DRAM & Flash Prices 2026 (Update) Part III: 2027 Outlook

2027 and the Outlook: What the Market Situation Means for Procurement and Planning

Although not every single product line in the market is already completely sold out, 2027 has long ceased to be an open planning year for strategic capacity, large buyers and prioritized product classes.

The reason lies in decisions that have already been made today: multi-year offtake agreements between manufacturers and major customers, as well as direct equity stakes through which market participants secure manufacturing capacity.

For the second half of 2026, we do not expect further price jumps on the scale of the first half – but we do expect a stabilization at a high, tight and allocation-driven level. The lowest daily price is therefore continuing to lose relevance as a point of reference.

2027: No Longer an Open Planning Year

From our perspective, the outlook for 2027 can be summed up in a single sentence: Although not every single product line in the market is already completely sold out, 2027 has long ceased to be an open planning year for strategic capacity, large buyers and prioritized product classes.

In our view, there are essentially two reasons for this: multi-year Strategic Customer Agreements and the purchase of DRAM capacity through equity stakes.

1. Multi-year Strategic Customer Agreements

Micron provides the strongest evidence here. According to its own statements, the company has concluded 16 multi-year Strategic Customer Agreements. These agreements typically run for five years, from calendar year 2026 to 2030, and represent around 20 percent of expected DRAM volume and roughly one third of expected NAND volume over that period. For 14 of these agreements, Micron cites cumulative minimum revenue of around USD 100 billion, along with customer deposits and financial commitments of around USD 22 billion.

2. Purchasing DRAM capacity through equity stakes

Capacity is being secured not only through take-or-pay contracts but also through direct equity stakes. One example beyond Micron: Taiwanese DRAM manufacturer Nanya Technology completed a private placement of around USD 2.5 billion in the first quarter of 2026, in which SanDisk, Kioxia, Solidigm (SK hynix) and Cisco Systems participated jointly – reportedly the first simultaneous equity investment by four global technology groups in Taiwan's memory industry.

Some of the investors additionally linked their stakes to multi-year supply agreements. The capital flows specifically into manufacturing capacity – evidence that NAND and SSD manufacturers are now buying directly into DRAM capacity to secure their own ability to supply (cf. TrendForce, 2026c).

In addition, there is the time lag on the supply side. Micron expects the first wafer output from its new ID1 fab in mid-2027, and from the ID2 fab not until late 2028. Additional capacity in Taiwan and advanced packaging capacity in Singapore will likewise become relevant step by step from 2027 onward.

2027 is therefore already being shaped by decisions that large customers and manufacturers are making now. For enterprise and data center products, this means: anyone who needs relevant volumes is no longer operating within a conventional short-term annual planning window. Forecast quality, technical qualification, manufacturer access and early project planning continue to gain importance.

The Spot Market Remains a Symptom, Not a Solution

In this environment, the spot market remains an important indicator, but not a reliable valve.

Our previous assessment already described how the spot market can no longer be understood as the actual lead market. It becomes relevant above all when regular allocations are insufficient, when projects need additional volumes at short notice, or when supply chains stall elsewhere.

What is remarkable here is the reversal of direction: in normal market phases, the spot market served large OEMs and buyers primarily as a valve for releasing surplus volumes. Today that inflow is drying up – and the same buyer groups are instead appearing as purchasers to close allocation gaps. The spot market has therefore reversed its direction of flow.

Little has changed in this respect as of mid-2026. On the contrary: the more large customers secure volumes on a long-term basis, the more the spot market becomes a residual market. It then no longer reflects the normal price of a healthy supply-demand situation, but indicates how scarce freely available product has become outside long-term contracts and allocations.

From a strategic perspective, the spot market therefore remains an early warning system. Rising spot prices or thinning availability are not the cause of the shortage, but a visible signal that upstream contract and allocation markets are already under strain.

Outlook for the Second Half of 2026

A differentiated picture emerges for the second half of 2026. Following the very strong price movements of the first half, further jumps of the same magnitude are not necessarily to be expected. Part of the repricing has already taken place, and manufacturers themselves expect the rate of price increases to moderate somewhat.

How differently the three largest suppliers comment on the transition into the second half of the year is shown by a direct comparison – even though reporting periods and currencies are not identical:

Sources: Micron, 2026a; SK hynix, 2026; Samsung, 2026b (preliminary estimates).

In essence, this means for us: all signs point to further price increases, and we expect a reasonably stable price corridor at a new, significantly higher level.

1. All signs point to further price increases

Despite differing reporting periods, currencies and accounting standards, all three suppliers paint the same picture: record margins, unbroken demand and an explicit expectation that price levels will not normalize beyond the current year. Any disagreement concerns the pace of further increases, not their direction.

2. A price corridor stable at a new, significantly higher level

There is little to suggest that the market will exit its new price corridor substantially on the downside. Demand remains broad and structural: AI, inference, hyperscaler expansion, traditional enterprise workloads, SSD migration, HPC and sovereign infrastructure projects are all taking effect simultaneously. On the supply side, cleanroom capacity, technology transitions, HBM prioritization, NAND-to-DRAM trade-offs and multi-year customer agreements remain limiting factors. Micron expects DRAM and NAND demand to continue to significantly exceed supply, with tightness persisting beyond calendar year 2027.

For server DRAM, high-capacity RDIMMs, enterprise SSDs and certain flash components, what is decisive is therefore not nominal global bit production, but effectively available, qualified and deliverable product. This distinction is precisely where the core of the current market situation lies.

A temporary easing of individual prices is possible. A broad return to the price logic of 2023, however, is not the base case. Both a massive short-term capacity boost and a clear decline in demand in the relevant data center and enterprise segments are currently absent.