Management volunteered that hyperscaler CapEx estimates rose from $500B in December to $740B post-earnings, an 48% move in eight weeks, and anchored their entire 2026 thesis to it.
Thesis: The compounding math on attach rate (mid-1s to 3+) and ASP ($3 to $4+) are structural trends that persist even if hyperscaler CapEx normalizes. Lattice is a fabless company on mature nodes with pricing power in small/mid-range FPGA and no advanced-node capex burden. The $20M beat versus consensus plus backlog through 2027 with scheduled orders confirms demand is real. The inventory build is manageable given long product life cycles and a debt-free balance sheet. Position sized for cycle risk, not full conviction.
Verdict: LONG — Conviction: MEDIUM
Catalyst: Q1 2026 prints $165M or above, confirming the acceleration thesis. Management also implied industrial channel inventory drops to 'the twos' in Q1, which would validate consumption tracking rather than channel stuffing.
Key Risk: Hyperscaler CapEx estimates that moved 48% in eight weeks reverse direction, collapsing the 2026 growth narrative while Lattice sits at peak internal inventory and peak confidence. Substrate and assembly lead times are stretching with management admitting 'very tight situation,' creating a Capacity Ceiling if demand holds but supply does not.
The Tell: Tamer told Quinn Bolton: 'if you use the high end of the guidance we gave you last quarter, you still don't get to the revenue we need to get to this.' This was unprompted escalation beyond his own 90-day-old forecast. He was telling you demand outran his model in one quarter, which is either genuine acceleration or order pull-forward.
Friction Level: MODERATE_FRICTION — Both sides agree cash flow is real and margins are expanding for genuine mix/pricing reasons. The disagreement is whether 2026 acceleration is structural (attach rate x ASP compounding) or purely a hyperscaler CapEx cycle peak, and whether internal inventory build during claimed demand surge is a yellow or red flag.
2025Q4 · Feb 10, 2026
5 of 5 readings agreed5 LONGRead the full call
Full-Year Growth Was 2.7%. The Q1 Guide Implies 37%. Both Numbers Are in the Same Document.
The compounding math on attach rate (mid-1s to 3+) and ASP ($3 to $4+) are structural trends that persist even if hyperscaler CapEx normalizes. Lattice is a fabless company on mature nodes with pricing power in small/mid-range FPGA and no advanced-node capex burden. The $20M beat versus consensus plus backlog through 2027 with scheduled orders confirms demand is real. The inventory build is manageable given long product life cycles and a debt-free balance sheet. Position sized for cycle risk, not full conviction.
Tamer told Quinn Bolton: 'if you use the high end of the guidance we gave you last quarter, you still don't get to the revenue we need to get to this.' This was unprompted escalation beyond his own 90-day-old forecast. He was telling you demand outran his model in one quarter, which is either genuine acceleration or order pull-forward.
Quinn Bolton noted the beat was nearly $20M versus consensus. Tamer acknowledged the prior quarter's segment ranges no longer reach the revenue needed. This was an upside surprise beyond even bullish buy-side expectations.
ASP moved from approximately $3 to above $4 while attach rate went from mid-1s to 3+ per server and server TAM grows from 15.3M to 16.5M units. Three compounding factors all rising simultaneously with volume holding.
Comms and compute booked through all of 2026 with orders extending five to six quarters out into 2027. Orders are scheduled throughout quarters rather than bunched, suggesting consumption-driven commitments.
Gross margin expanded 190bps to 69.3% on mix shift to higher-ASP companion chips. Operating margin up 340bps while OpEx was reduced 1%. Earnings growing 17% against 2.7% revenue growth confirms operating leverage is real, not narrative.
Incremental substrate and assembly orders placed since summer 2025, six months before guidance. Management is spending on supply ahead of confirmed demand, backing words with purchase commitments.
CFO states channel inventory is normalized yet the company is 'deliberately building inventory to support this growth environment.' Building stock while claiming orders run through 2027 creates tension between demand narrative and supply positioning.
Lead times increasing 'very, very fast' with substrate and assembly 'very tight.' Management's defense that orders are spread across quarters is exactly what hedge ordering looks like. Competitors being less forthcoming about lead times implies customers are ordering against supply fear.
Hyperscaler CapEx estimate went from $500B to $580B to $740B in eight weeks. Management anchored the entire growth thesis to this number. A figure that moves 48% in two months will eventually move down, and 2026 revenue depends on it.
Humanoids, robotaxis, aerospace defense all described as momentum and design wins destined for 2027. Tamer concedes he cannot quantify physical AI. Zero physical AI revenue in the 2026 guide. 100% of acceleration rides on data center CapEx.
