---
title: "SNDK: DC Boom, But LT Contracts Cap Price Hikes"
type: "Topics"
locale: "en"
url: "https://longbridge.com/en/dolphin/post/44653518.md"
description: "Beats no longer surprise: LT contracts floor the downside and cap the upside"
datetime: "2026-10-30T03:44:29.000Z"
locales:
  - [en](https://longbridge.com/en/dolphin/post/44653518.md)
  - [zh-CN](https://longbridge.com/zh-CN/dolphin/post/44653518.md)
  - [zh-HK](https://longbridge.com/zh-HK/dolphin/post/44653518.md)
author: "[Dolphin Research](https://longbridge.com/en/dolphin.md)"
generator: "portal-rs"
---

# SNDK: DC Boom, But LT Contracts Cap Price Hikes

$Sandisk(SNDK.US) reported FY27 Q1 (ended Oct 2, 2026) after market on Oct 29 ET: revenue, GPM and EPS all beat the top end of guidance, and next-quarter guidance also topped expectations. But this quarter’s beat leaned more on shipments than pricing, with less pricing traction than hoped. Details below:

**1\. Revenue topped:** revenue was $11.03bn, +23% QoQ, above the $10.3–10.8bn guide and ~consensus of $10.7bn (~3% higher).

The beat vs. guidance narrowed: last quarter it was ~9% above the high end, this quarter only ~2%. Buy-side expectations were already above the guide high (~$11.1bn), so a slight beat was largely in line.

**2\. Volume accelerated while LTAs capped price:** management said ~60% of the QoQ revenue lift came from shipments, ~40% from price.

By Dolphin’s estimate, bits rose ~14% QoQ, while ASP per GB rose only ~8%, below the ~10% consensus and less than half of NAND contract price gains (~+20%) in the same period. Management explained they shifted more supply to data center (+61% QoQ) while consumer declined 29% QoQ.

Dolphin views the supply shift to data center as a trade for long-term customers and share, which makes sense. However, the clear price underperformance vs. the industry likely ties to LTAs: about half of FY27 bits are under NBM contracts with monthly price locks and caps/floors. The trade-off of ‘certainty over elasticity’ under LTAs showed up clearly in the P&L this quarter.

**3\. GPM near ceiling; incremental margin slipping:** non-GAAP GPM was 85.1%, up only 50bps QoQ (85.0% on GAAP).

On estimates, incremental revenue converted to incremental GP at ~87%, down from ~97% last quarter. First, growth leaned more on volume; price hikes add little cost, while more units carry production costs. Second, unit cost per GB rose ~5% QoQ, likely due to a richer mix of higher-cost data center products.

This implies the profit driver is shifting from ‘pricing’ to ‘volume’. With bit growth capped in the mid-to-high teens by wafer plans, GPM has limited room to move higher.

**4\. Guidance: above expectations, but only ‘modest’ price increases**  
Next-quarter revenue mid-point is $12.4bn, ~2.5% above consensus; non-GAAP GPM mid-point at 85.0%, flat QoQ; EPS mid-point at $56, ~6% above, partly helped by buybacks reducing dilution (diluted shares down from 154mn to 151mn).

**More important than the numbers are management’s qualitative comments:**

① Next quarter both volume and price contribute, but price is only ‘modestly up through the year’, and the GPM guide only embeds ‘modest same-basis price increases’.

By Dolphin’s math, if ASP is +3%–5% QoQ, that implies bits +7%–9%, a further slowdown from ~8% ASP this quarter. TrendForce still expects Q4 contract prices to rise +15%–20%. The industry is still rallying, but the company’s pricing has shifted to ‘small steps’.

② The company gave no guide for the Mar quarter (FY3Q27) and urged conservative modeling from Q2 to Q3: Q3 is historically the weakest quarter, with past bit declines of up to 19% QoQ.

In 2H FY27, profit growth will rely mainly on bits, while bit growth is locked in the mid-to-high teens by wafer plans. Pricing elasticity has largely been ceded to LTAs.

**5\. No new NBM (LTA) signings:** NBM has 8 customers; the minimum contract revenue at floor prices is $93.9bn and customer financial guarantees total $16.5bn, unchanged from Investor Day; RPO was $88.9bn (vs. $91.1bn at Investor Day), down on deliveries this quarter. The two NBMs signed this quarter were pre-announced on the last call; no additions since.

