Nvidia survey shows AI driving revenue, cuts & spend

Nvidia has published its latest State of AI findings, pointing to wider use of artificial intelligence in business operations and a stronger focus on financial returns.

The annual surveys drew more than 3,200 responses worldwide across financial services, retail and consumer packaged goods, healthcare and life sciences, telecommunications, and manufacturing. The results suggest many organisations have moved beyond pilots and assessments and are now deploying AI at scale.

Across all respondents, 64% said their organisations actively use AI in operations. Another 28% remain in an assessment phase, while 8% reported no AI use and no plans to start.

Adoption varied by region. North America recorded 70% active usage, with 27% still assessing projects and 3% not using AI. In EMEA, 65% reported active use. Respondents in APAC reported 63% active use, while 15% said they do not use the technology.

Larger organisations reported higher usage and broader deployment. Among companies with more than 1,000 employees, 76% said they actively use AI and 2% said they do not use it at all. Smaller organisations reported lower adoption rates and placed greater emphasis on open source tools in their AI strategies.

From pilots

The results suggest a shift in how organisations manage AI programmes. Across most industry surveys, the share of respondents reporting active AI use increased, while the share in assessment declined. The findings indicate that internal governance, operational integration, and data readiness now play a larger role in determining whether projects progress.

Financial services showed strong engagement. The surveys describe AI being applied to workloads involving large volumes of text, numeric data, and documents. Nasdaq has built an AI platform for internal operations and external products.

Nasdaq outlined how it sees the technology's role in the business.

"At Nasdaq, we are a technology platform company, and AI has the ability for us to unite all the different businesses and products," said Michael O'Rourke, Senior Vice President and Head of AI and Emerging Technology at Nasdaq. "AI will help bring together data from all our businesses and technologies, and help us build better products and services."

Productivity focus

Operational efficiency and productivity ranked as the most common objectives for AI programmes. Creating operational efficiencies was the top goal for 34% of respondents, followed by improving employee productivity at 33%. Opening new business opportunities and revenue streams ranked third at 23%.

On operational impact, 53% said improved employee productivity was among AI's biggest effects on their operations. Another 42% pointed to operational efficiencies, and 34% reported new business and revenue opportunities.

Telecommunications stood out for reported productivity benefits. In the State of AI in Telecommunications results, 99% said AI improved employee productivity, with a quarter describing the improvement as major or significant.

Digital twins

Manufacturing and retail examples highlighted the use of simulation and 3D modelling in AI projects. Siemens has integrated AI into tools and applications used in industrial workflows, according to the survey narrative.

PepsiCo worked with Siemens and Nvidia on digital twins for selected US manufacturing and warehouse facilities. The project uses Siemens' Digital Twin Composer to recreate machines, conveyors, pallet routes, and operator paths. It combines physics-level simulation with AI agents for scenario testing.

Reported outcomes include a 20% increase in throughput in initial deployments. PepsiCo also reported faster design cycles, nearly 100% design validation, and 10-15% reductions in capital expenditure.

In retail, the survey referenced Lowe's use of AI-driven digital twins across more than 1,750 stores. The retailer also used AI for asset discovery and 3D model generation, creating 3D models from 2D product images at a reported cost of less than USD $1 per model.

Revenue and costs

The results place return on investment at the centre of enterprise AI decision-making. Overall, 88% said AI increased annual revenue in some or all parts of the business. Of those, 30% reported increases greater than 10%, 33% reported increases of 5-10%, and 25% reported increases of less than 5%.

Executives reported higher levels of revenue impact than the overall sample. A little over 40% of C-suite or vice president respondents reported annual revenue increases of more than 10%.

Cost reduction was similarly widespread. Overall, 87% said AI reduced annual costs, with 25% reporting decreases greater than 10%. Retail and consumer packaged goods had the highest share reporting cost reductions above 10%, at 37%.

Agentic AI

The surveys also tracked emerging interest in AI agents, described as autonomous tools that can reason, plan, and execute complex tasks based on high-level goals. Data collected from August through December 2025 showed 44% of companies either deploying or assessing agents last year.

Telecommunications reported the highest adoption of agentic AI at 48%, followed by retail and consumer packaged goods at 47%. Early 2026 deployments covered functions such as code development, legal and financial tasks, and administrative work.

One example cited was Mona by Clinomic, described as an onsite assistant for intensive-care units. The product consolidates, analyses, and visualises patient data in real time. Reported outcomes include a 68% reduction in documentation errors and a 33% reduction in perceived workload for clinical-care professionals.

Open source

Open source software and open-weight models featured as a central theme in AI strategy. Overall, 85% said open source was moderately to extremely important to their organisation's AI strategy, including 48% who rated it as very to extremely important.

Smaller companies reported greater reliance on open source. Among smaller organisations, 58% said open source was very to extremely important. Just over half of executives, at 51%, also rated it as highly important.

Budgets rise

Most respondents signalled higher spending in 2026. Overall, 86% said their AI budget will increase and 12% said it will remain unchanged. Nearly 40% expected budgets to rise by 10% or more.

North America showed the strongest expectation of larger increases, with 48% forecasting budget growth of at least 10%. Among executive-level respondents, 45% said the same.

Respondents prioritised optimisation and expansion of existing work. The top spending priority for 2026 was optimising AI workflows and production cycles (42%). Finding additional use cases followed at 31%, as did building and providing access to AI infrastructure, including on-premises data centres and cloud resources (31%).

Skills gap

The findings also highlight barriers to scaling AI. Data challenges ranked first, with 48% citing insufficient data or other data-related issues as a top challenge.

A shortage of specialists followed. Lack of AI experts and data scientists was a top challenge for 38% of respondents. Measuring business impact also remained difficult for some organisations, with 30% citing a lack of clarity on AI's return on investment as a top challenge.

Respondents came from a mix of roles: C-suite and vice presidents (27%), directors and managers (33%), and AI practitioners (40%). Nvidia expects organisations to direct increased budgets towards optimising deployed systems and extending AI use across more functions during 2026.