PENG · Penguin Solutions, Inc. · Stock research

Current Thesis

Two things happened in the same ten days and the market has only priced one of them.

On July 7 Penguin Solutions printed the quarter that validates the entire agentic-AI infrastructure thesis: net sales $478.71M against $405.53M consensus, up 48% year over year, with non-GAAP EPS of $0.84 versus $0.54 expected. Management lifted the FY26 net-sales growth guide from ~12% to 22% ±2% and EPS from ~$2.15 to $2.60 ±$0.05 the third upward revision inside one fiscal year then pre-guided FY27 at roughly 30% growth in both sales and non-GAAP EPS. AI infrastructure revenue roughly doubled. The stock ran to an all-time high of $89.86.

Then on July 13 the company announced a $650M convertible offering, upsized to $750M and closed July 17. Zero coupon, due 2031, initial conversion price $116.70 a 50% premium to the July 14 close of $77.80. Proceeds fund capped calls, the cash leg of exchanges on ~$135.5M of the 2% 2029s and ~$160M of the 2% 2030s, and a credit-facility paydown. Shares fell 11.6% after hours on the announcement and kept going, closing $60.41 on July 17, down 8.4% that session and roughly 33% below the high set eleven days earlier.

That drawdown is convertible-arb mechanics. Arb desks buy the paper and short delta against it; on a name with 51.24M shares outstanding and a beta of 2.83, that hedging flow overwhelms the tape. Short interest sits at 8.75M shares, 17.08% of shares outstanding, 2.69 days to cover. What the selloff is not is a fundamental revision nothing in the July 7 numbers or the FY27 framework was retracted.

The setup is a company whose growth rate doubled trading 33% below where it traded before the growth rate doubled, sitting precisely on its 50-day moving average at $60.77.

Bullish and bearish views on Penguin Solutions, Inc.

The model's bull view on Penguin Solutions, Inc. (PENG), in brief: Q3 FY26 (July 7): revenue $478.71M vs $405.53M consensus, +48% YoY. The bear view: The chart is broken on any short horizon: $89.86 ATH on July 8-ish, $77.80 on July 14, $60.41 on July 17. Both cases follow in full.

Bull Case

• Q3 FY26 (July 7): revenue $478.71M vs $405.53M consensus, +48% YoY. EPS $0.84 vs $0.54. A 18% revenue beat on a systems integrator is an order-book event, not a rounding item.
• FY26 guide raised from ~12% to 22% ±2% growth and EPS from ~$2.15 to $2.60 ±$0.05. Three raises in one fiscal year is the signature of demand arriving faster than the plan.
• Preliminary FY27 view: ~30% growth in sales and non-GAAP EPS. On $2.60 FY26 EPS that frames roughly $3.35–3.40 for FY27, putting $60.41 near 18x forward cheap against a business compounding at 30% with a re-rating ecosystem position.
• NVIDIA AI Factory Specialized Partner designation, June 23, invitation-only, layered on the AMD/Shell Houston HPC reference build. Two validated ecosystem credentials replace what was a single-logo story in the spring.
• Analyst revisions ran the other way from the tape: Citizens to $85, Needham $60→$80, Stifel $66→$75 on July 8. Consensus target ~$74.29 now sits ~23% above the market the reverse of the June setup, when price had outrun every published target.
• The convert removes refinancing risk through 2031 at a 0.00% coupon with capped calls limiting effective dilution. Balance-sheet capacity for the AI Factory Platform buildout is now funded at zero cash cost.
• 17% short interest against a $3.1B market cap and a 2.69-day cover ratio is fuel once issuance-related hedging finishes.

Bear Case

• The chart is broken on any short horizon: $89.86 ATH on July 8-ish, $77.80 on July 14, $60.41 on July 17. Down 15.8% in a week, down a third from the high, with no confirmed higher low since.
• Convertible-arb short pressure has no fixed expiry. Delta hedging continues as long as desks hold the paper, and further stock weakness mechanically forces more shorting. The mechanism that drove the drawdown can extend it.
• Governance is thin at the wrong moment. Nate Olmstead departed as CFO July 8 for The Trade Desk; Aaron Johnson took the interim seat July 9 with a permanent search still open. A $750M capital-markets transaction executed under an interim finance chief invites scrutiny of timing the offering landed six days after the best print in company history, which reads as issuing into strength.
• The exchange leg dilutes: the 2% 2029s and 2030s are being retired partly in common stock, adding shares at prices well below $116.70.
• Gross margins remain in the low-20s. This is integration and assembly economics, and a demand air-pocket compresses both revenue and multiple simultaneously.
• Barclays stayed Equal-Weight and raised its target only to $40 on July 9 roughly a third below market. The dispersion between a $40 and an $85 target on the same print is unusually wide and says the sell-side does not agree on what this business is worth.
• Trailing P/E ~43 with a beta of 2.83 means the drawdown math is unforgiving if FY27 slips.