**a. Coverage:** NBM covers ~1/2 of FY27 and ~2/3 of FY28 bits, consistent with Investor Day; management said coverage may ‘edge higher’, with the optimal ratio slightly above current.

**b. Terms:** WA duration >4 years, max 5; widespread price caps/floors; take-or-pay; monthly locks on volume, mix, and price.

**c. Guarantees:** guarantees equal ~17.6% of floor-price revenue (Micron ~21%), mostly off-balance sheet. This quarter, net inflows of on-balance prepayments and deposits were $465mn (vs. $1.938bn last quarter), which management said can be used for buybacks.

**d. Customers:** three are US hyperscalers; 95% counterparties are investment grade with aggregate operating cash flow of ~$130bn.

On the call, multiple customers were said to have underestimated NAND needs and are returning for second-round talks; one largest customer already signed a second tranche. The NBM pipeline exceeds the company’s supply through 2030, with demand signals extending beyond FY2030. But customer concentration bears watching: if AI funding tightens, changes in large customers’ credit would directly transmit to the company.

In Dolphin’s view, NBM turns NAND from a commodity repriced quarterly into a ‘contracted product’ with monthly locked volume and price, lifting the cycle floor on the way down. On the flip side, this quarter showed price caps give up upside elasticity in tight markets: industry contract prices rose ~20% in Jul–Sep, while ASP rose ~8%.

As LTAs roll out industry-wide, they reinforce the ‘this time is different’ narrative while also spurring capacity additions. That is the biggest overhang post-2028.

**6\. By segment: data center still surging**

**a.** Data center revenue was $4.80bn, +61% QoQ, mix rose from 33% to 43.5%, far above the $3.57bn the Street expected.

Management expects data center to exceed 55% of the NAND market in CY2027, with bit demand ~+40% YoY. By end-2030, KV cache could account for ~1/3 of AI data center flash demand.

Growth came from two lines: performance TLC eSSD (BiCS8 CBA architecture, used for KV cache and inference) and capacity QLC, with ‘Stargate’ fully qualified and ramping at customers. Hyperscaler TLC:QLC is ~7:3 and should broadly hold this fiscal year.

Customers are also discussing using SLC-mode NAND to replace part of DRAM (20–40x cost gap), and SLC vs. TLC/QLC has an approx. 3:1 effective capacity conversion. Real deployments would further tighten effective supply (still in architecture discussion).

**b.** Edge revenue was $5.83bn, +7% QoQ, slightly above the $5.53bn expectation.

**c.** Consumer revenue was $0.40bn, −29% QoQ, well below the $0.57bn expectation.

With total bits constrained by wafer plans, ‘allocating more to data center’ effectively means shifting supply away from Edge and consumer. At the same time, Edge and consumer are being ‘crowded out’: storage now accounts for 30%–50% of PC and smartphone BOM, so OEMs are cutting shipments and compressing device capacities.

Edge resilience comes from high-end and mainstream models, while low-end is shrinking. Auto and IoT are bright spots, and management believes auto could become one of the largest markets within Edge.

Management expects PC shipments to decline this year and fall again by low single digits next year, but capacity per device should rise (mix shift up). EB consumption in PCs and smartphones will still grow.

**7\. Extreme operating leverage:** revenue rose 23% QoQ while non-GAAP opex grew only 6% ($510mn, below guide); OPM reached 80.4%, already above the 75% FY28–FY30 long-term target from Investor Day.

Non-GAAP net income was $7.62bn, ~6% above consensus, driven mainly by revenue and GPM beats.

GAAP net income was $8.10bn, with the delta mainly from mark-to-market gains on the Nanya stake ($513mn), which fluctuates with Nanya’s share price and should not be viewed as recurring.

**8\. Capex, cash flow, and buybacks:** operating cash flow was $6.36bn (incl. $465mn of NBM prepayments and deposits), −11% QoQ, mainly due to A/R up $1.88bn to $6.59bn. Management said this reflects customer mix shifts, not changes in NBM payment terms.

