Inve US ▾
PEPENG

Penguin Solutions (PENG) Q4 FY26 earnings call: guidance and Q&A

Earnings call· Aug 2026 (Q4 FY26)· Information Technology· Oct 6, 2026

Source: earnings call transcript

Executive summary

Penguin Solutions Q4 FY26 — Record quarter as AI infrastructure and memory drive 68% sales growth; FY27 outlook raised

Penguin has pivoted from a flat first half to an AI-factory and data-center memory story, with growth broadening from memory into neo-cloud AI infrastructure and operating leverage outrunning revenue. The guide raise is backed by record but unquantified backlog; the debate shifts to cash conversion, mix-driven margin pressure and cyclicality of memory pricing.

Highlights

5
  • Q4 net sales $567M, up 68% YoY and 18% QoQ, with company records in net sales, gross profit, operating income, operating margin, net income and adjusted EBITDA

  • Non-GAAP diluted EPS $1.00, up 133% YoY; FY26 EPS $2.87, up 51% and $0.22 above the high end of the July outlook

  • Integrated memory record $341M, up 158% YoY and 24% QoQ; non-hyperscale AI infrastructure grew 99% YoY within advanced computing

  • Operating leverage: FY26 opex up 4% to $267M vs net sales up 26%; Q4 non-GAAP operating margin 15.8%, up 4.2pp YoY

  • FY27 net sales outlook raised to ~+40% at midpoint (~$2.43B) from a ~30% preliminary view; EPS guide ~$4.45

Concerns

5
  • Operating cash flow was -$163M in Q4 and -$152M for FY26 (vs +$113M in FY25) as working capital absorbed cash for memory/AI growth

  • Inventory rose to $749M from $255M a year ago and inventory days rose to 48 from 42 last quarter; accounts receivable $796M vs $308M

  • FY26 non-GAAP gross margin 29.3%, down 1.7pp on AI infrastructure hardware mix; FY27 gross margin guided ~28% ±2pp, with management flagging possible downward pressure in AI infrastructure

  • Services net sales down 11% YoY to $57M on lower hyperscale volumes; advanced computing FY26 sales down 14% to $559M after Penguin Edge wind-down

  • Wide FY27 ranges (±10pp sales, ±$0.70 EPS) reflect deployment timing, component lead times and memory cost/pricing variability; convertible issuance raised share count

Guidance & targets

CategoryTargetConfidence
FY27 net sales growth
approximately 40% at midpoint, plus or minus 10 percentage points (~$700M increase to ~$2.43B at midpoint)
high materiality
High
FY27 advanced computing net sales growth
approximately 40% at midpoint, plus or minus 10 percentage points
high materiality
High
FY27 integrated memory net sales growth
approximately 50% at midpoint, plus or minus 10 percentage points
high materiality
High
FY27 optimized LED net sales
relatively flat year-over-year
low materiality
Medium
FY27 non-GAAP gross margin
approximately 28%, plus or minus 2 percentage points
high materiality
Medium
FY27 non-GAAP operating expenses
approximately $275M, plus or minus $10M
high materiality
High
FY27 non-GAAP diluted EPS
approximately $4.45, plus or minus $0.70 (~55% growth at midpoint)
high materiality
High
FY27 diluted share count
approximately 63 million shares
medium materiality
High
FY27 non-GAAP effective tax rate
approximately 20%
medium materiality
Medium
FY27 operating margin and EPS vs sales growth
operating margin expansion; EPS growth outpacing net sales growth
medium materiality
Medium
FY27 capital expenditures
asset-light; capex well under 1% of net sales (continuation of FY26 pattern)
low materiality
Medium
FY27 working capital efficiency
improving working capital efficiency is an important priority
low materiality
Low

Segment performance

SegmentRevenueYoYQoQMargin
Advanced Computing
FY decline reflects wind-down of Penguin Edge and lower hyperscale hardware sales; AI factory platform growth more than offset traditional runoff.
Non-hyperscale AI infrastructure net sales growth: +99% YoY in Q4Non-hyperscale AI infrastructure share of segment net sales: 66% (vs 60% prior-year quarter)Non-hyperscale AI infrastructure FY26 growth: +72%New AI infrastructure customers: 6 in Q4 (4 neo clouds, 1 quant trading firm, 1 enterprise); 17 in FY26Existing AI infrastructure customers expanded: 12 in FY26
$154M in Q4 (27% of net sales); FY26 $559M+11% in Q4; FY26 -14%+12%Segment margin not disclosed; Q4 advanced computing margins lower sequentially
Integrated Memory
Both volume and pricing increased; data-center focused, with CXL memory expansion pipeline and bookings strengthening.
New memory customer logos: 2 in Q4; 15 in FY26Existing memory customers expanded: 29 in FY26
$341M in Q4 (60% of net sales); FY26 $924M (53% of net sales vs 34% in FY25)+158% in Q4; FY26 +99%+24%Segment margin not disclosed; profitability improved and stronger than expected
Optimized LED
Managed with discipline; not a strategic investment priority.
$72M in Q4 (13% of net sales); FY26 $249M+7% in Q4; FY26 -3%—Improved profitability, not quantified

