# How AI Is Reshaping the Global Semiconductor Patent Landscape > Source: > Published: 2026-09-01 16:32:49+00:00 The demands of AI workloads are pushing conventional chip designs up against limits in thermal management and memory bandwidth, according to a new report published today by IP management provider Anaqua. In response, legacy chipmakers and cloud hyperscalers are accelerating semiconductor patent filings around custom silicon, next-gen memory, and unconventional computing setups just to keep up with AI’s massive compute appetite. ### Macro trends and the timeline lag If current trends hold, AI could be shifting from a temporary tailwind to the central driver of long-term semiconductor R&D investment. Over the past five years, patent filings sitting directly at the intersection of AI and semiconductor increased by 114%, outpacing the broader combined sector’s 78% growth, according to Anaqua’s study. [View All](https://www.eetimes.com/category/sponsored-content/) “The numbers clearly show that the AI [and] semiconductor intersection is the area with the largest growth of patenting within the semiconductor industry,” Toni Nijm, chief product officer at Anaqua, told EE Times. “We believe that AI semiconductor patenting will continue to have an outsized impact over general semiconductor patenting.” Because of the typical 18-to-30-month publication lag at patent offices, the sharp 29% filing spike in the middle of this cycle directly mirrors the frantic boardroom decisions of the 2021 to 2022 chip crisis, when global shortages coincided with massive state subsidy rollouts such as the U.S. and EU Chips Acts. While annual growth has since slowed to 2%, it’s important not to mistake that for cooling interest. Instead, it signals that the market is settling into a highly funded steady state as patent offices work through their backlogs. “While growth numbers did plateau, it’s hard to calculate exactly why,” Nijm said. “It is probably due to a mixture of patent office backlogs as well as a focus on more specific inventions within each of the key AI [and] semiconductor sub-categories.” ### Silicon giants battle for dominance While the macro trends show an industry-wide rush to fund AI hardware, zooming in on individual patent portfolios reveals that the biggest players are taking vastly different paths to secure their dominance. “There are a number of companies that are well positioned, from a patenting perspective, to address key areas of growth within the AI infrastructure buildout,” Nijm said. “The report analyzed key categories where patenting investment is being made, as well as which companies are making those IP investments within the key subcategories.” Nvidia currently dominates data centers worldwide, yet it ranks seventh in GPU patents with 49 filings, and 15th in inference, with 501 filings, according to the Anaqua report. This apparent disconnect highlights a crucial reality: Nvidia’s true moat isn’t purely silicon; it lies in the proprietary CUDA software ecosystem it first built in 2006. On the other hand, Samsung is taking advantage of its expertise in memory to take the top spot in AI architecture filings with 1,194 patents, a 176% jump, while maintaining dominance in HBM and ranking second in inference, with 701 filings, according to Anaqua’s study. “The inference category is clearly where the largest growth is happening,” Nijm said. “This is likely due to the shift from training AI models to operating them at much higher levels of efficiency.” Meanwhile, Intel is mounting its own multi-front defense, leading all rivals with 188 GPU-titled filings for its Arc and data center platforms while maintaining a massive architectural footprint. Qualcomm isn’t far behind, expanding rapidly from on-device smartphone silicon straight into the enterprise data center, posting a 186% surge in architecture filings alongside AI200 and AI250 accelerators. ### The China surge Washington’s export controls were designed to starve China’s technology sector of advanced hardware, but Anaqua’s data illustrates an unintended backlash: a massive, state-backed race to develop domestic alternatives. With 5,387 filings over the past five years, Chinese government-affiliated entities, universities, and research institutes have effectively tripled the volume of their closest corporate rivals. ### Challengers, hyperscalers, and the frontier Beyond these state-driven geopolitical maneuvers, the broader competitive landscape is being reshaped from two sides: cloud giants designing their own silicon and deep-science innovators reimagining compute from the physics up. Data shows that hyperscalers are steadily cutting out merchant chip suppliers. For example, Alphabet is entrenching its custom TPU architecture into the top three for AI architecture filings, while Microsoft advances its in-house Maia accelerators. Meanwhile, IBM leads the charge in analog and neuromorphic computing with its phase-change memory and Spyre accelerator in a bid to address the von Neumann bottleneck. ### Industry outlook Anaqua’s data confirms that semiconductor IP is no longer just a routine corporate safeguard; it has become central to geopolitical leverage and technology dominance. And as standard chips start bumping up against hard limits in power and memory, the next chapter of AI won’t just be about cramming more transistors onto a die. Hyperscalers are cementing their vertical integration with bespoke silicon, while traditional semiconductor titans are racing to bring memory and processing logic closer together, and sovereign nations are building substantial IP portfolios. For hardware and software architects alike, the trend is clear: AI performance increasingly depends on the integration of silicon, memory, and specialized algorithms. You can download the full report [here](https://go.anaqua.com/2026-semiconductor-patent-report). ##### Also read: [Game, Set, Bot: Sony AI’s ‘Ace’ Serves Up a Defeat to Table Tennis Pros](https://www.eetimes.com/game-set-bot-sony-ais-ace-serves-up-a-defeat-to-table-tennis-pros/) [Synopsys Bets on AI-Enabled Ecosystem to Power Automotive Digital](https://www.eetimes.com/synopsys-bets-on-ai-to-power-automotive-digital-twins/) [Innatera Taps VLSI Expert to Lower Barriers to Neuromorphic Adoption](https://www.eetimes.com/innatera-taps-vlsi-expert-to-lower-barriers-to-neuromorphic-adoption/) [AI’s Booming Demand Meets a Semiconductor Reality Check](https://www.eetimes.com/ais-booming-demand-meets-a-semiconductor-reality-check/)