The current memory semiconductor boom will run for three to four years, according to a forecast reported by THE ELEC, Korea's electronics industry trade outlet. The projection frames the upcycle not as a short pricing spike but as a sustained expansion — and it names one variable above all others that could reshape the timeline: China's technological catch-up.

The forecast lands amid an industry still working through the aftereffects of the 2023 downturn, the deepest memory market contraction in a decade. Pricing recovered sharply through 2024, driven by tight supply across DRAM and NAND and by demand for high-bandwidth memory tied to AI accelerators. Against that backdrop, a three-to-four-year horizon would place the boom's end somewhere between roughly 2028 and 2029, assuming the cycle follows the projected path.

Why is China the key variable?

The forecast's central caveat concerns China. The report identifies Chinese technological catch-up — the narrowing gap between domestic Chinese producers and the incumbent memory makers in Korea and the United States — as the factor most likely to alter the shape of the cycle.

The mechanics are straightforward. Memory is a commodity-adjacent business where marginal supply determines pricing. If Chinese manufacturers succeed in scaling competitive NAND or DRAM output faster than expected, added capacity could pressure prices and shorten the boom. If they stall — whether on process technology, tooling access, or yield — the incumbent suppliers retain pricing power longer, extending the favorable period for the market leaders.

This is why analysts treat China not as a side issue but as the swing variable in any multi-year memory forecast. A catch-up measured in years rather than decades changes capital allocation decisions made now: fab construction timelines, equipment orders and capacity commitments all run on multi-year horizons, and a misread of Chinese capability carries real cost in either direction.

What does a 3–4 year horizon change?

Cycle length is the single most consequential number in memory industry planning. Suppliers that expect a two-year boom invest cautiously; suppliers that expect four years commit to aggressive capacity expansion, because new fabs take years to build and qualify.

A three-to-four-year projection supports:

• Sustained capital expenditure by the major memory producers on fab capacity and advanced process nodes
• Longer equipment order backlogs for semiconductor toolmakers serving the memory segment
• Extended pricing stability for DRAM and NAND buyers, who plan procurement around expected supply conditions
• Continued strength in high-value segments, where AI-related demand compounds the underlying cycle

The forecast implicitly assumes that demand tailwinds — notably AI-driven memory requirements — persist long enough to absorb supply additions through the projection window. Any material demand shock would compress the timeline independently of what happens on the Chinese supply side.

What could break the forecast?

Two failure modes stand out against the report's framing.

The first is faster-than-expected Chinese progress. Every quarter that Chinese producers close the technology gap pulls potential incremental supply forward, and memory pricing historically responds quickly when meaningful new capacity reaches the market.

The second is a demand break. Memory booms end when supply outruns consumption, and the report's multi-year horizon depends on AI and conventional demand continuing to absorb output growth through the late 2020s.

For now, the base case reported by THE ELEC is clear: a three-to-four-year expansion, with China's pace of technological catch-up as the variable to watch. Buyers, suppliers and equipment makers alike will treat that single variable as the industry's most important leading indicator for the remainder of the decade.