Orderbook & backlog

Company backlog Record; not quantified End of Q4 FY26

Record; bookings grew faster than net sales in AI infrastructure and integrated memory in Q4

Management declined to disclose how much of FY27 advanced computing outlook is covered by backlog; described as strong across memory and AI infrastructure

Integrated memory backlog Extends at least 4 quarters; higher than revenue recognized in the quarter End of Q4 FY26

Continues to strengthen

Supply-constrained portion of backlog previously cited as closer to even split between price and volume; no conversion % given

Product announcements

ProductTypeDetails
Cluster Ware AI agentic capabilities / AI factory operations agentlaunch
CXL memory expansion cards and Memory AI KV cache solutionupdate
Photonic high-bandwidth memory appliances and CXL AIC for DDR4roadmap

Deals & partnerships

Neo cloud backed by a leading South Korean technology company Design, build, deploy and manage NVIDIA GB-300 NVL-72 based platform Multiyear (not specified)

Counterparty unnamed.

Publicly traded neo cloud with >$3B in signed multiyear AI infrastructure contracts AI infrastructure deployment and 24/7 operation services supported by Cluster Ware AI Not disclosed (counterparty has >$3B signed contracts) Multiyear

Counterparty unnamed.

Lektra Full stack AI factory platform for distributed AI micro data centers powered by carbon-free energy

Validated reference designs, NVIDIA GPUs, services with Cluster Ware AI and support. Name as transcribed.

Neo cloud provider with $10B contracted compute from a leading AI lab Deploy and operate AI factory in Norway Not disclosed ($10B is the customer's contracted compute) Multiyear

Transcript reads 'a 36,000 AI factory in Norway' (likely 36,000 GPUs; unit garbled).

NVIDIA NVIDIA AI factory specialized partner (invitation-only designation) More than a decade of relationship

Cited as strengthening partnership and credibility.

Leading memory supplier (unnamed) New supply arrangement to support data center AI infrastructure portfolio

Also established relationship with another AI hardware supplier to expand supply availability.

Convertible note investors / 2029 and 2030 noteholders $750M 0% convertible notes due 2031, with capped call, and exchange of ~$296M of 2029/2030 notes $750M; ~$296M exchanged for cash and common stock Due 2031

Proceeds also used to repay $100M outstanding under credit agreement.

Risks & headwinds

Working capital absorption and negative operating cash flow FY27

Operating cash flow -$163M in Q4 and -$152M in FY26; AR $796M, inventory $749M, AP $817M; cash conversion cycle 42 days vs 33 last quarter

Mitigation:Improving working capital efficiency a stated FY27 priority; convertible raise provided liquidity ($647M cash)

Gross margin pressure from AI infrastructure hardware mix and memory costs FY27

FY26 gross margin 29.3%, down 1.7pp; FY27 guided ~28% ±2pp

Mitigation:High attach rate of Cluster Ware software and services; scale-driven gross profit dollar growth

Memory market cyclicality and pricing/cost assumptions FY27

Memory is 60% of Q4 sales; FY27 memory growth guided ~50% ±10pp

Mitigation:Data-center focus, negotiated pricing, asset-light model, backlog of at least 4 quarters

Component supply, lead times and deployment timing FY27 quarterly cadence

Certain components have extended lead times; FY27 sales range ±10pp, EPS ±$0.70

Mitigation:New memory and AI hardware supply arrangements

Hyperscale and services softness FY27

Services net sales -11% YoY to $57M; advanced computing FY26 -14%; lower hyperscale sales expected in FY27

Mitigation:Growth in multiyear AI factory managed services and non-hyperscale AI infrastructure

Inventory build ahead of backlog Near term

Inventory $749M vs $255M a year ago; inventory days 48 vs 42 last quarter

Mitigation:Framed as positioning against record backlog and expected customer deployments

Dilution from convertible financing Through 2031

Conversion price ~$116.70; FY27 diluted share count ~63M

Mitigation:Capped call up to ~$175.05 per share

What to watch in Q1 FY27

Operating cash flow and working capital

H1 FY27
Current Q4 operating cash flow -$163M; cash conversion cycle 42 days; inventory $749M
Target Improving cash conversion and reduced cash burn

Why it matters

Growth is being funded by working capital; sustained outflows would pressure liquidity despite the convertible raise

We believe our capital structure and liquidity provide flexibility to fund this growth while improving working capital efficiency remains an important priority for fiscal 2027.

