Two things happened simultaneously on the evening of May 26th. Xiaomi released its Q1 2026 financial report, with total revenue of 99.1 billion yuan, a year-on-year decrease of 10.9%; adjusted net profit was 6.07 billion yuan, a year-on-year plunge of 43.1%. Mobile phone business revenue was 44.3 billion yuan, a year-on-year decrease of 12.5%, and the gross profit margin fell to 10.1%, a decrease of 2.3 percentage points compared with the same period last year.
At the financial report conference call, Xiaomi Group President Lu Weibing said that the price of the same version of memory has soared nearly 4 times compared with the same period last year. For a mobile phone configured with 12GB LPDDR5 + 512GB UFS, the memory cost alone has increased by about 1,500.00 yuan. He said that Xiaomi “will not pass on the cost increase of memory to consumers,” but predicted that the price increase cycle will continue until 2027 or even 2028. In order to survive, Xiaomi took the initiative to cut entry-level models, and quarterly shipments fell to 33.80 million units.
The second thing was that Micron Technology surged more than 19% in a single day, and its market value exceeded 1 trillion US dollars. UBS directly raised Micron’s target price from 535.00 USD to 1,625.00 USD, a one-time increase of approximately 204%, becoming the highest target price among the 46 brokerages currently covering Micron. A few days ago, Citi just raised Micron’s target price from 425.00 USD to 840.00 USD, and HSBC also raised it from 750.00 USD to 1,100.00 USD. Wall Street has not had such a unified opinion on the same cyclical stock for a long time. 12 months ago, Micron’s stock price was less than 110.00 USD, and it has increased by 8 times within a year.
On the same day, the trillion-dollar carnival of selling memory and the halved profits of buying memory occurred. Goldman Sachs played a thought-provoking role in this carnival. In December 2025, Goldman Sachs gave Micron a neutral rating with a target price of 205.00 USD. In the first quarter of 2026, Goldman Sachs reduced its Micron position by nearly 20%. On March 19, the day Micron released its financial report, Goldman Sachs raised its target price from 360.00 USD to 400.00 USD, but maintained a neutral rating. At this time, the stock price had far exceeded 400.00 USD. Subsequently, Micron surged 40% in a week, and Goldman Sachs accurately missed the opportunity. On May 17, Goldman Sachs issued a storage industry report, concluding that it was “the most severe supply shortage in 15 years,” and raised the overall rating of the storage industry, but remained neutral on Micron, with a target price of 400.00 USD.
01 Why is it going crazy? A new story called LTA?
UBS analyst Timothy Arcuri’s May 26th research report’s core argument is that Long-Term Agreements (LTA) are fundamentally eliminating the cyclical nature of the storage industry. Storage chips are the most commodity-like variety in the semiconductor industry. The prices of DRAM and NAND have followed the cruel rule of rising for two years and falling for two years for forty years. UBS believes that the “curse of cycles” for these companies will be broken by AI. Cloud vendors such as Microsoft, Google, Amazon, and Meta, in order to lock in HBM and DDR5 supply in the AI arms race, have begun to actively sign 3 to 5-year fixed-price long-term contracts with storage manufacturers, with advance payments. The power relationship of the industrial chain has been reversed.
UBS’s model calculation shows that if LTA is included in Micron’s earnings forecast, even if DRAM spot prices plummet by 50% in fiscal year 2029, Micron’s full-year earnings per share can still remain above 100.00 USD. From the perspective of the valuation system, if the cyclicality disappears, storage stocks should not be valued as cyclical stocks, but as infrastructure utilities, jumping from 8 to 15 times PE to 20 to 30 times PE. JPMorgan Chase also issued a research report with similar conclusions in mid-May, while Citi’s logic is that HBM production will squeeze ordinary DRAM wafer capacity, leading to a long-term shortage of general storage.
02 This storage is not that storage
Wall Street uses the “storage super cycle” to tell a unified bull market narrative, but the 2026 storage market presents three levels of differentiation. The first layer is AI storage (HBM, server DDR5, enterprise-level SSD), where price increases, stockouts, and long-term agreements to lock in production capacity occur simultaneously, which is the story of Micron’s trillion-dollar market value. The second layer is mobile phone and embedded storage, where price increases are also fierce, and the proportion of memory in the mobile phone BOM has risen to 30% to 40%. Xiaomi is in this layer and is forced to accept price increases. The third layer is PC retail spot, where there is a reverse fluctuation. Due to sufficient channel inventory, the prices of some products have fallen. The essence of this differentiation is that the three major storage manufacturers are shifting wafer capacity from consumer-grade to AI.
03 Can long-term agreements really eliminate cycles?
The logic of long-term agreements seems solid, but there is a prerequisite: the demand side does not collapse. There is nothing in the physical world that grows at more than 40% forever. Once the growth rate of AI infrastructure investment slows down, the supply and demand balance of storage chips may reverse within 18 months. The three storage manufacturers are now expanding production like crazy, and Micron’s revenue growth mainly depends on price elasticity rather than sales elasticity. Multiplying the profit at the top of the cycle by a seemingly “reasonable” multiple to get a seemingly “not expensive” valuation is the most classic valuation trap when cyclical stocks peak. When the entire Wall Street shouts “this time is different” at the same time, historical experience is worth vigilance.
