SanDisk Investor Day Slogan: How Goldman Sachs Interprets It

TL;DR · Goldman Sachs maintains a buy rating on SanDisk with a target price of $2,200, representing a 63.7% upside from the closing price of $1,344.29 on August 12. · The core of the valuation is the target gross margin of about 80% from FY2028 to FY2030, along with revenue and price visibility from a $94 billion long-term customer agreement. · Low capital expenditures and the remaining $15.5 billion in buyback authorization are expected to amplify EPS, but long-term agreements are still not enough to completely eliminate the NAND cycle. · HBF provides additional imagination space for AI inference storage, but the first product samples are expected to be released only by 2027.

After holding its 2026 Investor Day on August 13, SanDisk maintained its buy rating and a 12-month target price of $2,200. Compared to the closing price of $1,344.29 on August 12, this target price corresponds to an upside of approximately 63.7%.

The high target price is supported not only by NAND prices remaining high. Goldman Sachs places more importance on SanDisk's efforts to reduce cyclical fluctuations in its traditional storage business through long-term customer agreements, low capital expenditures, and large-scale buybacks, converting high profits into more stable free cash flow.

The long-term targets set by the company for FY2028 to FY2030 are quite aggressive: revenue growth is maintained at about 15% to 19%, non-GAAP gross margin at about 80%, non-GAAP operating margin at about 75%, and adjusted free cash flow margin at about 50%, while capital expenditure intensity is controlled at mid-single-digit percentages.

Whether these targets can be achieved hinges not on whether NAND prices can continue to rise indefinitely, but on whether SanDisk can maintain price and demand floors through long-term agreements and convert more profits into shareholder returns through manufacturing efficiency.

Target Price of $2,200: Betting on Sustained High Profits

Goldman Sachs's target price of $2,200 is based on a 20x price-to-earnings ratio and a normalized EPS of $110.

At the time of the report's release, Goldman Sachs noted that SanDisk's stock price rose about 15% after the Investor Day, but believes that the company's higher-than-expected long-term financial targets, capital return policy, and HBF technology potential have not yet been fully factored into the market valuation.

According to Goldman Sachs's model, SanDisk's revenue is expected to grow from $20.248 billion in FY2026 to $71.041 billion in FY2028, while EPS is projected to rise from $70.16 to $284.98. During the same period, EV/EBITDA is expected to drop from 6.3x to 2.5x, and free cash flow yield is expected to increase from 13.4% to 19.5%.

However, the EPS close to $285 in FY2028 is not the profit base directly used for Goldman Sachs's target price. Instead, Goldman Sachs uses a normalized EPS of $110, indicating that its valuation has already discounted the current high profit levels of NAND.

What is truly aggressive is SanDisk's own long-term operational targets. A gross margin of about 80% and an operating margin of 75% imply that the company believes its current profitability does not solely stem from short-term price increases but may be sustained through customer agreements, product mix, and manufacturing efficiency.

This is also the core of market contention: Is SanDisk merely in an upcycle for NAND, or has it established a new business model capable of reducing cyclical fluctuations?

$94 Billion Long-Term Agreements Set Revenue and Price Floors

The core of SanDisk's new business model is NBM, or multi-year customer agreements.

According to the company, the total contract value of these agreements, calculated at contract floor prices, is approximately $94 billion, with remaining performance obligations of about $91 billion, covering 8 customers and including about $16.5 billion in financial guarantees. Among the customers are 3 large-scale cloud providers in the U.S., with a weighted average contract duration of about 4 years.

These agreements enhance visibility into future shipments. SanDisk expects that about 50% of the planned shipments for FY2027 will be covered by NBM, with coverage for FY2028 expected to be about two-thirds.

The pricing method also differs from traditional spot trading. Recent contracts typically use fixed prices, while those for further years set price floors and ceilings. According to the company's statements at the Investor Day, even if executed at contract floor prices, the gross margin for the related business can still reach about 80%.

This means that under normal contract execution, SanDisk is no longer fully exposed to fluctuations in NAND spot prices, and the revenue and profit floors will be clearer than in the past.

However, Goldman Sachs has not directly concluded that long-term agreements have eliminated the industry cycle. The report clearly states that whether NBM can truly mitigate the NAND cycle remains to be verified, and its effects will take longer to reflect in stock valuations.

Long-term agreements lock in certain customers, shipment volumes, and price ranges, but not the entire NAND market. For FY2027 and FY2028, about 50% and one-third of the planned shipment volumes, respectively, remain uncovered. Supply expansion, demand decline, and changes in customer procurement rhythms may still affect overall prices and profits.

Therefore, the $94 billion contract serves more as a buffer for the cycle rather than eliminating it entirely.

Low Capital Expenditures and Buybacks Convert Profits into EPS

In addition to long-term agreements, another advantage for SanDisk comes from manufacturing capital efficiency.