After $54mn in PP&E, FCF was $6.31bn; excluding NBM-related items and Flash Ventures settlements, Adj. FCF was $5.70bn, with an FCF margin of 51.6%, in line with the 50% Investor Day target.

The company repurchased $5.0bn this quarter, using ~90% of Adj. FCF, and reiterated a 100% return of excess cash. At a market cap of ~$195bn, the annualized buyback yield is ~10%.

The remaining $10.5bn authorization covers roughly two quarters; renewal will determine whether buybacks can stay at this pace. On capex, the company is driving bit growth via BiCS node migrations (bits per wafer up ~27% annualized), not adding new wafer starts, in sharp contrast to Micron’s aggressive cleanroom buildout.

Total capex was $380mn this quarter (3.4% of revenue), and the company expects FY27 to rise to a mid-single-digit % of revenue. On $48–50bn of revenue, that implies roughly $2–3bn.

**Supply side:** at end-Aug, Kioxia and the company announced a Japan JV expansion of ~JPY 5tn (~$31bn), construction 2026–2032, with scaled output in late-2027 to 2028. This overlaps with Dolphin’s view that price increases persist until 2H27 and 2028 is the biggest debate.

This implies higher capex and D&A over the next few years. But the company will not accelerate BiCS8→9→10 migrations for short-term bit growth, maintaining mid-to-high teens bit growth plans.

9\. For call notes, see ‘[**Data Center is on Fire, but LTAs are Tapping the Brakes on Pricing!**](https://longbridge.com/zh-CN/dolphin/post/44653518)’

**Dolphin’s take:**

At its Aug Investor Day, the company sketched an ‘AI-era cash machine’: mid-to-high teens revenue CAGR in FY28–FY30, 80% GPM, 75% OPM, 50% Adj. FCF margin, and 100% return of excess cash. This is the first full quarter post-NBM rollout, and the market wants to see if pricing, profit, and cash flow track the script.

Results beat again, but quarterly beats are no longer the key stock driver. What matters is not ‘how much above,’ but ‘how’:

a. ~60% of the revenue delta came from volume and ~40% from price; ASP rose ~8% this quarter and only ‘modestly’ next quarter, while industry contract prices are still rising double-digits. LTAs are flattening the profit curve.

b. 2H FY27 sequential momentum looks softer; the Mar quarter seasonally weak, and management proactively guided conservative modeling.

c. No new NBM signings; the $93.9bn minimum contract revenue and $16.5bn guarantees are unchanged. A ‘strong pipeline’ has yet to translate into new numbers.

The real debate is post-2028: management’s model assumes flat prices, mid-to-high teens bit growth, and normalized 80% GPM. The market doubts the cycle can be fully smoothed.

**This quarter’s results are fine; the issue is capped pricing upside and uncertainty on the post-2028 cycle.**

In Dolphin’s Oct deep-dive ‘[AI Can’t Change the DNA: How Long Can the Rally Last?](https://longbridge.cn/en/dolphin/post/44203122?app_id=longbridge&utm_source=longbridge_app_share&share_track_id=386520b2-5153-4a06-875d-57dad8bbf0a0&invite-code=4NOXYT&locale=zh-CN&community_badge=1&profile_following_followers_activities=1)**’**, the base case was: this upcycle likely runs to end-2027; supply loosens in 2028 after YMTC Fab 3 ramps; in 2029, Samsung P5, Kioxia/the company’s northern expansion, and SK Hynix M17 ramp together, driving a visible price pullback. Under the neutral case, net income peaks in FY28 (~$41.6bn), then falls to ~$21.6bn by FY30.

**New intra-quarter data reinforces this path on both price and supply:**

**Pricing:** TrendForce shifted the NAND contract price peak from Q4 2026 to Q4 2027, implying a longer upcycle but a gentler slope. Q4 is still +15%–20%, while early-Oct spot saw a single-week pullback of ~2.9%; the company’s own pricing slope is even flatter.

**Supply:** Kioxia/the company JV expansion, Micron upping FY27 capex to >$5.0bn, and YMTC Wuhan Fab 3 connected tools in Sep all point to scaled output around late-2027 to 2028.

