235M OCF, CapEx at 6% of revenue with Thailand qualifying this year. Operating leverage printing: OpEx grew 7% against 21% revenue growth in 2025. I&M genuinely recovering with three sequential quarters of growth, channel inventory down, backlog up.
Semi grew 6% in 2025 against mid-teens WFE growth. AE's own market grew more than twice as fast as they did. The Q4 beat came from a 10-point forecast swing in one quarter that management didn't predict, and forecast revisions reverse. Data center approached 40% of revenue from low-20s in twelve months, derivative of a single hyperscaler capex cycle, gated by GPU, ASIC, and memory allocation that 'dictate the number of boxes' they can ship. New product revenue is double-digit millions against $1.8B with revenue deferred to 2027-2029 by management's own framing. EPS flat at $1.94 on $10M of sequential revenue growth from tax rate reset to 16-17% and convertible pushing shares to 39.7M. Both AE and its customers are building safety stock off the same forecast-driven semi recovery. Thailand's $1B sits empty until Q4 at earliest. CFO says 'well over 40% ex-tariffs' but guides 39.5-40% flat.
Say/Do Gap on gross margin
CFO says 'clearly well over 40% excluding tariffs' but guides Q1 at 39.5-40% flat and tariff headwinds called out three times on the call. Management frames as one-time, the persistent guidance suggests recurring.
Inventory Divergence
AE building strategic inventory with turns guided down from 2.9x while customers hold more safety stock per Kelley's explicit confirmation. Both sides buffering off the same forecast-driven semi recovery. If H2 forecasts slip, double destock.
EPS flat despite revenue growth
Q1 guide: $500M revenue (up from $489M) but EPS flat at $1.94. Tax rate reset from 14.7% actual to 16-17% guided and convertible note pushing share count to 39.7M absorbs the operating leverage.
Mix Shift Spin
Data center approaching 40% of revenue flagged as potential margin headwind by CFO, offset only by new products still in pilot at double-digit millions of revenue against $1.8B total.
LONG, medium conviction. The execution is real: $235M OCF, 240bps structural margin expansion, three straight beats, capacity in the ground. The Q4 beat was fragile (10-point forecast swing) and the channel double-buffer is a real reversal risk. Size for the risk, own for the execution. The asymmetry favors long because the bear's concerns are timing-dependent reversals in future quarters, not present deterioration in the numbers. Add on any semi-driven pullback ahead of H2 2026 proof point. Trim if inventory turns deteriorate beyond the guided tenth or two.