Hook
On August 13, SanDisk—a name synonymous with NAND flash—announced a 2028-2030 financial target that sent its stock up 6.3%: 80% non-GAAP gross margins, 75% non-GAAP operating margins, and 100% excess cash return to shareholders. For context, the NAND industry has never seen margins above 60% even during its best cycles. The announcement wasn't just a number—it was a strategic declaration. SanDisk is effectively saying: We are no longer a commodity memory manufacturer. We are a high-margin infrastructure play, betting on AI demand and supply scarcity.
But as a decentralized protocol PM who has spent years watching the blockchain storage space—Filecoin, Arweave, Storj—I see a deeper tension. SanDisk’s target is built on a foundation that decentralized storage projects claim to disrupt: centralized control, heavy capex, and proprietary lock-in. Yet the target itself exposes a glaring vulnerability in the decentralized narrative. If a centralized giant can promise 80% margins by pivoting to AI storage, what does that mean for protocols that promise “cheaper, more open” storage? The numbers don’t lie, but the soul of the industry remains quiet.
Context
SanDisk, after its spin-off from Western Digital, is a pure-play NAND manufacturer with a joint venture with Kioxia (formerly Toshiba Memory). Its core technology is BiCS FLASH 3D NAND, currently at BiCS8 (218 layers), transitioning to BiCS9 (300+ layers). The company’s target is predicated on a radical shift: from selling NAND wafers and consumer SSDs to dominating enterprise AI SSDs (eSSD). The 80% margin implies that by 2028, SanDisk will derive more than 40% of revenue from high-value AI data center storage, where pricing is far less elastic than the consumer market.
But the hidden signal is more profound. The pledge to return 100% of excess cash to shareholders means SanDisk is effectively abandoning the traditional NAND capital expenditure cycle. It will not build new fabs at scale. Instead, it will rely on Kioxia’s capacity and its own controller/firmware ecosystem to extract value. This is a “fab-lite” pivot—a move that reduces risk but also exposes the company to supply chain dependencies. For the blockchain storage world, this is a mirror. Decentralized storage networks also claim to be “fab-lite”: they don’t own the hardware; they aggregate it. But they lack the same pricing power.
Core
Let me break down SanDisk’s target through the lens of the seven dimensions I’ve used to analyze blockchain protocols. Each dimension reveals a lesson for decentralized storage.
1. Technology SanDisk’s 80% margin hinges on stacking layers—literally. The 3D NAND stack is a physical moat: each new layer (BiCS9, 300+) requires years of R&D and billions in investment. The yield at high layers is notoriously low; only a handful of companies can do it. Decentralized storage protocols, by contrast, have no such technological moat. They rely on commodity hardware (HDDs, SSDs) that anyone can buy. The barrier to entry is low, which means margin compression is inevitable. As I wrote in my Gitcoin days, “The code enforces fairness, but it cannot enforce scarcity.” Without a technical bottleneck, prices will race to the bottom.
2. Supply Chain SanDisk’s supply chain is concentrated in Japan (Kioxia JV) and the US. It faces low geopolitical risk because NAND doesn’t require EUV. But the 80% margin assumes that Kioxia will continue to supply wafers at favorable prices. This is a fragile handshake. In decentralized storage, the supply chain is distributed across thousands of miners. That distribution is a strength in resilience but a weakness in cost control. The protocol cannot dictate pricing; it must rely on market forces. When I audited smart contracts for Gitcoin, I saw how quadratic funding could align incentives, but it couldn’t guarantee a floor price. The same applies to storage: without a central coordinator, margins are left to the invisible hand—which often slaps hard.
3. Capex and Depreciation SanDisk’s 100% cash return target implies it will slash capex to maintenance levels only. By 2028, its older fabs will be fully depreciated, dramatically lowering the cost of goods sold. This is a classic “cash cow” strategy. Blockchain storage protocols, however, are perpetual capex machines. Miners must constantly upgrade hardware to stay competitive. The protocol itself doesn’t own the hardware, so it doesn’t benefit from depreciation. Instead, the network fees must cover the miners’ capital costs—and those costs are unpredictable. In the Terra/Luna collapse, I saw how reliance on external capital can be fatal. Storage protocols face a similar fragility: if the token price drops, miners leave, and the network becomes unreliable.