‘Strong prints now, softer sequential momentum, and looser forward supply’ explains why the multiple struggles to rerate. The market is paying up for FY27 profits but discounting FY29–FY30 mean reversion; and now another layer: LTAs cap downside but also cap upside.

**Post-print, Dolphin will watch four things:**

**1) NBM round-two pricing:** while no new deals this quarter, multiple customers are discussing second rounds. Whether new/renewed contract floors move up with spot will determine how much elasticity can be clawed back.

**2) Supply discipline:** FY27 capex intensity is tracking from this quarter’s 3.4% toward mid-single digits. Watch the funding cadence for the Kioxia JV expansion, BiCS9/10 ramp timing, and YMTC/Samsung/Micron expansion moves.

**3) 2028 demand visibility:** can FY28 NBM coverage rise from ~2/3, and can new use cases such as KV cache, QLC data lakes, and SLC substituting DRAM translate into orders?

**4) HBF progress:** the first HBF memory die has taped out and entered manufacturing; sampling to customers is planned for 2027. Next steps are taping out the controller base die and assembling with the memory die for customer qual. HBF and 3D Matrix Memory were not in the Investor Day model, representing upside options, but management is limiting customer exposure and not discussing TAM yet, so orders are still distant.

**Stock:** from the Sep 22 peak of $1,887, shares fell to $1,330 pre-earnings, a ~30% pullback (post-market ~$1,280), while Micron fell only ~15%.

The sector faces three pressures:

a. ‘Peak pricing’ worries: spot pullback and slower Q4 contract price gains drove a mid-Oct selloff in KR/JP memory names. b. AI funding concerns: on Oct 22, Jensen Huang warned of ‘capital scarcity’ for AI expansion, and the stock fell ~8.5% that day.

c. Forward supply concerns: major players’ new capacity clusters around late-2027 to 2028. On the day of earnings, memory stocks fell again post-Samsung’s print.

The stock fell more because it is a pure NAND play and most sensitive to price. Concerns that LTAs cap upside also make the market reluctant to pay for its pricing elasticity.

**Valuation:** at the pre-print market cap of ~$194.7bn (share price $1,330), the stock trades at ~6.2x FY27 (to Jun 2027) and ~5.1x FY28. On Dolphin’s FY22–FY30 cycle-average net income (~$15.9bn), that’s ~12x.

Under a neutral scenario (15x on cycle-average), implied market cap is ~$239bn (about $1,632 per share), or ~7.7x FY27 and ~6.2x FY28. If there is evidence of higher round-two NBM floors or materially higher FY28 coverage, the multiple could rise to 16x, implying ~$254.4bn (~$1,737 per share), ~8.2x FY27 and ~6.6x FY28, or ~23%–31% upside.

Bear case (no LTA premium, 8–10x cycle-average) implies ~$127–159bn (~$870–1,090 per share), ~4.1–5.1x FY27 and ~3.3–4.2x FY28, or ~18%–35% downside from current.

Currently, the share price is nearly 20% below the neutral entry line of $1,632 from the Oct deep-dive.

This print confirms high FY27 profitability and strong cash returns ($5bn buyback in a single quarter; ~52% Adj. FCF margin), while exposing the LTA trade-off: upside pricing elasticity is capped, and 2H growth must come from bits, not price.

While the report does not answer whether prices ‘break’ post-2028, the current ~12x cycle PE already discounts that risk. In the neutral case, the current level is within an accumulation range, but near-term catalysts are lacking; better to build positions in tranches between bearish and neutral levels, not chase. A better add point may await clarity on round-two NBM floors and how soft the Mar quarter gets.

\<End here>

Risk disclosure and statements for this article: [Dolphin Research Disclaimer and General Disclosure](https://support.longbridge.global/topics/misc/dolphin-disclaimer)

Dolphin Research storage coverage recap:

**Recent AI storage analyses:**

[‘Why Maligned HDDs Became AI’s Data Endpoint?’](https://longbridge.com/en/dolphin/post/43941280?channel=OWNN00110)

DRAM industry:

[‘From Compute to Memory: Why Did HBM Break Out?’](https://longbridge.com/en/dolphin/post/43941280?channel=OWNN00110)

[‘From HBM Caps to NAND Volume: Who Holds Whom, Nvidia vs. Memory?’](https://longbridge.com/en/dolphin/post/44108509?channel=OWNN00110)

