Can U.S. Companies Maintain Their Lead in AI innovation?

Advanced artificial intelligence systems have the potential to contribute $15.7 trillion to the global economy by 2030, according to PwC estimates, and the race is on to develop them. While U.S. companies like OpenAI, Google, and Anthropic lead the world in the development of new AI models that could address highly complex challenges in many different industries, their Chinese competitors are closing the gap, according to three recently published reports.

The U.S. is still ahead in the number of significant AI models released. According to the April 2025 Artificial Intelligence Index Report from Stanford University’s Institute for Human-Centered AI, U.S. institutions produced 40 notable AI models in 2024, compared to China’s 15 and Europe’s 3, with the vast majority of U.S. models coming from the U.S. corporate sector.

The gap is narrower with respect to performance. Chinese generative AI models are only three to six months behind their U.S. competitors, and that gap is closing, according to a May report by Insikt Group, the research and threat analysis division of Recorded Future.

In January of this year, Chinese AI startup DeepSeek generated headlines around the world when it launched R1 – a high performance, open source, large language model (LLM) – for a fraction of the cost of the latest U.S. models and requiring far less computational resources.

DeepSeek was just one of 302 generative AI service models registered in China as of January 2025, according to the Insikt report, from companies such as Alibaba, Baidu, and Infinigence AI. The report, Measuring the U.S.-China AI Gap, finds that Chinese companies are much more likely to use open source technology for their AI models than their U.S. counterparts, resulting in performance gains. These performance gains and a focus on cost efficiency at these companies have made them globally competitive. 

Competing AI Ecosystems

However, the development of new AI models is only one part of the competitiveness story. The current dominance of U.S. companies in AI is backed by the country’s elite research institutions, deep pool of talent, lead in semiconductor manufacturing, energy production, and AI computing infrastructure, as well as an abundance of private funding.

The latter is one area where the U.S. companies have a vast advantage over their international competitors. According to Stanford’s AI Index Report, U.S. private AI investment grew to $109.1 billion in 2024—nearly 12 times China’s $9.3 billion and 24 times the U.K.’s $4.5 billion. The U.S. also had over 10 times more newly funded AI companies than its nearest rival did in 2024. “When you look at what type of countries in the world actually lead on this investment, it’s the U.S. that’s far away ahead of other geographic areas like Europe and China,” said Nestor Maslej, research manager at Stanford’s Institute for Human-Centered AI, during a presentation of the report’s findings.

However, China’s AI ecosystem is also growing rapidly, supported by government-led investment initiatives, a growing pool of high-quality talent, the largest amount of published research globally, increasing links between Chinese universities and the AI industry, and advances in semiconductor manufacturing. The Chinese government’s plan to become the world’s leader in AI by 2030, launched nearly a decade ago in 2017, is gaining momentum.

The Race to Adoption

An equally important part of the competitiveness story is how successfully AI products are distributed to and adopted by consumers, businesses, and industries to the benefit of the national economy.

According to the Insikt report, it’s hard to determine which country leads on this front. Cost-efficient, open-source Chinese AI models are being adopted domestically and abroad, while more patents for AI applications are being filed in China than anywhere else, and in key industries, such as software, finance, and energy. General optimism about AI products is also higher in China. Some 83% of people think AI services are more beneficial than harmful in China, according to the AI Intelligence Index Report, compared to just 39% in the U.S.

However, the U.S. leads in other areas. More organizations use AI in North America than in China, according to the Stanford report, and the U.S. also dominates when it comes to the computing infrastructure used to run AI products, and access to data on which new AI models are trained.

Large tech companies in both China and the U.S. distribute new AI models to the large numbers of businesses and consumers that use their Software-as-a-Service platforms. In the U.S., for example, Amazon Web Services distributes Anthropic’s Claude AI models through AWS Bedrock, while Microsoft’s AI Azure Foundry distributes OpenAI models.

According to a report, published in May from Georgetown University’s Center for Security and Emerging Technology (CSET), the dominance of large tech companies potentially creates a risk for the U.S. AI industry. “Incumbents can effectively pick winners and losers in the AI market, potentially preventing disruptive upstarts and their inventions from reaching the market,” wrote senior research analyst Jack Corrigan, in Promoting AI Innovation Through Competition. “If left unchecked, this behavior could undermine the long-term innovation capacity and resiliency of the U.S. AI ecosystem.”

Challenges Ahead

The U.S could lose some of its research competitiveness should its ability to attract and retain top talent diminish. On the flip side, China’s AI industry also faces significant challenges. China’s government has imposed AI regulations for Chinese companies that are developing public-facing products, which can slow down development and deployment. China’s growing semiconductor industry is also unlikely to scale quickly enough to meet rapidly growing domestic demand for AI accelerator chips, according to the Insikt report.

The AI Outlook

What does this mean for the future competitiveness of AI companies in the U.S.? Based on an analysis of all the key drivers, including government and venture capital funding, industry regulation, talent, technology diffusion, model performance, and compute capacity (the computational resources required to train and operate AI models), Insikt Group believes that China is not likely to surpass the U.S. by 2030. In fact, it finds that if planned U.S. investments in semiconductors, energy, and AI computing infrastructure all go ahead, the computing gap in particular, will likely continue widening by 2030.

However, the report’s authors don’t consider U.S. dominance a foregone conclusion. “U.S.-China AI competition will almost certainly intensify, particularly in relation to generative AI and LLMs,” they write. “Unless U.S. companies embrace open source and drastically lower their prices, Chinese models will almost certainly see increased adoption worldwide while making continued improvements.”

What could shift this assessment, finds the report, would be either Chinese or U.S. companies developing new AI architectures and techniques that transform model training and performance, or developing agentic (autonomous) and collaborative AI systems with new capabilities, especially in robotics.

In the high-stakes race to develop the next generation of AI models, U.S. companies have much at risk.