Non-volatile memory market seen doubling to $191.9 billion by 2035

The global non-volatile memory market is projected to rise from $85.5 billion in 2026 to $191.9 billion by 2035, driven by AI workloads, hyperscale data center spending and automotive electrification. The forecast points to faster growth for next-generation memory types and continued dominance by Asia-Pacific suppliers.

Why it matters: - The market is moving toward nearly 2.5 times its 2025 size, reflecting a broader shift to advanced memory as core infrastructure for AI, cloud and vehicles. - NAND Flash remains the biggest revenue pool, but faster-growth technologies like MRAM, FeRAM and phase-change memory are gaining commercial traction. - Automotive, enterprise storage and hyperscale data centers are turning memory demand into a long-cycle capital investment theme.

What happened: - The global non-volatile memory market was estimated at $78.2 billion in 2025. - The market is projected to grow to $85.5 billion in 2026 and reach $191.9 billion by 2035. - The forecast implies a 9.4% compound annual growth rate from 2026 to 2035. - Market Research Future published the report and offered a full report and sample copy.

The details: - AI and machine learning workload expansion adds about 2.5 percentage points to the market’s CAGR. - Hyperscale data-center capital expenditure adds about 2.0 percentage points. - Automotive electrification and advanced driver-assistance systems add about 1.5 percentage points. - 3D NAND layer scaling adds about 1.2 percentage points. - NAND Flash accounts for roughly 58% of market revenue. - Next-generation memory technologies including MRAM, PCM and ReRAM are forecast to post the fastest segment CAGR at 18.6% through 2035. - NOR Flash is valued at about $9.4 billion in 2025. - EEPROM holds an 8% market share. - Consumer electronics represents about 35% of total market share. - Enterprise storage is valued at about $21.9 billion in 2025. - Automotive applications are projected to grow at a 14.2% CAGR through 2035. - Asia-Pacific holds about 42% of global market revenue. - North America has roughly a 28% share. - Europe accounts for about 20% of global revenue. - Samsung, SK Hynix and Micron have collectively committed more than $100 billion to memory fab construction through 2028. - The U.S. CHIPS and Science Act provides $52.7 billion in incentives. - The European Chips Act provides EUR 43 billion.

Between the lines: - The growth story is less about one product cycle and more about a structural reset in how data is stored, moved and processed. - Legacy planar NAND is being replaced by 200-plus-layer 3D NAND, which lowers cost per bit and supports new use cases. - Large cloud operators and AI infrastructure buyers are shaping demand more directly than consumer device cycles. - Memory is also becoming more strategically regional as governments push domestic chip production and supply-chain localization. - Competitive advantage is shifting toward layer-count leadership, cost per bit and the ability to fund very expensive fabs.

What's next: - AI workloads are expected to keep rising by an estimated 40% annually through 2028. - By 2030, projections indicate 60% of enterprise servers will include dedicated AI accelerators. - Compute Express Link 3.0 is expected to accelerate memory pooling and disaggregation across servers. - Processing-in-memory products may gain traction as vendors push lower-energy AI inference designs. - Automotive memory content could expand further as software-defined vehicles adopt more onboard storage. - Additional capacity investments and new 300-plus-layer NAND roadmaps are likely to influence pricing and market share over the next several years.

The bottom line: - Non-volatile memory is shifting from a commodity component to a strategic enabling technology for AI, cloud and connected vehicles, with the biggest gains likely to accrue to suppliers that can scale capacity and advanced architectures fastest.