Management claims company-wide ASP growth while gross margin hits 59.8%, the low end of guidance, and inventory builds 30% against a flat revenue outlook.
Thesis: Ambarella has a real structural position: 42M installed HAI SoCs, 370+ production customers, 5nm to 4nm to 2nm progression, clean FCF and DSO. But days of inventory rose from 76 to 99 in a single quarter while the Q1 guide implies another sequential decline, gross margin sits at 59.8% despite claimed ASP growth, and the won portion of the $13B auto pipeline is flat year over year. The beat is real; the demand durability question is not yet answered.
Verdict: HOLD — Conviction: MEDIUM
Catalyst: CV7 4nm revenue ramp beginning in Q4 FY2027 with premium ASP finalized, plus first ASIC silicon revenue early FY2028 proving the semi-custom model carries real unit economics.
Key Risk: Q1 revenue misses the $97M floor or gross margin breaks below 59%, confirming inventory build was channel stuffing rather than supply chain timing.
The Tell: Fermi Wang volunteered '$19,000,000,000' for the auto pipeline unprompted, then CFO John Young corrected it to '$13,000,000,000' on the same call. Wang had already answered a pipeline growth question using the wrong number, suggesting the figure management wants to anchor is higher than the one they're reporting.
Friction Level: MODERATE_FRICTION — Both sides agree on the numbers: clean DSO, honest FCF, inventory spike, flat won pipeline, margin compression. Bull reads installed base and node transitions as structural moat. Bear reads inventory divergence and Say/Do Gap as demand softening masked by narrative.
2026Q4 · Feb 26, 2026
3 of 5 readings agreed3 HOLD · 2 AVOIDRead the full call
Inventory Up 30% in One Quarter, Guide Implies Sequential Revenue Decline
Ambarella has a real structural position: 42M installed HAI SoCs, 370+ production customers, 5nm to 4nm to 2nm progression, clean FCF and DSO. But days of inventory rose from 76 to 99 in a single quarter while the Q1 guide implies another sequential decline, gross margin sits at 59.8% despite claimed ASP growth, and the won portion of the $13B auto pipeline is flat year over year. The beat is real; the demand durability question is not yet answered.
Fermi Wang volunteered '$19,000,000,000' for the auto pipeline unprompted, then CFO John Young corrected it to '$13,000,000,000' on the same call. Wang had already answered a pipeline growth question using the wrong number, suggesting the figure management wants to anchor is higher than the one they're reporting.
Days of inventory went from 76 to 99 in one quarter, a 30% build, while Q4 revenue fell 7% sequentially and the Q1 midpoint guide of $100M implies a further sequential decline. Management's explanation is 'to support our current level of business.'
Management claims company-wide ASP growth in both FY26 and FY27, yet non-GAAP gross margin fell from 62.7% to 60.7% on 37% revenue growth, with Q4 landing at 59.8%, the low end of guidance. Either ASPs rose and foundry/mask costs ate the increase, or the ASP claim covers mix shift.
CEO stated $19B auto pipeline, CFO corrected to $13B on the same call. The won portion, the only part that converts to revenue, is flat year over year. Management stopped reporting the probability-weighted figure (~$2B previously) they used to quote.
FY26 revenue $390.7M beat and exceeded by 37.2% growth. Q4 at $100.9M beat the midpoint of the $97-103M guide. FY27 guide of 10-15% growth with unit and ASP increases. Seventeenth consecutive year of positive FCF at $58M.
42 million HAI SoCs installed, 370+ unique customers in production, 200+ model architectures running on Cooper platform. Customers rebuild software stacks to switch, creating real lock-in across the design-in base.
-28
CAUTIOUS"Fermi Wang on FY27 growth drivers: 'We are trying to understand, and working with customers to understand, their new product ramp-up and how it is going to impact our growth this year.' Guidance carried by 'we believe' rather than 'we are confident.'"
Specific unit counts, customer names, balance sheet detail, and pipeline numbers disclosed. Deflections on ASIC gross margin economics, ISV channel revenue, and channel business model leave key unit economics unknown.
CV7 4nm revenue ramp beginning in Q4 FY2027 with premium ASP finalized, plus first ASIC silicon revenue early FY2028 proving the semi-custom model carries real unit economics.
Q1 revenue misses the $97M floor or gross margin breaks below 59%, confirming inventory build was channel stuffing rather than supply chain timing.
FY26 revenue of $390.7M up 37.2%, HAI up 50%, with beat-and-raise execution across the full year. 42M HAI SoCs in the field across 370+ production customers and 200+ model architectures on Cooper creates real switching costs. FCF of $58M on 15% of revenue, DSO flat at 36, cash up $62M to $312.6M. Node progression from 5nm to 4nm to 2nm GAA positions ASP expansion. CV7 carries premium ASP versus CV5 with design wins engaged. First ASIC silicon revenue lands early FY28 with significant NRE upfront. Warehouse robotics deployment in production, drone franchise in first full production quarter. Management guides FY27 for both unit and ASP growth inside the 59-62% GM range. 25 million units shipped across 15+ SoCs despite industry supply constraints. This is a fabless company with pricing power and a compounding installed base.
Days of inventory surged from 76 to 99 in one quarter, a 30% build, against a Q1 guide implying further sequential revenue decline. Gross margin at 59.8% sits at the low end despite claimed company-wide ASP growth, a Say/Do Gap. OpEx accelerated to 12.9% growth against 6.5% prior, making operating leverage negative at 10-15% revenue growth. The $13B auto pipeline's won portion is flat year over year, and management dropped the probability-weighted metric they previously disclosed. ISV channel has 'no meaningful revenue this year.' ASIC economics undisclosed with first silicon revenue only early FY28. CV7 has no material revenue this year with ASP negotiations unfinished. WT Microelectronics at 73.1% of Q4 revenue creates partner concentration risk. IoT is roughly 50% consumer/prosumer including drones and 360 cameras, directly exposed to DRAM cost inflation that management claims won't affect them.
Inventory Divergence
DOI jumped from 76 to 99 days (+30%) in one quarter while Q1 guide implies sequential decline. Management's 'to support our current level of business' is the standard channel stuffing script when demand warrants no buildup.
Backlog Quality
CEO said $19B pipeline, CFO corrected to $13B same call. Won portion flat YoY. Probability-weighted figure (~$2B) quietly discontinued. Only the won portion converts to revenue and it's not growing.
Say/Do Gap
Claiming company-wide ASP growth while non-GAAP gross margin compressed 200bps to 60.7% for the year and hit 59.8% in Q4. Either node transition costs ate the ASP gain or the claim is mix shift spin.
Mix Shift Spin
Margin compression from 62.7% to 60.7% attributed to 'mix shift into new product generations' while management simultaneously claims ASP growth. The two claims are in tension.
HOLD. Cash flow is clean, DSO is flat, and the installed base is real. But inventory up 30% against declining guide, margin at low end despite ASP claims, and won pipeline flat means the demand durability question is open. Add on proof: Q1 revenue above $103M or CV7 ASP finalized above CV5. Short if Q1 misses $97M floor or GM breaks 59%.