Single-name notes:

Company (Part I): [‘AI Inference Boom: Can the ‘Unlucky’ Name Reborn Like a Phoenix?’](https://longbridge.com/en/dolphin/post/43891402?channel=OWNN00110)

Company (Part II): [‘NAND is ‘Inherently Prolific’: How to Defend 80% GPM?’](https://longbridge.com/en/dolphin/post/43966452?channel=OWNN00110)

Company (Part III): ‘[From Sidekick to Star: How AI Inference is ‘Rewriting’ NAND’s Fate?](https://longbridge.cn/en/dolphin/post/44124083?app_id=longbridge&utm_source=longbridge_app_share&share_track_id=e17f4a6c-b5d9-407e-9241-0f8012b8106d&invite-code=4NOXYT&locale=zh-CN&community_badge=1&profile_following_followers_activities=1&channel=OWSN00110)’

Company (Part IV): ‘[AI Can’t Change the ‘DNA’: How Long Can the Rally Last?](https://longbridge.cn/en/dolphin/post/44203122?app_id=longbridge&utm_source=longbridge_app_share&share_track_id=386520b2-5153-4a06-875d-57dad8bbf0a0&invite-code=4NOXYT&locale=zh-CN&community_badge=1&profile_following_followers_activities=1)**’**

**II. Dolphin AI DC Interconnect series:**

**Network architectures**

Nvidia networking: ‘[AI-era DC Interconnect: Team-of-Chips Above a Single Die — Is There a ‘China’ Opportunity?](https://longbridge.cn/en/dolphin/post/43798848?channel=OWNN00110)’

AMD vs. Nvidia in DC networking: ‘[AMD Arm-Wrestles Nvidia: Is Helios Up to It?](https://longbridge.cn/en/dolphin/post/43906382?channel=OWSN00110)’

Google networking: ‘[Challenging Nvidia’s Hegemony: What Makes Google’s ‘Photonic Network’ Work?](https://longbridge.cn/en/dolphin/post/43994983?channel=OWNN00110)’

**Single names**

Lumentum (Part I): [‘From Optical Veteran to ‘All-Purpose Water Vendor’: What Powers Lumentum?’](https://longbridge.cn/en/dolphin/post/44059432?channel=OWNN00110)

Lumentum (Part II): [‘Lumentum: Capacity Under Strain — What Secret Sauce Remains for the ‘All-Purpose Water Vendor’?](https://longbridge.cn/en/dolphin/post/44191355?channel=OWNN00110)’

**III. Dolphin AI XPU series**

Agents and CPUs: [‘With ‘Muse’ Going Viral, Is This CPU’s ChatGPT Moment?’](https://longbridge.cn/en/dolphin/post/44095954?channel=OWNN00110)

…More at the [Longbridge Dolphin Research website](https://longbridge.cn/en/profiles/2106669?channel=OWNN00030).

### Related Stocks

- [MU.US](https://longbridge.com/en/quote/MU.US.md)
- [SNDK.US](https://longbridge.com/en/quote/SNDK.US.md)
- [MUU.US](https://longbridge.com/en/quote/MUU.US.md)
- [MUD.US](https://longbridge.com/en/quote/MUD.US.md)
- [MULL.US](https://longbridge.com/en/quote/MULL.US.md)
- [MUZ.US](https://longbridge.com/en/quote/MUZ.US.md)
- [SNDU.US](https://longbridge.com/en/quote/SNDU.US.md)
- [SNXX.US](https://longbridge.com/en/quote/SNXX.US.md)
- [SNDG.US](https://longbridge.com/en/quote/SNDG.US.md)
- [SNDQ.US](https://longbridge.com/en/quote/SNDQ.US.md)

## Comments (4)

- **MaxSureWin · 2026-11-01T01:27:56.000Z**: hush ahhh
- **yanny · 2026-10-31T22:40:49.000Z**: Good
- **NewUser_asNDXg · 2026-10-31T01:00:19.000Z**: Wow
- **CrazyOne · 2026-10-31T00:22:15.000Z**: lets go


---
> **Disclaimer: This article is for reference only and does not constitute any investment advice.**