Nintendo's 41% Unit Drop Is a Memory-Cost Event — and the Unchanged 16.5M Forecast Is the Real Risk

Nintendo shipped 2.67 million Switch 2 consoles in the three months to September 30, 2026, fell 41% to 2.67 million units year over year and below the 3.82 million shipped in the June quarter. The company had raised the console's price in every major market first, and its own fiscal 2027 forecast now carries roughly ¥100 billion of higher component costs, mostly memory, inside cost of sales.

Volume broke; revenue did not. Quarterly net sales fell 15% to ¥448.2 billion, roughly one-third of the unit decline, because the price increase carried part of the load and because 26.35 million installed consoles keep converting into software. Nintendo lost boxes, not money — the distinction that makes this the cleanest elasticity case in consumer hardware.

Guidance is where this gets uncomfortable. Nintendo left its full-year forecast exactly as it stood on May 8: net sales of ¥2,050.0 billion, operating profit of ¥370.0 billion and 16.50 million Switch 2 units. With 6.49 million units sold in the first half, the target now requires 10.01 million units in the second half, 5.5% more than the 9.49 million shipped in the same half of fiscal 2026.

Nintendo's first-half operating profit rose 49.1% to ¥216.5 billion while net sales fell 12.1%. Divide the profit by the consoles shipped and the economics invert: operating profit per Switch 2 unit more than doubled to ¥33,400 from ¥14,000 a year earlier. The hardware got more expensive to build and to buy; the software on it got far more profitable.

The software leg did the heavy lifting: Switch 2 software shipments rose 18.8% to 24.49 million units in the first half while hardware units fell 37.4%, lifting software sold per console from 1.99 to 3.77. Digital sales reached ¥249.6 billion, up 60.5%, and the installed base now stands at 26.35 million consoles carrying 73.21 million software units. A weak software slate would have turned this into a profit collapse — it did not.

Nintendo's ¥100 billion component line is somebody else's record quarter. Micron Technology reported fiscal fourth-quarter revenue of $54.23 billion against $11.32 billion a year earlier, with DRAM revenue of $39.8 billion and NAND of $14.1 billion. SK hynix posted third-quarter revenue of KRW 104.6 trillion at an 80% operating margin, and Samsung Electronics reported third-quarter operating profit of KRW 107.4 trillion, up 783% year over year.

The mechanism is allocation, not luck. SK hynix expects the supply-demand imbalance to persist through 2030 because leading-edge capacity takes years to build, and it is shifting lines toward HBM4 and server products. That is what a console maker competes with: the same wafer capacity, priced by data-centre demand, with mobile and console memory left as the residual. TrendForce expects conventional DRAM contract prices to rise 10–15% sequentially in the December quarter and NAND 15–20%.

The same memory bill lands differently depending on what the buyer can defer. A $500 console is the most deferrable purchase in the chain, and it broke. Sony Group raised the US PlayStation 5 to $649.99 on April 2, 2026 on memory costs, then walked most of that increase back to $549.99 by August 21 — a competitor confirming the same elasticity in real time.

The pattern repeats: every company that raised price kept revenue and lost units, and every lost unit was replaced by a higher average selling price. Apple's September quarter is the proof that the squeeze is survivable when buyers trade up rather than out — $114.5 billion of revenue, up 12%, and a record September for iPhone — but even Apple guided December-quarter gross margin down to 45–46%, with most of the sequential decline attributed to memory. Worldwide PC shipments fell 4% year over year to 65 million units in the June quarter.

Over the next three quarters the swing factor is supply, not software. The December quarter is Nintendo's first full period at the new US and European prices, 4Q26 contract prices set the cost floor Nintendo plans against, and February 2027 is when the 16.50 million-unit target gets confirmed or cut. Nobody has to believe in the software slate to reach that date — the arithmetic does the work.

Over one to three years the question changes from elasticity to architecture. Capacity is migrating to HBM4 and server parts, which keeps the consumer memory floor high and the console price floor with it. For Nintendo, that means a structurally higher price, a longer payback period on every console, and a business that has to earn its margin on software rather than on the box. If it can hold ¥370 billion of operating profit on 16.5 million consoles, the memory cycle has changed what a console cycle is worth, not merely how many units it ships.

Nintendo closed at ¥8,289 on November 5, 2026, 43% below its 52-week high of ¥14,630, on 21.7 times trailing earnings with net cash and a 2.5% dividend yield. The memory complex trades at a fraction of that: SK hynix at 3.9 times trailing enterprise value to EBITDA with a 20.4% earnings yield, Samsung Electronics at 3.5 times, Micron Technology at 8.0 times on a 92% return on equity.

That dispersion is the investable expression of the thesis. Memory suppliers are being paid for scarcity; assemblers are being asked to absorb it, and the market has already made its choice about who deserves a premium. Nintendo's own numbers are the argument in miniature: revenue guided11.4% lower, operating profit guided 2.8% higher. The company that lost a third of its console volume and still grew profit is now valued at a scarcity-asset multiple while the sellers of the scarce input trade at 3 to 4 times earnings.