[Tencent Technology]
Memory Market Disruption: Implications for Crypto’s Infrastructure Layer
The simultaneous occurrence of Xiaomi’s profit collapse and Micron’s trillion-dollar valuation represents a fundamental market realignment with significant implications for the crypto ecosystem. This dichotomy—where memory producers thrive while consumers struggle—mirrors emerging dynamics in blockchain infrastructure and token economics.
The Memory Market Paradigm Shift
The memory chip sector is undergoing its most significant transformation in four decades. The traditional “two years up, two years down” cycle appears broken due to Long-Term Agreements (LTAs) between cloud giants and memory manufacturers. These 3-5 year fixed-price contracts with advance payments have reversed industry power dynamics, creating what UBS analysts call the “elimination of cyclicality.”
For crypto investors, this narrative should resonate strongly. The parallels between traditional infrastructure and blockchain are striking: just as Microsoft, Google, Amazon, and Meta are locking in memory supply for AI expansion, we’re seeing similar dynamics where Layer 1 protocols and infrastructure projects secure long-term partnerships to stabilize their operations.
Direct Impacts on Crypto Mining
The memory cost surge directly affects mining economics. With memory components accounting for 30-40% of mobile phone BOM costs (according to Xiaomi), similar cost pressures are impacting mining rigs. This particularly affects:
- Memory-hard algorithms – Projects like Ergo and Neoxa that rely on memory-intensive operations face increased hardware barriers
- GPU mining – The most accessible form of mining for retail participants is becoming prohibitively expensive
- Mining profitability – The mining sector may face consolidation, accelerating the shift toward professional operations and reducing network decentralization
The result could be a mining landscape dominated by well-capitalized entities, potentially compromising the decentralization ethos that underpins many cryptocurrencies.
AI-Crypto Convergence
The AI storage boom (HBM, server DDR5, enterprise SSD) represents a significant opportunity for crypto projects focused on:
- Decentralized AI infrastructure – Projects like Fetch.ai and SingularityNET that provide alternatives to centralized AI solutions
- Oracles – Data providers for AI models on blockchain could benefit from increased AI adoption
- Computational marketplaces – Platforms that tokenize computing resources may see increased demand
However, investors must exercise caution. The current Wall Street consensus on memory chips—with UBS raising Micron’s target price by 204%—echoes the “this time is different” narrative that often precedes market peaks. Crypto investors should be particularly wary of similar enthusiasm in AI-related tokens, which may be disconnected from fundamental value.
Decentralized Storage Opportunity
The most compelling crypto opportunity arising from the memory market disruption lies in decentralized storage solutions. As traditional cloud providers scramble to secure memory through LTAs and face rising costs, decentralized alternatives become increasingly attractive:
- Filecoin – Benefits from the growing need for cost-effective storage solutions
- Arweave – Its one-time payment model contrasts sharply with the rising subscription costs of traditional cloud
- Storj – Offers an alternative for organizations seeking more transparent and potentially cheaper storage
These projects address the core issue highlighted by Xiaomi’s experience: the inability to pass rising infrastructure costs to end users. Decentralized storage offers a fundamentally different economic model that could prove more sustainable in a high-cost memory environment.
Token Valuation Implications
The memory market’s transformation offers important lessons for crypto token valuation:
- Infrastructure vs. Application – Just as Micron’s valuation shifted from cyclical (8-15x PE) to infrastructure-like (20-30x PE), blockchain infrastructure tokens may warrant different valuation metrics than application tokens
- LTAs in Crypto – Protocols that establish long-term partnerships and stable revenue streams may be valued more favorably than those with highly variable tokenomics
- Resource Efficiency – In a high-cost memory environment, protocols that minimize resource requirements could gain competitive advantages
The memory market’s differentiation between AI storage, mobile/embedded storage, and PC retail spot mirrors the segmentation within crypto between infrastructure, DeFi, and NFT projects. Each segment faces different economic pressures and opportunities.
Risk Considerations
Several risks should temper enthusiasm:
- Demand Collapse – As noted in the report, nothing grows at >40% forever. If AI infrastructure investment slows, the memory market could reverse quickly, creating ripple effects in related sectors
- Production Expansion – Memory manufacturers are aggressively expanding capacity, which could lead to oversupply if demand doesn’t materialize as expected
- Correlation Risk – As crypto becomes increasingly correlated with traditional tech markets, volatility in the semiconductor sector could impact crypto markets
Conclusion
The memory market disruption represents both challenges and opportunities for the crypto ecosystem. While rising memory costs could pressure mining economics and development expenses, the underlying trends toward AI integration and decentralized infrastructure create compelling investment opportunities. Crypto investors should focus on projects that address the core issues highlighted by this market realignment: the need for sustainable infrastructure economics, the convergence of AI and blockchain, and the value proposition of decentralized alternatives to centralized services.
The memory market’s transformation serves as a valuable case study for crypto investors, demonstrating how fundamental shifts in traditional infrastructure markets can create both risks and opportunities for emerging technologies. As always, a focus on fundamentals and a diversified approach will be key to navigating these complex dynamics.