YMTC’s 70% Domestic-Tool Target Is Not New NAND Supply Yet

YMTC’s planned fourth Wuhan NAND fab targets up to 70% domestic equipment, according to Reuters reporting summarized by AI Weekly. The report describes at least 60% as the intended domestic share, compared with just over 50% at the third fab. Supplier discussions remain preliminary, with no formal orders reported.

The facility is designed for approximately 100,000 wafers per month and more than 300 NAND layers. Construction is expected in 2027. These are planning and design figures, not a statement that equipment is installed or memory is available for purchase.

For an AI infrastructure buyer, the news belongs in supply-chain monitoring rather than an immediate storage-cost revision. Equipment sourcing can change how a factory is built without establishing when a particular customer will receive qualified products.

Domestic sourcing is a target with an unfinished bill of materials

The reported move from just over 50% domestic equipment at the third fab toward at least 60%, and potentially 70%, at the fourth describes a sourcing ambition. It does not establish a completed purchase plan. The explicit absence of formal orders limits what a buyer can infer about supplier selection or equipment delivery.

Lithography and inspection remain foreign in the reported plan. That qualification matters because a domestic share is not the same as independence from foreign tools. A procurement discussion should preserve the remaining dependencies rather than compress the story into a claim that the fab can operate entirely on domestic equipment.

The available facts also do not explain the denominator used to measure equipment share. A buyer should request that definition before treating the percentage as a precise measure of technical independence. Equipment count, spending and production importance should not be silently assumed to describe the same thing.

Design capacity does not identify a delivery date

Approximately 100,000 wafers per month is the reported design capacity. It is not reported commissioned output, qualified product availability or a commitment to a named buyer. Capacity planning needs those distinctions because a factory specification and an executable purchase order answer different operational questions.

Construction expected in 2027 is likewise not a disclosed date for customer shipments. The supplied reporting does not establish a commissioning schedule or a production ramp. Converting the construction expectation into a delivery promise would erase the uncertainty that should be visible in an infrastructure plan.

More than 300 NAND layers describes a technical characteristic of the planned manufacturing capability. It does not supply a supported estimate for finished storage prices, usable inventory or products a specific buyer will be able to qualify. Those decisions require evidence closer to the eventual product and transaction.

The trade-off is continued dependency during a sourcing change

Increasing domestic equipment content can be significant while still leaving critical purchasing questions unresolved. The reported foreign lithography and inspection requirements mean the sourcing plan should not be read as a clean break from outside supply. The evidence does not establish that remaining dependencies are immaterial to commissioning.

Nor does the target demonstrate a delivery problem. The report describes preliminary supplier discussions, not a failed installation or a quantified disruption. Risk analysis should avoid inventing an outage just as carefully as it avoids inventing available capacity. A target is a reason to monitor the dependency, not proof of either success or failure.

For a buyer, the cost of uncertainty is maintaining a distinction between strategic options and committed supply. A future fab can remain relevant to sourcing strategy without receiving credit in the current operating plan. Do not reduce the contingency assigned to existing suppliers merely because an additional design-capacity figure has entered the news.

Market share is context, not a customer allocation

The report cites Omdia figures putting China’s Q2 NAND share at 38.5%, versus 28% in 2025. That comparison supplies market context for the factory plans. It does not identify who receives the output, which products meet a buyer’s requirements or whether a particular supply agreement becomes easier to obtain.

The share figures also should not be substituted for a commissioning measure. A market position can change while a separate proposed facility remains unbuilt. Combining them into a narrative of immediately available additional supply would attach an operating conclusion to figures that describe different subjects.

A useful monitoring note should retain each claim in its original category. Record the sourcing percentage as a target, construction as an expectation and wafer throughput as a design specification. Keep the Omdia share comparison in the market-context field. The separation makes later updates easier to interpret without rewriting the original reporting.

What To Do About It

• Label the capacity correctly: Keep planned factory throughput outside current available inventory and identify it explicitly as design capacity of approximately 100,000 wafers per month.
• Track an order milestone: Ask suppliers for evidence of formal equipment orders before treating preliminary sourcing discussions as a completed plan for up to 70% domestic equipment.
• Request commissioning evidence: Separate construction expectations from qualified customer shipments rather than using a presumed production start in 2027.
• Preserve the dependency map: Keep foreign lithography and inspection visible in the sourcing review instead of treating the target as complete independence at 70%.
• Keep pricing evidence local: Ask for actual product quotations and delivery commitments rather than projecting a storage-price change from the cited Q2 share of 38.5%.

The Bottom Line

YMTC’s reported 70% domestic-tool target is a supply-chain signal, not a statement that new memory is ready to ship. Approximately 100,000 wafers per month remains design capacity, and construction expected in 2027 does not establish a delivery date. Buyers should track formal orders and commissioning while preserving the reported foreign-tool dependencies. Change a purchasing plan when supply evidence changes, not when a factory target appears.