Find out whether PENG delivers this. Inve re-checks your case every morning.

Q&A highlights

What changed in 90 days to lift advanced computing from mid-teens to ~40%, and how do neo-cloud unit economics differ from enterprise?

Bookings exceeded expectations with a strong pipeline, driving AI infrastructure and company-level guide increases. Neo-cloud unit economics are similar to enterprise on-prem factories; differentiation is end-to-end build/operate with Cluster Ware AI for SLAs, over 3-5 year engagements.

“But all in all, similar unit economics, further expansion and pretty unique product market fit with new cloud customers.”

asked by Katherine Murphy (Goldman Sachs) · answered by Kash Shaikh

3 min read 6 chapters

Detailed narrative

Record Q4 and the second-half inflection

After roughly flat first-half sales, growth accelerated to 48% in Q3 and 68% in Q4, giving second-half growth of 58%; FY26 net sales were about $1.73B, up about 26%. Management attributes the acceleration to the AI factory platform strategy launched mid-year, a sharper data-center focus and go-to-market effort aimed at neo clouds and enterprise.

Q4 net sales and EPS exceeded the outlook, and FY26 operating income reached a record $241M, up 44%. Bookings grew faster than net sales in AI infrastructure and memory and the company ended the year with record backlog.

AI infrastructure and neo-cloud momentum

Advanced computing grew 11% YoY in Q4, but non-hyperscale AI infrastructure grew 99% and rose to 66% of the segment, offsetting traditional business runoff. Six new AI infrastructure customers were added in Q4, including four neo clouds, a quant trading firm and an enterprise.

Named wins include a Korean-tech-backed neo cloud on GB-300 NVL72, a public neo cloud with over $3B of signed contracts, Lektra, and a neo cloud with $10B contracted compute for a Norway deployment. Management says neo-cloud unit economics are similar to enterprise on-prem factories, and that most projects have already started.

Integrated memory: price and volume

Memory hit a record $341M, 60% of company sales, with both volume and pricing rising on data-center and agentic-AI inference demand. FY26 memory sales were $924M, up 99%, lifting memory to 53% of sales from 34%. Management says backlog is well above revenue recognized in the quarter and extends at least four quarters.

CXL memory expansion cards, KV-cache solutions and a new supply arrangement with a leading memory supplier support continued growth, though management concedes memory is cyclical.

Margins and operating leverage

Gross margin improved 70bps sequentially to 28.8% on memory profitability, partly offset by lower advanced computing margins; FY gross margin fell 1.7pp to 29.3% on AI infrastructure hardware mix. Opex grew only 4% for the year, a 3.4pp decline as a share of sales, which management ties to India-based engineering (95%+), Taiwan memory engineering, Malaysia manufacturing and internal AI productivity tools.

Segment margins are not disclosed, though management said AI infrastructure hardware wins may pressure margins, offset by high cluster software and services attach.

Working capital, balance sheet and financing

Receivables, inventory and payables all scaled with gross billings and higher memory costs, producing operating cash outflow in Q4 and FY26. The company closed a $750M 0% convertible due 2031 with capped call, exchanged about $296M of 2029/2030 converts for cash and shares, and repaid the $100M credit facility, ending with $647M cash.

Capex remained asset-light at $12M for the year. The new convert increased the share count, which is why sequential EPS growth trailed operating income growth.

FY27 outlook and leadership

FY27 guidance was raised well above the ~30% preliminary view, with advanced computing and memory both guided higher and LED flat; EPS is expected to grow faster than sales. Management says the earlier view was directional and that bookings and conversion visibility now support a formal outlook.

Stephen Comin (transcribed as '[ Stephen Coming ]') joins as CFO; Aaron Johnson, interim CFO since July, returns to Vice President of Finance and Accounting. Management cited extended component lead times and deployment timing as quarterly cadence risks.

Summary of the company's earnings call; see how call summaries are prepared. Not investment advice.