Negative free cash flow of $29.6M in one quarter against an $800M ten-year LTA that management calls 'potential revenue.'
Thesis: Ambarella is signing 5-10 year customer LTAs with no matching foundry supply commitment, while burning $29.6M in free cash to build 145 days of inventory against 0.5% sequential revenue growth. The CV7/CV72 5nm ramp and automotive record are genuine, but the LTAs are 'potential' revenue averaging $80M/year against a $400M base, and the inventory build must convert to revenue in Q2 or the stuffing pattern is confirmed.
Verdict: HOLD — Conviction: MEDIUM
Catalyst: Q2 days of inventory falling below 120 alongside revenue at or above $108M midpoint, proving the inventory build converts to recognized revenue and the CV7 ramp is real.
Key Risk: Days of inventory holding at 145 with Q2 revenue at $108M midpoint. That confirms stuffing and implies FCF burn continues through the year, draining the $277.8M cash position.
The Tell: When pressed on the 46-day inventory build, Louis Gerhardy answered first with a general 'we wanted to build a bit of inventory heading into this year,' then Fermi Wang interjected unprompted: 'Joe, it's Fermi. Just want to add another color on this.' He then attributed the build to TSMC tightening supply, contradicting his earlier script claim that 'we have secured our supply.' Two speakers and a self-correction on a balance sheet question is a tell.
Friction Level: MODERATE_FRICTION — Both sides agree the LTAs, CV7/CV8 ramp, and automotive record are real. The split is whether a 46-day inventory build with negative $29.6M FCF is prudent positioning for product ramps or stuffing ahead of guidance that can't support it. Q2 inventory days settle it.
2027Q1 · May 28, 2026
3 of 5 readings agreed3 HOLD · 1 AVOID · 1 LONGRead the full call
Inventory Up 46 Days While Revenue Grew 0.5% Sequentially
Ambarella is signing 5-10 year customer LTAs with no matching foundry supply commitment, while burning $29.6M in free cash to build 145 days of inventory against 0.5% sequential revenue growth. The CV7/CV72 5nm ramp and automotive record are genuine, but the LTAs are 'potential' revenue averaging $80M/year against a $400M base, and the inventory build must convert to revenue in Q2 or the stuffing pattern is confirmed.
When pressed on the 46-day inventory build, Louis Gerhardy answered first with a general 'we wanted to build a bit of inventory heading into this year,' then Fermi Wang interjected unprompted: 'Joe, it's Fermi. Just want to add another color on this.' He then attributed the build to TSMC tightening supply, contradicting his earlier script claim that 'we have secured our supply.' Two speakers and a self-correction on a balance sheet question is a tell.
Days of inventory rose from 99 to 145 in one quarter (+46%) while revenue declined 0.5% sequentially. $34.8M of cash consumed by inventory builds not matched by revenue acceleration. Management called it 'prudent' positioning for product ramps but could not quantify what portion maps to CV7/CV8 vs general stocking.
Management says record demand, LTAs provide visibility, and CV7 ramp is 'steep.' Operating cash flow was negative $25.6M against $5M non-GAAP net income. Free cash outflow of $29.6M in a single quarter. Q2 guide of $108M midpoint is seasonally normal, not validating the inventory position.
Delivered on all three guide metrics: revenue above midpoint at $100.4M, gross margin at top of range at 59.9%, opEx below midpoint at $56.4M. Q2 guide of $105-111M implies accelerating sequential growth in both auto and IoT. Consistent pattern of conservative guidance met with actual outperformance.
Corporate ASP of $15 on legacy parts while every new CV7x and CV8 product commands 'meaningfully higher pricing' per management. Gross margin held at 59.9% despite mix shift toward automotive, and long-term model of 59-62% reaffirmed. ASP inflection tied to node transition from 10nm to 5nm to 2nm.
Hanwha LTA of $800M over 10+ years is characterized as 'potential revenue,' not guaranteed. Averages $80M/year against a ~$400M annual revenue base. Two LTAs announced but management declined to quantify pipeline: 'There's no other concrete information we can disclose at this point.'
NRE funding from customers deepens the integration cost of switching. Hanwha relationship spans 15 years with co-development of two silicon generations. 12 SoCs under one SDK means customer software investment locks them across the portfolio. DRAM shortage response requires deeper FAE engagement, further embedding Ambarella.
No long-term supply contract with any foundry, confirmed directly by CEO: 'We don't have any long-term contract with any suppliers.' Signing 5-10 year customer commitments while relying on annual wafer allocation from Samsung at 2nm, where Samsung trails TSMC in node maturity. 18 years without a shortage is history, not a contract.
