Investors put about $560 billion into startups in the first half of 2026, more than in the whole of 2025. And most of it went to a single technology: AI took 77% of global venture capital in those six months.
The figures come from WIPO’s Global Innovation Index 2026 (GII), published in September. At first glance, the report looks like a story of stability. Switzerland, Sweden, the United States, the Republic of Korea and Singapore top the ranking again, just as they did last year.
Look closer, though, and a lot is moving. The money is concentrating, research is at a record high, and a new generation of science-based startups is building its business on patents. All of this shows up in patent data, often years before it shows up in products.
Why the GII is worth reading as an IP professional
The GII ranks nearly 140 economies on how well they innovate, looking at everything from R&D spending and research output to high-tech exports and IP licensing. It’s one of the most widely used benchmarks of its kind and is independently audited by the European Commission.
Patent data plays a central role. WIPO’s ranking of the world’s top 100 innovation clusters is partly based on where the inventors behind PCT applications (international patent applications filed through WIPO’s Patent Cooperation Treaty) are located. WIPO uses patents to see where innovation happens, and IP teams can use the same data to see where their next competitors are coming from.
More money, fewer bets
According to the report’s Global Innovation Tracker, venture capital grew 28% in 2025, yet the number of deals fell for the fourth year in a row. Investors are writing bigger cheques to fewer companies, and a growing share of that money goes to AI.
Companies are spending more on research too. Corporate R&D reached a record $1.5 trillion in 2025, driven largely by software and AI companies, along with defence-related research.
Well-funded AI companies aren’t staying in their own lane. They’re applying their technology to manufacturing, medical devices, vehicles and much more. If you only monitor your traditional competitors, you may miss some of the most relevant new filings in your field.
Research is booming, international filing less so
Scientific publishing hit a new record in 2025, but international patent filing grew only slightly: PCT applications rose by less than 1%. Behind that flat total, the balance is shifting. Filings from China kept growing, while filings from the United States fell for the third year running.
Patenting isn’t slowing down everywhere. It’s moving between countries and concentrating in specific technologies, which makes it more important to know where to look.
Deep-science startups are built on patents
This year’s GII takes a close look at deep-science startups: companies turning frontier research in fields like semiconductors, robotics, quantum computing, life sciences and advanced materials into products. WIPO counts more than 30,000 of them founded since 2000.
The most interesting finding for IP teams is how much these companies rely on patents. About half of deep-science startups hold patents, compared with just 15% of other startups. They also file more as they grow, and most of them use the PCT route to seek protection in several countries.
The reason is simple. These companies can wait five to fifteen years between invention and revenue, and during that time their patents are often the main asset they can show investors and partners. That’s why they tend to appear in patent data long before they have a product on the market.
The map is getting wider
China and the United States host nearly half of the world’s top 100 innovation clusters. Europe holds its own with 30, led by Germany. Sweden, ranked second overall in the GII, also saw Stockholm move into the top 30 clusters this year.
At the same time, newer innovation economies such as India, Viet Nam, Türkiye and Indonesia have climbed the rankings quickly over the past decade, and deep-science startups are growing fastest outside the traditional centers. As a result, more relevant prior art and more competing claims are being published in languages and patent offices that many IP teams rarely search.
What this means for your IP strategy
None of these trends call for a completely new approach. But together they show where many monitoring setups have blind spots, and where a few changes can make a real difference.
Rethink who your competitors are
Most competitor lists are built around the companies you meet in the market today: the ones you bid against, see at trade fairs and read about in industry news. The GII suggests that list is getting less complete.
AI companies with large budgets are moving into new industries, and deep-science startups are developing technologies that may compete with yours years from now. Neither group is likely to be on your radar yet. It’s worth reviewing your watch list at least once a year and asking who is filing in your technology areas, not just who you compete with today.
Monitor technologies, not only company names
Monitoring by company name only finds the companies you already know about. Monitoring by technology, using keywords or classification codes such as IPC and CPC (the international systems patent offices use to group inventions by subject), picks up filings from anyone working in your field.
The two approaches work best together. Name-based monitoring keeps you up to date on known competitors, while technology-based monitoring shows you who is new. When an unfamiliar applicant appears several times in your technology monitoring, that’s a signal worth looking into.
Use patents as an early warning system
Products are the last stage of innovation. Patent filings come much earlier, which makes them one of the best ways to see where a technology or competitor is heading.
That’s especially true for deep-science startups. They often patent years before they launch, so their filings show what they’re working on long before the market does. Spotting them early gives you options: you can assess the risk to your own products, adjust your R&D direction, or start a conversation about licensing or partnership while it’s still on your terms.
Look beyond your usual jurisdictions
Many teams focus their searches on the European and US patent offices, often because those filings are in languages they can read. As innovation spreads to more countries, that habit leaves a growing gap. Important filings now appear in Chinese, Korean, Portuguese and many other languages.
Including these offices in your searches and monitoring doesn’t have to mean hiring specialists. Machine translation makes foreign-language filings readable enough to judge whether they’re relevant, and only the important ones need a closer professional look.
Plan how your team will handle the volume
Wider monitoring means more documents to review. Without a clear process, that quickly becomes overwhelming. Agree on who reviews what, how relevance is decided, and how findings are shared with R&D and management, so new information leads to decisions rather than piling up in inboxes.
How IamIP can help
IamIP is built for this kind of work. With weekly competitive monitoring, you can track technologies as well as named companies, read filings from any country with built-in machine translation, and use the AI Patent Categorizer to focus your review on what’s relevant as volumes grow. Shared portfolios and workflows make it easy to bring R&D, IP and management into the same process.
The GII shows where innovation is heading, and patent filings are often the first place that direction becomes visible. The sooner your team sees it, the more time you have to respond.
Related reading: Patent Monitoring Isn’t Optional—Here’s Why – IamIP