AI Infrastructure Optimism Will Eventually Expose NAND Pricing And Margin Fragility

Catalysts

About Sandisk

Sandisk develops and sells NAND flash storage products for data center, edge devices and consumer markets.

What are the underlying business or industry changes driving this perspective?

• Data center storage needs tied to AI training and inference are rising quickly. However, if hyperscalers slow or reprioritize their US$1t plus long term AI infrastructure ambitions, Sandisk's enterprise SSD ramp, especially around BiCS8 and Stargate products, could support less revenue growth and leave recent capacity and R&D spending harder to absorb in earnings.
• Management is planning bit supply growth in line with a mid to high teens demand compound annual growth rate. Any easing in AI driven exabyte demand or faster industry wide wafer and node additions could pressure NAND pricing, which would weigh directly on revenue and gross margin progression.
• Customer conversations are extending into 2027 with interest in multi quarter volume and price arrangements. If large buyers eventually push for tougher long term pricing to secure supply, Sandisk could face a ceiling on gross margin expansion just as non GAAP operating expenses remain in the mid US$400m range, limiting operating margin and EPS upside.
• The company is shifting its mix toward data center, where it is currently underrepresented. Slower than expected qualifications for high capacity QLC SSDs such as 128 TB and 256 TB drives, or delays in high bandwidth flash adoption for inference, could leave utilization of BiCS8 focused capex below expectations and cap revenue and free cash flow growth.
• Edge and consumer businesses are currently benefiting from PC refresh, smartphone capacity increases and gaming storage demand. If device upgrade cycles or AI centric premium phone sales soften after calendar 2026, Sandisk's NAND content gains per device may not fully offset weaker units, which would pressure segment revenue and keep net margins more volatile through cycles.

Assumptions

This narrative explores a more pessimistic perspective on Sandisk compared to the consensus, based on a Fair Value that aligns with the bearish cohort of analysts. How have these above catalysts been quantified?

• The bearish analysts are assuming Sandisk's revenue will grow by 16.3% annually over the next 3 years.
• The bearish analysts assume that profit margins will increase from -22.4% today to 13.8% in 3 years time.
• The bearish analysts expect earnings to reach $1.7 billion (and earnings per share of $11.52) by about January 2029, up from $-1.7 billion today. However, there is some disagreement amongst the analysts with the more bullish ones expecting earnings as high as $5.7 billion.
• In order for the above numbers to justify the price target of the more bearish analyst cohort, the company would need to trade at a PE ratio of 21.3x on those 2029 earnings, up from -31.8x today. This future PE is lower than the current PE for the US Tech industry at 26.4x.
• The bearish analysts expect the number of shares outstanding to grow by 7.0% per year for the next 3 years.
• To value all of this in today's terms, we will use a discount rate of 8.24%, as per the Simply Wall St company report.

Risks

What could happen that would invalidate this narrative?

• Customer demand for long term volume and price commitments, including visibility out to calendar 2027 from multiple hyperscalers, could help smooth volatility and support more predictable bit shipments, which may support revenue and earnings resilience even if sentiment on the stock weakens.
• Management expects demand for Sandisk products to exceed supply through the end of calendar 2026 and believes this could extend beyond that period. If this supply constrained backdrop persists while fabs run at full utilization, it may support pricing and gross margins more than a bearish share price view assumes.
• The rapid adoption of BiCS8, with an expectation that it will be Sandisk's most significant node by the end of fiscal 2026 and a majority of bit production exiting that year, along with growing QLC mix toward 40% of bits, could sustain cost per bit improvements and support gross margin and earnings power over time.
• Long term trends such as AI infrastructure investments expected to surpass US$1t by 2030, rising data center exabyte needs and higher NAND content in PCs and premium smartphones, combined with Sandisk's broad presence across data center, edge and consumer, may underpin multi year growth in revenue and free cash flow.
• Sandisk has already moved to a net cash position with US$1.4b in cash and cash equivalents, generated adjusted free cash flow margin of 19.4% in the quarter and plans to continue investing while returning cash to shareholders, which could support shareholder returns and limit downside to equity value versus a more pessimistic outlook.

Valuation

How have all the factors above been brought together to estimate a fair value?

• The assumed bearish price target for Sandisk is $157.81, which represents up to two standard deviations below the consensus price target of $276.32. This valuation is based on what can be assumed as the expectations of Sandisk's future earnings growth, profit margins and other risk factors from analysts on the more bearish end of the spectrum.
  
• However, there is a degree of disagreement amongst analysts, with the most bullish reporting a price target of $408.0, and the most bearish reporting a price target of just $135.0.
  
• In order for you to agree with the more bearish analyst cohort, you'd need to believe that by 2029, revenues will be $12.2 billion, earnings will come to $1.7 billion, and it would be trading on a PE ratio of 21.3x, assuming you use a discount rate of 8.2%.
  
• Given the current share price of $377.41, the analyst price target of $157.81 is 139.2% lower. Despite analysts expecting the underlying business to improve, they seem to believe the market's expectations are too high.
  
• We always encourage you to reach your own conclusions though. So sense check these analyst numbers against your own assumptions and expectations based on your understanding of the business and what you believe is probable.