Setup & Price Structure

Price closed $60.41 on July 17 against a 50-day moving average of $60.77 and a 200-day at $31.16. The stock is at the 50-DMA, roughly 94% above the 200-DMA, and 33% below the $89.86 high.

The level that matters is the $58–62 shelf. That zone contains the June 7 low near $59.86 and the June 26 close near $62.28 the base the June–July advance launched from. Reclaiming and holding it converts the convert selloff into a retest of prior breakout support. Losing it on a daily close means the July 7 print marked the high and the mechanical selling has become distribution.

Above, the map is $70 (mid-July consolidation), then $77.80 (the July 14 conversion reference), then $89.86. The 52-week range remains $16.04 to $89.86, with the stock still up roughly 237% year to date.

There is no earnings binary inside the next month, which is the cleanest feature of this setup. Fiscal year ends late August; the Q4 print lands around October. That leaves several weeks for the arb flow to clear and for the tape to prove whether $58–62 holds, without a print in the middle of it.

Where this sits in the trap matrix: not peak retail sentiment (sentiment is negative post-offering), not stretched above moving averages (it is on the 50-DMA), no earnings inside three days. The live risk is a falling knife with a mechanical seller still working which argues for waiting on a reclaim rather than catching the down-move.

Catalyst Calendar (next 30 days)

• ~2026-07-24 to 2026-08-10 (est.) 10-Q filing for Q3 FY26. First look at segment-level AI Infrastructure vs Integrated Memory revenue splits and backlog language behind the headline beat.
• ~2026-08-11 (est.) Mid-August 13F season; watch for institutional additions or the SKT position showing any 13D/G amendment.
• Ongoing Form 4 filings. Insider buying into the $58–62 zone from anyone other than the interim CFO would be a meaningful signal on how management reads the offering-driven drawdown.
• ~2026-10 (est.) Q4 FY26 print and the first formal FY27 guide. The ~30% framework becomes a number management has to defend.
• No earnings catalyst inside the next 30 days. The July 7 binary has resolved.

Elapsed catalysts

• 2026-07-17 (complete) $750M convertible offering closed, full $100M option exercised. Issuance-related hedging typically compresses over the following 5–15 sessions; watch whether daily volume normalizes toward the pre-announcement baseline. _(passed 2d ago)_

What Would Change Our Mind

• A daily close below $58. That loses the June 7 higher-low shelf near $59.86 and the 50-DMA at ~$60.77 in one move, and reframes the convert selloff as the start of a downtrend rather than a hedging artifact.
• Sustained daily volume well above the pre-offering baseline two-plus weeks after the July 17 close, indicating the seller is not arb flow finishing but real distribution.
• Any softening of the ~30% FY27 growth framework an "approximately" turning into a range, or a segment-mix caveat in the 10-Q or at the October print.
• Loss of the NVIDIA specialization, or a competing integrator winning a named AI Factory build that Penguin was expected to take.
• A permanent CFO hire accompanied by guidance revision. New finance chiefs reset expectations; that is the standard risk when the seat has been interim through a capital-markets event.
• Conversely, the thesis strengthens on a weekly close back above $70 with the shelf intact that would signal the arb overhang has cleared and the fundamental re-rate is being priced again.

Correlation Notes

• Trades with the AI-infrastructure buildout complex rather than with semiconductor cycle names closest read-throughs are SMCI, DELL server orders, and VRT. Micron's July capex commentary ($250B US expansion referenced July 12) supports the memory-integration side of the model.
• Beta 2.83 versus the S&P. Index-level risk-off amplifies here by roughly 3x; a 2% SPX day is a 6% day in this name before any idiosyncratic news.
• Hedging with hyperscaler or GPU-maker puts is unreliable. Single-day correlation breaks down on float-driven and issuance-driven moves the July 13–17 drawdown happened while broad AI infrastructure was flat to higher.
• The convertible arb short base creates a correlation asymmetry worth understanding: on the downside the name trades heavier than peers because hedging accelerates; on a reclaim it trades lighter because the same desks buy back delta.
• Watch NVDA partner-ecosystem news flow as a leading indicator. Specialization-tier announcements have moved this name more reliably than its own product releases.