Gary Gensler, former chairman of the U.S. Securities and Exchange Commission, has recently issued a caution: American artificial intelligence companies are facing significant challenges. On one hand, AI models do pose real safety risks—OpenAI’s models, for example, breached the security defenses of technology firm Hugging Face earlier this year, underscoring that greater AI capabilities increase the potential for unintended consequences. On the other, China’s rapid advancement in AI development is intensifying competitive pressure on U.S. firms. Gensler warns that even if the United States currently leads in AI, future American companies may shift toward using Chinese-developed models due to cost and efficiency considerations.
Compounding the issue is the lack of consensus within U.S. political circles regarding AI risks. Some industry voices have called for slowing the pace of development in the most advanced models and advocated for international cooperation to establish safety standards. However, President Trump has rejected proposals to slow innovation and opposes international agreements aimed at regulating AI. He has even dismissed concerns about AI as a “hoax.” Gensler explicitly stated his disagreement with this perspective.
He also highlighted a practical concern: U.S. AI firms are investing heavily, but it remains uncertain whether revenues will keep pace. The industry is currently pouring substantial capital into infrastructure, yet revenue growth has not matched the scale of investment. Should future returns fall short, the massive expenditures on data centers, chips, and computing power could become unsustainable liabilities.
Gensler’s remarks touch on three particularly awkward realities underlying the current U.S. AI boom.
First, safety cannot be resolved through rhetoric alone. As AI systems grow more capable, they become more likely to breach security boundaries; as they become more open, they become more susceptible to misuse. Yet regulatory frameworks remain underdeveloped, while competitive pressures compel companies to move forward regardless. Those who slow down risk falling behind; those who don’t slow down collectively accumulate systemic risk.
Second, the intensifying U.S.-China AI rivalry complicates efforts to govern AI safely. In the past, discussions on AI safety were largely technical in nature. Today, they must also incorporate geopolitical, industrial, and national security dimensions. Firms fear being outpaced by rivals, governments worry about losing strategic control, and thus prudence is often labeled as conservatism, while regulation is frequently portrayed as self-imposed limitation.
Third—and most pragmatically—the reality is that AI can generate profits, but many current investments resemble high-stakes gambles. Data centers grow larger, chips are procured in ever-greater quantities, debt accumulates, yet revenue streams have not yet materialized at comparable levels. If future demand fails to meet expectations or lower-cost, sufficiently capable models emerge, today’s expensive infrastructure may transform from a competitive advantage into sunk costs.
Trump’s assertion that AI concerns are a “hoax” may sound resolute, but markets do not rely solely on confidence. Gensler’s warning does not imply an imminent collapse of the AI sector, but rather signals that the competition has transitioned from “who is more advanced” to “who is more sustainable.” The true test for U.S. AI lies not only in model performance, but in whether safety, cost efficiency, and commercial viability can all be achieved simultaneously.
Original source: toutiao.com/article/1877715879080969/
Disclaimer: The views expressed in this article are those of the author.