SanDisk's joint manufacturing arrangement with Kioxia has been extended to 2034. The company states that between 2021 and 2025, the SanDisk-Kioxia joint system contributed 29% of the industry's bit growth while accounting for only 13% of the industry's capital expenditures during the same period.

Based on the capital investment required for each additional 1EB output in 2025, the industry average is about 2.7 times that of the SanDisk-Kioxia system. The company expects to maintain mid-double-digit bit growth while controlling capital expenditure intensity at mid-single-digit percentages by increasing wafer output, improving production efficiency, and better utilizing existing equipment and clean rooms.

This is particularly important for storage companies. Traditional NAND manufacturers often need to add substantial capital expenditures during upcycles, which can lead to oversupply. Lower capital expenditure intensity means that if SanDisk's profit targets are met, more cash can remain in the company rather than being reinvested in the next round of expansion.

SanDisk has clarified its capital allocation sequence: first, maintain technological leadership and invest in the business; second, maintain a debt-free and sufficiently liquid balance sheet; and finally, return 100% of excess cash after business reinvestment to shareholders, primarily through share buybacks.

The company's board previously authorized a $6 billion buyback, of which approximately $4.5 billion has been executed. After an additional $14 billion authorization, the remaining buyback capacity is about $15.5 billion.

If the adjusted free cash flow rate can indeed approach 50%, continuous buybacks will reduce the outstanding share count and further amplify earnings per share. However, buybacks are merely a method of cash flow distribution and cannot replace the operational targets themselves. Only with sustained profits and free cash flow can buybacks provide sustainable support for valuations.

HBF Opens Up AI Inference Space, But Revenue Realization Is Still Far Off

Beyond traditional NAND, SanDisk is also attempting to enter the higher-value storage segment of AI inference systems through HBF, or high-bandwidth flash.

As model context lengths increase and inference chains extend, KV Cache needs to store and quickly retrieve more intermediate computation results. GPUs not only require computing power but also need larger storage capacity and higher data reading bandwidth. SanDisk refers to this issue as the "memory wall" in AI inference.

The company estimates that by 2032, the total addressable market for AI data center storage will reach 1.2ZB, with KV Cache accounting for about 35%.

HBF aims to fill the gap between HBM and SSD. According to SanDisk's technical vision, HBF can provide reading bandwidth close to that of HBM while increasing capacity to 8 to 16 times that of HBM.

In SanDisk's own simulations, an HBF-only architecture can achieve the same token output with about half the GPU capacity of an HBM-only solution. If this result can be replicated in real systems, HBF could not only increase storage capacity but also reduce the number of GPUs required for AI inference and overall capital expenditures.

However, this data currently comes mainly from company simulations and does not equate to validation through customer deployment. Goldman Sachs reports that SanDisk's first HBF product has completed wafer fabrication, with the first samples expected to be released in 2027. After that, it will still need to go through customer certification, system architecture adaptation, cost assessment, and ramp-up for mass production.

Therefore, HBF is better viewed as a long-term upside option in SanDisk's valuation rather than a large-scale business that can already be accounted for in recent revenues.

Goldman Sachs is Bullish, But Long-Term Agreements Still Need Cycle Testing

Goldman Sachs remains bullish on SanDisk, primarily because recent NAND supply additions are limited, and the company's product mix and customer structure are still improving; the longer-term logic is that NBM agreements, low capital expenditures, and buybacks may collectively enhance the stability of profits.

This logic can be summarized as follows: long-term agreements enhance visibility into shipment volumes and prices, manufacturing efficiency converts profits into free cash flow, buybacks then convert cash flow into EPS, and HBF provides additional growth space for AI inference storage.

However, each link still requires verification.

While NBM covers about 50% of planned shipments for FY2027 and about two-thirds for FY2028, the remaining business will still be affected by NAND prices and customer procurement rhythms. Competitors like YMTC continuing to advance their technology routes may also alter the industry's supply, pricing, and share dynamics.

The approximately 80% long-term gross margin cannot rely solely on contracts; it also depends on product mix, manufacturing efficiency, and overall supply and demand for NAND. If industry prices decline significantly or customer agreements do not meet expectations, the company's targets may still face pressure.

HBF needs to prove that its technical simulations can translate into real deployment advantages. There remains a gap between 2027 samples and large-scale revenue, involving customer certification, supply chain support, and system-level cost validation.

Therefore, this report does not conclude that "the NAND cycle has ended" but rather presents a conditional high-elasticity judgment: if SanDisk can execute the $94 billion long-term agreements as promised, maintain gross margins around 80%, and control capital expenditure intensity at mid-single digits, the current stock price may still underestimate its long-term cash flow; conversely, if the execution of agreements, industry supply and demand, or HBF deployment falls short of expectations, the optimistic assumptions embedded in the $2,200 target price will be repriced.