SANM · Sanmina Corporation · Stock research

Current Thesis

The legacy-EMS AI re-rate is now in its correction phase with the binary sitting directly ahead. After FQ2 FY2026 (2026-04-27) delivered revenue $4.01B (+102% YoY, +22% vs $3.29B consensus) and non-GAAP EPS $3.16 vs $2.40, the stock did not stall near $190 it ran to a $282.72 closing all-time high on 2026-06-03 on S&P MidCap 400 inclusion and Bank of America's 2026-06-05 target raise to $280, then rolled over ~16% to a last documented ~$236 (2026-06-26, a 6.7% single-day drop) on higher Treasury yields and softer datacenter-demand reads. The narrative Sanmina as AMD's named preferred US NPI partner for cloud rack and cluster-scale AI via the ~$3B ZT Systems acquisition (closed 2025-10-27) is intact but no longer accelerating. The decisive change since last review is timing: the FQ3 print (est. ~2026-07-27, quarter ended 2026-06-27) is now roughly nine days out, converting a post-parabolic correction into a stand-aside-into-binary. That print tests whether FQ2's surge was H2 order pull-forward or a durable run-rate.

Bullish and bearish views on Sanmina Corporation

The model's bull view on Sanmina Corporation (SANM), in brief: FQ2 FY2026 (2026-04-27): revenue $4.01B (+102% YoY, +22% vs $3.29B consensus), non-GAAP EPS $3.16 vs $2.40, free cash flow $342M, non-GAAP pretax ROIC 34.7% vs 23.0% a year prior. The bear view: Pull-forward is the entire debate: management attributed FQ2 strength to "accelerated compute shipments previously expected in the second half" moving into Q2. Both cases follow in full.

Bull Case

• FQ2 FY2026 (2026-04-27): revenue $4.01B (+102% YoY, +22% vs $3.29B consensus), non-GAAP EPS $3.16 vs $2.40, free cash flow $342M, non-GAAP pretax ROIC 34.7% vs 23.0% a year prior.
• FY2026 guide raised (2026-04-27) to adj EPS $10.75–11.35 (vs $10.02 est) on sales $13.7–14.3B; stated FY2027 revenue target $16B+ against FY2025's $8.1B base.
• ZT Systems adds New Jersey, Texas and Netherlands datacenter-manufacturing capacity plus high-power and liquid-cooling capability for next-gen AI racks, anchored by AMD as named preferred NPI partner.
• Structural passive bid: added to the S&P MidCap 400 in the June 2026 quarterly rebalance (up from SmallCap 600), a permanent flow change.
• Analyst catch-up unfinished BofA raised to $280 from $234 (2026-06-05) while the 13-analyst mean still sits ~$224–230, below the last ~$236 tape, leaving upgrade room if FQ3 confirms run-rate.
• Drawdown history is favorable: of 15 sharp (≥20% in a month) declines since 2010, the stock was higher 12 months later in 11, median forward return +27%.

Bear Case

• Pull-forward is the entire debate: management attributed FQ2 strength to "accelerated compute shipments previously expected in the second half" moving into Q2. FQ3 guide is $3.2–3.5B revenue (vs $3.522B consensus) a sequential step DOWN from $4.01B so the print can disappoint even against a lowered bar.
• Post-parabolic structure: ~16% off the $282.72 ATH (2026-06-03) to ~$236 (2026-06-26), including a 6.7% single-session drop. The easy momentum leg has already been paid out.
• AMD/ZT single-customer concentration JPM's Neutral explicitly flags "AMD rack concentration risk." If AMD second-sources rack/NPI manufacturing or loses accelerator share, the growth engine stalls.
• Thin margins: FY2026 operating-margin guide of 6.3–6.6% means revenue near-doubling at low-single-digit margins, which argues for a lower multiple than the ~50x trailing P/E paid at the highs.
• The 13-analyst mean PT (~$224–230) trades below the last tape; three weeks ago that gap read as bullish sell-side lag, into a binary it reads as price ahead of fundamentals. Sell-side floor targets (JPM $145, Susquehanna $135) sit far under the tape.

Setup & Price Structure

• Last documented ~$236 (2026-06-26), down ~16% from the $282.72 closing all-time high (2026-06-03); 52-week range roughly $75–283. Price requires re-validation against the current tape before acting.
• The June rally was a blow-off, not a base a parabolic advance from the sub-$190 April post-earnings shelf to $282.72 in five weeks, now unwinding. There is no confirmed higher-low base yet.
• Preferred structure for any re-engagement is a higher low that reclaims the 20-EMA, or post-print run-rate confirmation not a falling-knife catch mid-correction.
• Valuation anchors: NTM EV/EBITDA ~11x vs Jabil 14x / Celestica 29x; fwd P/E ~18.6; net debt/EBITDA 1.48x; $600M buyback open-ended.
• With the print ~9 days out, this sits in the pre-earnings blackout window where fresh entries carry binary gap risk; the disciplined read is to stand aside until the print clears.

Catalyst Calendar (next 30 days)

• ~2026-07-27 (est.) FQ3 FY2026 earnings, quarter ended 2026-06-27. The binary: revenue guide $3.2–3.5B (vs $3.522B consensus) and adj EPS $2.55–2.85, plus whether management raises or holds the FY2026 $13.7–14.3B / $10.75–11.35 framework. Watch language on H2 order timing and AMD/ZT run-rate.
• ~2026-07-27 FQ3 earnings call, same session: read on rack/cluster backlog, liquid-cooling capacity utilization, and customer concentration.
• Into/after 2026-07-27 potential analyst upgrade or PT-revision cluster as the mean (~$224–230) reconciles to the print; an upgrade wave is the un-played catch-up catalyst if the run-rate holds.

What Would Change Our Mind

• Bullish re-engagement: an FQ3 print that holds or beats the $3.2–3.5B guide with FY2026 raised and management framing datacenter demand as run-rate rather than pull-forward, followed by a higher-low base that reclaims the 20-EMA that reopens the trend leg.
• Bearish confirmation: revenue printing under the $3.2B guide floor, or explicit commentary that FQ2's +102% YoY was H2 shipments pulled forward, would confirm the pull-forward bear case and validate the correction as the start of a de-rate.
• The invalidation level below is the price scoreboard that separates a routine post-parabolic pullback from giving back the entire June re-rate.

Correlation Notes

• Trades as a high-beta AMD / AI-datacenter derivative single-customer concentration means SANM moves with AMD accelerator sentiment and hyperscaler capex reads more than with its own diversified EMS book.
• Correlated with EMS peers Jabil (JBL), Celestica (CLS) and Flex (FLEX); relative multiple (~11x NTM EV/EBITDA) sits well below CLS (~29x), so peer re-rating in the group can pull the whole cohort.
• Rate-sensitive: the 2026-06-26 6.7% drop coincided with higher Treasury yields, so a renewed yield spike pressures the long-duration AI-growth cohort including this name.
• S&P MidCap 400 membership ties a portion of the float to index/passive flows and rebalance mechanics rather than fundamentals.