Semiconductor equipment demand is rising even as worries about an artificial intelligence bubble curb the industry's broader enthusiasm, DigiTimes reports. The finding puts tool makers in an unusual position. Their order books improve while the customers funding the AI build-out face sharper questions about returns.

The report frames 2025 as a year of two opposing forces. On one side, chipmakers continue to buy deposition, lithography and etch systems to serve AI-related demand for advanced logic and high-bandwidth memory. On the other, investors and analysts increasingly ask whether the capital being committed to AI infrastructure will generate matching revenue.

DigiTimes does not publish full shipment or revenue forecasts in the summary of its reporting. The core claim is directional: equipment demand is up, and AI-related risk is now a visible brake on sentiment rather than on orders, at least for now.

Why does equipment demand hold up?

Equipment orders tend to lag end-demand by several quarters. Foundries and memory makers commit to tool purchases long before finished chips reach the market. That lag explains why equipment makers can see rising demand even when commentators warn about overinvestment in AI compute.

Capacity already under construction must be equipped regardless of how the demand debate resolves. A fab shell without tools produces nothing. Once a construction milestone passes, the tool orders follow.

This dynamic has played out before. In prior cycles, equipment makers kept booking orders for two to three quarters after analysts flagged demand risk, because committed fab projects could not be halted cheaply.

What risks does the report identify?

According to DigiTimes, the risks center on the sustainability of AI spending. The concern is not that AI demand has stalled. The concern is that the pace of investment may outrun the revenue the AI ecosystem can actually generate.

If that gap widens, the consequences would move through the chain in sequence:

• Cloud providers and AI developers would slow capital spending first.
• Chip designers would then revise order volumes with foundries.
• Foundries and memory makers would defer capacity expansions.
• Equipment makers would see cancellations and push-outs last.

The equipment segment sits at the end of that transmission line. That position protects near-term orders and exposes long-term ones.

How should readers weigh the claim?

The DigiTimes report reflects supply-chain reporting from within the Taiwanese electronics ecosystem, where much of the world's advanced foundry capacity sits. Its reporting on equipment demand frequently draws on order and shipment activity among suppliers to the leading foundries.

Readers should distinguish two categories of information in this story. The reported rise in equipment demand reflects current commercial activity. The assessment of AI bubble risk is a forward-looking judgment about how long that activity can continue.

Neither element in the available reporting carries a specific capacity figure, deal value or shipment number. Any precise forecast attached to this story would come from other sources, and this report does not cite them.

What comes next?

The commercial question is whether AI-related equipment orders keep rising through the next two quarters. If cloud capital spending holds, tool makers extend their run. If spending cracks, the equipment segment will feel it last, but it will feel it.