US Semiconductors: Read Through From UBS AI Software Conference: Open-Source Models Momentum

Research evidence excerpt

US Semiconductors: Read Through From UBS AI Software Conference: Open-Source Models Momentum

Global Research

17 July 2026ab

US Semiconductors Equities

AmericasRead Through From UBS AI Software

Conference: Open-Source Models Momentum Semiconductors

Timothy Arcuri

Analyst

timothy.arcuri@ubs.com

Summary +1-415-352 5676

UBS hosted the Private Software AI Companies Conference this week and the key Natalia Winkler, CFA

takeaway was continued strong AI demand, robust revenue growth expectations, and Analyst

accelerating adoption of open-source models. As enterprise use cases move from testing natalia.winkler@ubs.com

into production and large-scale deployment, organizations are becoming increasingly +1-415-352 4626

focused on optimizing token costs and maximizing ROI—particularly as some vendors

(such as MSFT Copilot) transition some clients from seat-based licensing to

consumption-based pricing. We do not view this as reducing demand. Rather, we think

demand is increasingly being reallocated toward the most cost-effective models and

workloads. The focus on optimization is also driving adoption of open-source models,

which have narrowed the performance gap versus frontier solutions over the past year

and are increasingly being deployed for specific steps within agentic workflows and

consumer-facing applications. We see this trend as particularly favorable for NVDA given

its software leadership in open-source model development, the Nemotron family, and

the fact that most open-source models continue to be trained and fine-tuned on NVDA

hardware.

Details

AI demand remains exceptionally strong and appears to be broadening rather than

slowing. Nearly every company pointed to expanding usage despite increased

scrutiny around ROI.