Lattice is fabless on mature nodes and does not control the bottleneck. Substrate and assembly are constrained industry-wide. Management admits lead times are worsening quarter over quarter. If demand holds, supply wall arrives before demand wall.
+85
CONFIDENT"Tamer gave specific server unit TAM (15.3M to 16.5M), attach rates (mid-1s to 3+), ASP ($3 to $4+), and the exact CapEx estimate chain. CFO said 'tight grip on the knobs and levers.' Zero hedge words in prepared remarks. 'Strong' used 30+ times."
Rich quantitative detail: exact attach rates, ASPs, server unit TAM, segment growth breakdowns, CapEx estimates. Double-ordering question answered directly with specifics. Gaps: physical AI unquantified, guidance bracket widened to $14M without clean explanation.
Q1 2026 prints $165M or above, confirming the acceleration thesis. Management also implied industrial channel inventory drops to 'the twos' in Q1, which would validate consumption tracking rather than channel stuffing.
Hyperscaler CapEx estimates that moved 48% in eight weeks reverse direction, collapsing the 2026 growth narrative while Lattice sits at peak internal inventory and peak confidence. Substrate and assembly lead times are stretching with management admitting 'very tight situation,' creating a Capacity Ceiling if demand holds but supply does not.
“In Q4, we delivered $145.8 million in revenue, up 9.3% sequentially - our strongest sequential performance in seven years - and up 24.2% year-over-year.”
“In Q1 2026, we expect revenue to grow into the range of $158 million to $172 million.”
Figures are extracted from the transcripts and checked in code: every quote appears word for word. Descriptive only: none of this feeds the verdict.
Three compounding drivers are all moving up simultaneously: server units 15.3M to 16.5M, attach rate mid-1s to 3+, ASP $3 to $4+. This is structural, not cyclical. Full-year 2025 EPS grew 17% on 2.7% revenue growth, proving operating leverage is real. Cash flow confirms earnings quality: $175.1M operating cash flow on $523.3M revenue, FCF margin 25.3%. Debt-free with $250M buyback. Beat was $20M above consensus. Backlog through 2027 with orders scheduled across quarters. Channel inventory normalized so 2026 tracks consumption. Management placed incremental supply orders since summer, showing capital conviction. Design wins hit records for four straight quarters supporting 25%+ growth. Four IP tuck-ins completed. No mix shift spin, no receivables anomaly, no China masking detected.
Full-year growth was 2.7%. The Q1 guide implies 37% acceleration, entirely dependent on hyperscaler CapEx estimates that quadrupled upward in two months. Inventory build during claimed demand surge is a structural contradiction. Lead times stretching with explicit double-order risk denial and competitors withholding supply information is textbook pre-overorder behavior. Guidance bracket widened to 8.5% of midpoint versus 6.7% historically, and CFO's 'comfort with midpoint' language typically signals front-loaded quarters. Physical AI is unquantifiable by management's own admission with zero revenue in 2026 guide. Lattice is fabless on mature nodes and does not control the substrate or assembly bottleneck. Industrial and automotive declined 18% in 2025 and recovery assumes consumption tracks normalization, which last year's data proved was not guaranteed. Peak confidence language at peak CapEx estimates.
Inventory Divergence
CFO says channel inventory is normalized while simultaneously stating the company is 'deliberately building inventory to support this growth environment.' If orders run through 2027 as claimed, internal stockpiling contradicts the demand narrative.
Unsustainable Trend Confidence
Management anchored the growth thesis to hyperscaler CapEx that moved from $500B to $740B in eight weeks. A 48% swing in estimates in two months is not a trend, it is a repricing event. When it normalizes, the attached revenue forecast drops with it.
Backlog Quality
Lead times increasing 'very, very fast' with substrate and assembly 'very tight.' Orders spread across quarters is what hedge ordering looks like. Competitors withholding lead time information implies customers ordering against fear, not forecast.
AI Label Without Substance
Humanoids, robotaxis, aerospace defense, medical all dated to 2027 with zero revenue attached to 2026 guide. Tamer concedes he cannot quantify physical AI. The entire acceleration thesis rides on hyperscaler data center CapEx alone.
Capacity Ceiling
Fabless on mature nodes with no control over substrate or assembly bottlenecks. Management admits lead times worsening quarter over quarter. If demand is as strong as claimed, supply wall arrives before demand wall.
LONG at medium conviction. The compounding math on attach rate and ASP is structural and persists even if hyperscaler CapEx normalizes. Cash flow is clean, balance sheet is debt-free, margins expand on real mix shift, and the $20M beat confirms buy-side expectations were too low. Size for cycle risk. The single-driver dependency on hyperscaler CapEx and the inventory build during claimed demand surge argue against a full position. Add on Q1 print at $165M+. Cut if Q2 backlog quality questions surface or CapEx estimates reverse.