New OECD data shows diverging trajectories between public and private investment in research and development (R&D).

Government R&D budgets across OECD countries fell by 2,7% in inflation adjusted (real) terms in 2025, marking a second consecutive year of decline. By contrast, R&D spending by the largest corporate R&D investors increased in 2025 and accelerated sharply over the first half of 2026, driven by firms in digital industries and their supply chains. R&D spending grew by 13,9% year-on-year in real terms in the first half of 2026, up from 5,6% in 2025, and led by companies in software and computer services (23%), and IT hardware and electronics (20%).

 

Outside digital industries and their supply chains, corporate R&D growth remains weak

New data from the OECD Short-term Financial Tracker of Business R&D (SwiFTBeRD) – a panel tracking the quarterly R&D reports of 60 of the largest global R&D investors – show that the growth in corporate R&D in 2026 was concentrated in a few sectors.  Beyond software and computer services, and IT hardware and electronics, growth was much more modest: R&D spending by life sciences firms rose by 2% year-on-year in the first half of 2026, while R&D spending by firms in automotive and aerospace declined for the fourth consecutive semester.

At company level, Nvidia, Meta and Samsung Electronics recorded the highest year-on-year growth in nominal R&D spending among the SwiFTBeRD panel in the first half of 2026, reflecting strong momentum in IT hardware and software R&D linked – as implied in corporate reports and disclosures – to the development and deployment of frontier AI models.

Several other fast-growing companies operating in this area have seen growth in R&D levels approaching them to potential future inclusion in the SwiFTBeRD panel.

 

Government R&D budgets fall below 2020 levels

Data from the OECD Research and Development Statistics (RDS) database, based on national official statistics, show that the 2,7% fall in government budget allocations for R&D (GBARD) in 2025 followed a 3,7% reduction in 2024, bringing inflation-adjusted OECD-area R&D budgets in 2025 below their 2020 levels.

Cuts were widespread but varied in size, from -7,8% in the US and -3,7% in Japan to -0,9% in Germany.

Across the European Union, aggregate R&D budgets fell by 0,7%, as reductions in several of its larger economies more than offset strong growth in several other countries. This estimate excludes the R&D support programmes covered by the EU budget, currently not covered in GBARD statistics.

 

Share of government R&D targeted at specific policy goals falls

At the same time, a smaller share of government R&D budgets was directed towards supporting specific government functions such as health, defence and energy as opposed to the goal of advancing general knowledge through cross-cutting research funding councils and unearmarked public funds that universities use for R&D.

Public R&D funding is identified as directional when is targeted to serve an underlying policy goal. Across the OECD area, policy goal-targeted (ie. directional) support for R&D declined from 68,0% of GBARD in 2024 to 66,4% in 2025, and across the EU area from 45,4% to 44,2%.

Directionality is significantly lower when R&D tax incentives are taken into account.

By convention, these incentives are excluded from GBARD, and since they rarely prioritise R&D projects relevant to a goal compared to another, they are not directional. Adding the cost of R&D tax incentives (GTARD) to the total cost of public support in GBARD results in a significantly lower estimate of directionality of 55,6% for the OECD area in 2024, the latest year for which GTARD data are available.

Longer-term data shows that R&D tax incentives, which represented 18% of total OECD governments’ support for R&D in 2024, up from 6% in 2000, have been a major contributor to the decline in the directionality of government R&D support over the past two decades.