+5
NEUTRAL"Mostly direct and specific: Samsung 2nm commitment volunteered with a named press release, wafer supply answered without hedging, competitive landscape addressed with 12-SoC platform specifics. The inventory question required two speakers, Louis Gerhardy then Fermi adding 'another color,' which reads as defensive elaboration on a balance sheet item."
Full transcript with specific financials, cash flow detail, LTA terms, and foundry position disclosed. Gaps: no quantified LTA pipeline beyond Hanwha, no foundry supply contract detail, no breakdown of inventory by product line to validate the ramp positioning claim.
Q2 days of inventory falling below 120 alongside revenue at or above $108M midpoint, proving the inventory build converts to recognized revenue and the CV7 ramp is real.
Days of inventory holding at 145 with Q2 revenue at $108M midpoint. That confirms stuffing and implies FCF burn continues through the year, draining the $277.8M cash position.
“For fiscal Q1, revenue was $100.4 million, slightly above the midpoint of our prior guidance range of $97 million-$103 million.”
“We forecast a seasonally strong fiscal second quarter, with revenue in the range of $105 million-$111 million, or $108 million at the midpoint.”
Changed more than 79% of calls
Wording compared with the Feb 26, 2026 call, ranked among 97 calls reporting on Q1 2026. A measure of how much the story moved, not a signal about the stock.
Figures are extracted from the transcripts and checked in code: every quote appears word for word. Language change is computed from word frequencies, with no model involved. Descriptive only: none of this feeds the verdict.
Beat all three guide metrics simultaneously: revenue $100.4M above midpoint, gross margin 59.9% at top of range, opEx $56.4M below midpoint. Automotive posted an all-time record on commercial fleet telematics where 10-20% of a 100M-unit installed base is AI-enabled. Q2 guide of $105-111M implies accelerating sequential growth. CV75/CV72 5nm ramp is 'steep' and driving material incremental revenue this fiscal year. Hanwha LTA alone represents $800M+ potential over 10 years, with customer NRE funding the development. Samsung officially announced Ambarella alongside NVIDIA as a 2nm committed customer in a company-wide press release. CV8 2nm production starts H1 FY2028 with ASP above the $15 baseline. Robotics carries 15+ design wins with $100M+ lifetime revenue across 30+ customers, mostly on 5nm. Every new product commands higher ASP than legacy. Gross margin model of 59-62% reaffirmed despite automotive mix shift.
Free cash outflow of $29.6M in one quarter. Operating cash flow negative $25.6M against $5M non-GAAP net income. Cash and marketable securities down $34.8M sequentially to $277.8M. Days of inventory up from 99 to 145, a 46% jump, against 0.5% sequential revenue growth. $34.8M of cash consumed by inventory builds revenue hasn't justified. Management needed two speakers to answer the inventory question and Fermi contradicted his own script by citing TSMC supply tightening after earlier saying supply was secured. The $800M Hanwha LTA is 'potential,' averages $80M/year against $400M base, and management confirmed no other LTA pipeline detail. No foundry supply contract with any supplier, admitted directly by CEO. Signing 5-10 year customer commitments with zero foundry commitment. Q2 guide midpoint of $108M is seasonally normal and doesn't validate 145 days of inventory.
Inventory Divergence
Days of inventory rose 46% in one quarter (99 to 145) while revenue declined 0.5% sequentially. $34.8M cash consumed. Management attributed it to product ramps and supply tightening but offered no quantification of inventory by product line or expected conversion timeline.
Say/Do Gap
Management says record demand and LTA visibility while operating cash flow is negative $25.6M and free cash flow negative $29.6M against $5M net income. The cash burn contradicts the demand narrative until Q2 revenue converts the inventory build.
Backlog Quality
Hanwha $800M over 10+ years is explicitly 'potential revenue,' not guaranteed. Averages $80M/year against ~$400M base. Management declined to quantify pipeline: 'There's no other concrete information we can disclose.' LTAs signed faster than they convert to recognized revenue.
Capacity Ceiling
CEO confirmed no long-term foundry supply contract with any supplier while signing 5-10 year customer LTAs. Samsung 2nm trail TSMC in maturity. 18 years without shortage is a track record, not a contractual guarantee of capacity for CV8 production in H1 FY2028.
HOLD. The beat-and-raise pattern and pricing power on CV7x/CV8 are real. The inventory build and negative FCF are also real. The asymmetry hinges on Q2: if days of inventory drops below 120 with revenue at $108M+, the ramp thesis holds and this re-rates on LTA visibility. If inventory holds at 145, the stuffing pattern is confirmed and cash drains to $240M by year-end. Wait for Q2. No edge either side today.