Subscribe to The Value Anchor Topic / The Trend Catcher Topic —unlock the full historical archive and never miss a weekly pick again.

A single Investor Day has forced Wall Street to re-underwrite Sandisk Corporation(SNDK.US). After the stock had come under pressure post-earnings, a barrage of disclosures at the event flipped sentiment almost overnight, driving shares up nearly 14%.

What thrilled investors wasn't any one new product. Management attempted to prove that SanDisk is no longer just a cyclical company riding rising NAND prices, but a business that — through long-term contracts, technology upgrades and aggressive buybacks — can become a "compounding machine" throwing off durable cash flow. Four catalysts stood out.

1. An ~80% gross margin target

SanDisk laid out a multi-year financial framework spanning fiscal 2028–2030, headlined by a roughly 80% gross margin goal. NAND has always been a textbook cyclical industry — demand rises, prices climb, capacity floods in, then oversupply crushes margins. The key detail here: 80% is an average target across FY2028–2030, signaling that management believes the margin uplift is structural rather than a fleeting price spike.

Fiscal Q4 2026 already hinted at that earnings power: roughly $36B annualized revenue, an 84.6% non-GAAP gross margin, and about $20B in annualized adjusted free cash flow. Bank of America estimates that, if the framework holds, SanDisk could generate around $100B in cumulative cash flow over FY2028–2030. The debate has shifted from "can SanDisk make money" to "how long can margins this high last?"

2. NBM long-term contracts — an attempt to rewrite the cycle

SanDisk's confidence rests largely on its NBM (multi-year) contracts, which upgrade customer relationships from quarterly price haggling to deep, multi-year partnerships. These deals lock in product mix, volume, and pricing mechanisms — fixed prices in some years, floors and ceilings further out — backed by customer financial guarantees.

So far, SanDisk has signed NBMs with 8 data-center-focused customers, including 3 US hyperscalers, with 2 already expanding their agreements. The headline numbers:

• Weighted-average term of over 4 years (up to 5)
• ~half of FY2027 and ~two-thirds of FY2028 shipment bits already committed
• $93.9B total contract value at the price floor
• $91.1B in remaining performance obligations
• $16.5B in customer financial guarantees

Because $93.9B is calculated at the floor, realized revenue could run higher. NBM can't erase the cycle, but it dampens earnings sensitivity to quarterly pricing — the core pitch being that the market should value SanDisk on long-term orders and predictable cash flow, not spot prices.

Supply has also tightened structurally: global NAND wafer capacity is down ~30% from the 2022 peak (~560k monthly wafers gone), with utilization back near 100%. SanDisk intends to grow shipments through process upgrades rather than new fabs. Its Kioxia joint venture is renewed to 2034, and together the pair delivered ~29% of industry NAND output on just ~13% of capex from 2021–2025. From BiCS5 to BiCS11, bit density per wafer rises ~54% per generation on average — enabling mid-to-high double-digit shipment growth at mid-single-digit capital intensity.

3. Returning 100% of excess cash

Capital-allocation priorities are explicit: invest in the business and technology; keep a healthy, debt-free balance sheet; then return 100% of excess cash to shareholders — with a clear preference for buybacks over dividends. After completing ~$4.5B of a $6B authorization and adding a fresh $14B, roughly $15.5B remains. BofA notes that if ~$100B of cash flow goes mainly to repurchases, SanDisk could buy back close to half its current market cap. TMTB estimates sustained buybacks could lift CY2030 EPS from ~$345 to a range of $467–$787, versus a Street consensus that had EPS fading toward $164 as the cycle rolled over.

4. HBF — a valuation option on the AI inference era

High Bandwidth Flash (HBF) is SanDisk's architecture aimed at the "memory wall." It offers read bandwidth approaching HBM with 8–16x the capacity, making it well-suited to large MoE models, long context windows, and massive KV caches. In internal simulations, a single HBF-equipped GPU can stand in for eight HBM GPUs (~8x capital efficiency). The first HBF die has taped out, with sample inference products expected in 2027. The HBF alliance already includes SK Hynix, Google, Meta and Tenstorrent, with legendary chip architect Jim Keller joining its advisory board. Goldman Sachs frames HBF as an "upside option" on top of the core NAND franchise — one that could reposition SanDisk from a memory maker into an AI-memory platform.

Has SanDisk really escaped the cycle?

Not proven yet. Several caveats remain: the company's forecast that the flash market swells from a historical average near $60B to over $300B in 2026 and close to $500B in 2027 leans heavily on sustained AI data-center demand and elevated prices; NBM has not been stress-tested through a full downturn, and non-NBM volumes still float with the market; new supply from Chinese manufacturers, price mean-reversion, and any cloud-capex slowdown remain live risks; and HBF revenue is still years out.

What genuinely changed is the narrative: lock demand with long-term orders, control supply through process upgrades, release cash via low capital intensity, and convert that cash into EPS through buybacks. The one thing left for SanDisk to prove is whether this is simply a super memory cycle — or a true structural re-rating of the NAND business model.

US-listed ways to play the theme

The names below are ideas tied to the story, not recommendations.

• Sandisk Corporation(SNDK.US) — the direct play; the re-rating thesis lives or dies on NBM execution, the ~80% margin target and buyback pace.
• Micron Technology, Inc.(MU.US) — the closest US-listed proxy for the NAND/DRAM upcycle, and a key HBM supplier levered to the same AI-memory demand.
• Western Digital Corporation(WDC.US) — SanDisk's former parent; benefits from the broader storage/data-center capacity buildout.
• NVIDIA Corporation(NVDA.US)— the AI-compute anchor whose accelerators HBF is designed to complement; a barometer for inference demand.
• Alphabet Inc. Class C(GOOG.US) / Meta Platforms(META.US) — hyperscalers named in the HBF alliance and among the data-center buyers driving NAND demand.
• Broadcom Limited(AVGO.US)— custom AI-silicon leverage to the same inference build-out.
• Semiconductor ETFs (VanEck Vectors Semiconductor ETF(SMH.US), PHLX Sox Semiconductor Sector Ishares(SOXX.US)) — diversified exposure for those wary of single-name memory volatility.

Disclaimer: This article is a summary for informational purposes only and does not constitute investment advice. Memory names are highly cyclical and volatile; do your own research and consider position sizing and risk tolerance before investing.