4. Market Demand SanDisk’s target is a bet on AI-driven storage demand. The thesis is that AI servers require 3-5x more NAND than traditional servers, and that demand will outstrip supply for years. This is a reasonable bet. But decentralized storage is not positioned for high-performance AI workloads. Filecoin and Arweave are optimized for cold storage—archival, backup, media. The latency and throughput are orders of magnitude worse than a centralized eSSD. The AI boom will benefit centralized storage first, leaving decentralized protocols to fight over the leftovers. As I said in my Nifty Gateway stand, “Decentralization must mean economic justice, not just technological novelty.” But the market is voting with its wallet: AI is pouring money into centralized solutions, not decentralized ones.
5. Geopolitics SanDisk benefits from being a US-based company with a Japanese manufacturing base. It can sell to China (with some restrictions), but it’s insulated from the worst trade wars. Decentralized storage protocols are global by design, but that exposes them to regulatory fragmentation. For example, Filecoin’s miners are concentrated in China, and if the US restricts Chinese hardware or data flows, the network could be split. The 2025 Bitcoin ETF regulatory work taught me that clarity can enhance decentralization, but it can also compartmentalize it. SanDisk’s margins are protected by its home-government alignment; decentralized protocols have no such umbrella.
6. Competition SanDisk’s main competition is Samsung, SK Hynix, Micron, and Kioxia. These are all massive, well-funded incumbents. But SanDisk’s pivot to AI eSSD creates a differentiated niche. In decentralized storage, the competition is primarily other protocols (Filecoin vs. Arweave vs. Storj) and increasingly centralized cloud providers (AWS, Azure). The latter have infinite resources and better performance. The former are fragmented. The 80% margin target assumes that SanDisk can escape the commodity trap. For decentralized storage, the commodity trap is the default state. The protocols are competing on price, not on performance or moat. That’s a race to the bottom.
7. Financials SanDisk’s 80% gross margin is unprecedented in storage. It implies a net income margin of 60-70% after operating expenses. For a hardware company, this is closer to a software company’s margin. The only way to achieve it is to have pricing power akin to a monopoly. SanDisk is betting that by 2028, the AI storage market will be so tight that buyers will pay a premium for its vertically integrated solution. Decentralized storage protocols, by contrast, have no pricing power. The token price is determined by speculation and utility, but the actual storage cost is driven by supply and demand of hardware. The protocols take a small fee (e.g., Filecoin’s 20% of mining rewards), but that’s still a fraction of the value. The real value accrues to the miners, not the protocol. This is the opposite of SanDisk’s model: SanDisk captures value at the product level; decentralized storage leaks value to the infrastructure layer.
Contrarian
Here’s the counter-intuitive angle: SanDisk’s target might actually be a validation of the decentralized storage thesis, not a refutation. The 80% margin relies on scarcity—both of technology (high-layer NAND) and of capital (other firms not expanding). If SanDisk succeeds, it will draw even more attention to the AI data center boom, which will eventually create a need for secondary and archival storage. That’s where decentralized storage excels. The 2022 Terra debacle taught me that when the bubble bursts, the survivors are those with real utility. SanDisk is betting on a future of hyper-concentrated storage; but the decentralized alternative is already being built for a future where data is abundant and censorship is a risk.
Moreover, SanDisk’s fab-lite strategy is a double-edged sword. By relying on Kioxia, it cedes control over its own destiny. If Kioxia decides to prioritize its own brand or raise prices, SanDisk’s margins vanish. The partnership is a fragile alliance. In contrast, decentralized storage networks are built on open protocols with no single point of failure. The miners are independent, and the protocol is governed by a distributed community. While this lack of coordination leads to inefficiency, it also prevents a single player from extracting rent. The 80% margin is a rent extraction number. Decentralized storage, by design, cannot extract that much rent—it must pass most of the value to the participants. That might be a feature, not a bug.
Takeaway
SanDisk’s announcement is a wake-up call for the blockchain storage sector. It shows that the real money is in high-performance, high-margin storage, not in the commodity tier. If decentralized protocols want to capture value, they must either (a) build a technological moat (e.g., specialized hardware, zero-knowledge proofs for storage verification) or (b) pivot to a niche that centralized players ignore—like permanent data, verifiable compute, or provenance. The path of “cheaper decentralized storage” is a dead end when the centralized alternative is both cheap and fast.
When the graph spikes, the soul remains quiet. The NAND industry has seen booms and busts, and SanDisk’s target may be the peak of the peak. For blockchain storage, the question is not whether we can match SanDisk’s margins, but whether we can survive the transition from hype to utility. The numbers are a challenge, but the soul of the industry—the belief that open, owner-controlled data is a human right—remains the real compass. Trust, not code, will be the final currency.