December 5, 2025

Aidan Gomez, CEO of Toronto‑based startup Cohere, argued that the United States and Canada are better positioned than China to lead the global AI race. While Chinese companies have rolled out impressive models, Gomez believes the key advantage lies in commercializing the technology at scale. He notes that liberal democracies prefer to rely on domestic or allied suppliers for critical infrastructure and that Western firms are investing billions to build AI infrastructure. Gomez also dismissed apocalyptic “Terminator” narratives about AI and warned that returns on investment drop as models get bigger. His comments highlight the strategic choices facing AI startups and policymakers.

Full Article:
The artificial intelligence arms race has often been framed as a sprint to build the biggest, most powerful models. In this narrative, whichever country releases the first artificial general intelligence will dominate the future. Aidan Gomez, co‑founder and CEO of Cohere, offers a different perspective. Speaking at Reuters NEXT, Gomez acknowledged that Chinese companies such as DeepSeek and tech giants like Alibaba and Baidu have produced state‑of‑the‑art language models. However, he argued that mere technical prowess isn’t enough. “It’s not who gets the technology first,” he said, “but who commercializes it at scale.” In his view, the United States and Canada hold a decisive advantage because they are trusted partners for economies around the world, whereas many liberal democracies are hesitant to build critical infrastructure on Chinese technology

Gomez’s stance contrasts sharply with the widely reported view of Nvidia CEO Jensen Huang, who warned that China is “nanoseconds behind America” and could win the AI race. Gomez acknowledges the rapid progress of Chinese labs but stresses that adoption is shaped by geopolitics as much as performance. Countries concerned about national security, privacy and supply chains may choose to work with U.S. or Canadian firms even if Chinese models are technically comparable. This preference gives Western startups like Cohere an opportunity to become trusted suppliers of enterprise AI, particularly in sectors such as finance, healthcare and government.

The CEO also emphasized that building bigger models isn’t always better. “Spending an incremental $10 billion a year to improve your model does not deliver the return on investment … over the past few years,” Gomez said. As models scale, improvements in accuracy and capability taper off, while the costs of training and running them skyrocket. Meanwhile, enterprise customers demand reliability, security and ease of integration, attributes that don’t necessarily correlate with model size. Cohere’s strategy has been to build specialized models tuned for specific industries rather than chasing benchmark records. For entrepreneurs, this suggests there is room for smaller, leaner AI systems that serve defined niches more efficiently than general‑purpose behemoths.

Another point Gomez raised is the importance of infrastructure. To maintain U.S. leadership in AI, American companies are spending billions on data centers, custom chips and fiber networks. This investment isn’t just about capacity; it’s also about resilience. By owning the infrastructure that powers AI applications, U.S. firms can ensure that supply chains remain under friendly control. The U.S. government’s restrictions on exporting advanced chips to China underscore how hardware access shapes AI competition. Entrepreneurs building AI products should pay attention to supply chain dynamics and consider partnerships that secure compute resources.

Gomez also touched on the role of liberal democracies. He argued that democracies around the world prefer to rely on allied technologies for critical infrastructure. That means startups operating in these ecosystems may find more receptive markets if they emphasize transparency, privacy and compliance with democratic norms. For example, Cohere offers its models through cloud providers that comply with data sovereignty laws, making it an attractive option for governments and companies concerned about data location.

Addressing fears about runaway AI, Gomez expressed skepticism about doomsday scenarios. “I personally don’t believe a lot of these stories of ‘Terminators’ and doomsdays,” he said. Such narratives have lost popularity as people interact with real AI systems and see their benefits. However, he acknowledged that tech investors are increasingly demanding better returns on AI investments and greater transparency about risks. For founders, this means that building trust and delivering tangible business value may be more important than hyping speculative future capabilities.

From an entrepreneurial standpoint, Gomez’s remarks contain several strategic insights. First, focus on commercialization: the real battle isn’t to build the biggest model but to deliver useful solutions at scale. Second, align with geopolitical realities: businesses that offer trustworthy, compliant AI services may win contracts even against technically superior rivals. Third, control your infrastructure: secure compute and data resources are a competitive advantage. Fourth, resist the temptation to overspend on model size; incremental improvements may not justify the cost. Lastly, cultivate a sober narrative about AI, one that acknowledges both its potential and its limitations, because customers and regulators are increasingly wary of hype.

In the years ahead, the AI race will likely be less about sprinting to the finish and more about building a robust ecosystem of technology, trust and talent. For future billionaires, the winners may not be those who brag about nanoseconds but those who think deeply about where value is created—and who deliver it consistently to their partners and customers.

 Subscribe to our newsletter to stay up to date with the latest breakthroughs in AI, Business, Technology, and Mindset.