Henry Paulson on AI Risk, U.S.–China Relations and Crisis Leadership

Jim Esposito introduces Henry “Hank” Paulson, former Goldman Sachs CEO and U.S. Treasury secretary, at Citadel Securities’ Future of Global Markets event. Esposito recalls watching the 2008 crisis from Goldman and thanks Paulson for his public service.

Paulson thanks Esposito and praises Ken Griffin’s announced gift to support a new Miami campus focused on technology and education. He frames the investment in talent as relevant to the next generation of businesses.

Asked about the global economy, Paulson says he sees unusually high uncertainty and risks. He points to sovereign and private debt, arguing that high debt can increase vulnerability to financial turbulence while reducing room to respond.

He also worries that economies and monetary policies are moving in different directions, making international coordination harder. He contrasts the United States and China and describes them as out of sync.

Paulson describes challenges he sees in China, including weak consumption, industrial deflation, unemployment and excess capacity. His assessment and figures are his own view at the October 2026 event, not a comprehensive account of China’s economy.

He describes the U.S. economy as resilient despite inflation, trade and immigration-related headwinds. He also notes the demand for capital from data-center construction and says monetary policy and energy prices add uncertainty.

Paulson warns that rapid AI investment and interconnected financing could lead to a severe correction. He distinguishes that possibility from a safety incident that could abruptly disrupt development. These are risk scenarios, not forecasts of an inevitable outcome.

He identifies climate change and biodiversity loss as longer-term economic and social risks. Paulson gives specific extinction estimates, but the interview does not establish their methodology; his broader argument is that damage to nature can impose large, hard-to-measure costs.

Esposito asks about U.S.–China relations. Paulson calls the relationship consequential and fraught: the countries compete strategically and economically while their economies remain deeply connected.

Paulson describes a form of stability based less on trust than on the cost of mutual economic disruption. Each country can hurt the other, he argues, and neither can easily afford uncontrolled escalation.

He advocates regular, purposeful dialogue to monitor shared risks, including AI safety, nuclear proliferation, pandemics and financial crises. In his view, guardrails and an understanding of each side’s red lines can help prevent competition becoming conflict.

Paulson offers a positive assessment of recent leader-level meetings and discusses prospective economic and AI safety dialogue. Those descriptions reflect his remarks at the event; the future outcome of the talks was not known in the interview.

He argues that competition between the two systems will also be decided domestically. The United States has important strengths, he says, but political polarization, debt and the management of AI development are challenges it must address.

Esposito asks about China’s energy position and AI. Paulson says China has expanded electricity generation and transmission rapidly across multiple sources, while he sees advanced chips as a more significant constraint for its AI ambitions.

He notes China’s dependence on imported oil and points to electrification and strategic reserves as parts of its energy-security response. His comparison treats energy security as a national-security issue.

For the United States, Paulson distinguishes abundant energy resources from the ability to deliver electricity where it is needed. He identifies permitting, transmission and project timelines as bottlenecks and says energy policy has become relevant to AI policy.

Esposito asks about Paulson’s opinion piece with former Treasury secretary Robert Rubin. Paulson says their central concern is AI safety and security, including cyberattacks, threats to infrastructure and possible misuse of biological technologies.

He argues that extraordinary investment incentives can make risks harder to assess objectively, and doubts that competing companies can provide sufficient oversight entirely on their own. His case is for constraints and accountability beyond the firms developing the technology.

Paulson says history often involves recognizing a risk but underestimating its severity. He cites incidents involving frontier AI as warning signs, although the conversation does not independently establish the details of those incidents.

He and Rubin have advocated cooperation between the United States and China on AI risks. Paulson compares the rationale to arms-control arrangements between distrustful rivals, while acknowledging that monitoring AI is different and difficult.

Paulson also expects AI to affect employment, particularly white-collar work in advanced economies. The timing and scale are uncertain. He calls for government, businesses and academia to help workers adapt, including incentives for training and retraining.

Esposito turns to fiscal policy: why do large deficits persist when the economy and markets appear relatively strong, and will a crisis be required before elected officials act? He frames the issue as a question about political will and timing.

Paulson argues that an unchecked debt trajectory could harm prosperity and national security. He says Congress often struggles to make politically difficult choices before a crisis is visible, and that delay can make adjustment more costly.

He says he cannot know when a fiscal crisis might occur or what would trigger one. His strong concern about sustainability is a judgment about risk, not a prediction of a particular date or market event.

Given the possibility of delayed action, Paulson urges the government to prepare a contingency plan with targeted short-term responses. He recalls seeking congressional authority during the Fannie Mae and Freddie Mac emergency without knowing whether lawmakers would grant it.

Esposito asks what Paulson learned about leadership in the 2008 financial crisis. Paulson starts by acknowledging the human cost: people lost jobs and homes, and the recovery and asset-price effects were distributed unevenly.

Paulson distinguishes the causes of the crisis from the emergency response. He cites mortgage-market incentives, regulatory shortcomings and global imbalances, and argues that the actions officials took prevented a still worse outcome. That counterfactual is his assessment.

His first lesson is that leaders rarely have all the information they want when a crisis demands action. Waiting for certainty can mean acting too late, so decisions must be revisited as new facts arrive.

He gives Fannie Mae and Freddie Mac as an example. He initially sought broad authority to reassure markets, then says he had to use it as conditions deteriorated. He argues the intervention helped preserve mortgage funding, even though many households still lost their homes.

A second reversal concerned bank capital. Paulson says he originally planned to buy troubled assets and resisted direct capital injections, partly because he feared nationalization. As bank failures spread, he changed course and proposed injecting capital through non-voting preferred shares.

He recalls President George W. Bush asking how he would explain the reversal. Paulson says he answered that he would acknowledge the mistake and explain the new plan. He credits the intervention with stabilizing the system, a subject that remains open to historical assessment.

Paulson argues that the public cost and returns of particular interventions should be examined alongside their distributional consequences. He says some support was repaid with returns, while earlier acknowledging that the crisis and recovery harmed households unevenly.

His next lesson concerns authority and persuasion. A Treasury secretary depends on presidential support and must work with Congress and independent regulators. Paulson says relationships built before the crisis made cooperation possible when pressure intensified.

He recalls building relationships across party lines, including with Senate leader Harry Reid and House figures Barney Frank and Nancy Pelosi. The point, he says, was not that agreement was easy, but that trust and a working relationship existed before an emergency.

Paulson credits cooperation with Ben Bernanke and Tim Geithner, whose teams brought different authorities and capabilities. He believes public fiscal resources remain necessary to contain some financial panics; this is his policy judgment rather than an uncontested rule.

He contrasts the public reaction to capital injections with less visible guarantees and notes that some tools developed in the financial crisis were later used during COVID. For him, an enduring problem is gaining authority before a crisis makes action unavoidable.

International coordination was another lesson. Paulson says relationships with Chinese officials allowed him to call quickly as financial markets came under pressure. He describes those channels as useful for containing immediate disruption and coordinating later action.

He says Chinese lending and stimulus helped support global demand after the crisis, while also arguing that excessive stimulus later damaged China’s economy. This is his interpretation of a complex historical episode.

Paulson recalls Bush’s support during difficult days, including anxiety about Citigroup and the use of TARP funds. He says the president encouraged him by pointing to the team’s experience and the time they had already spent working together.

Esposito asks for leadership principles that apply across the public and private sectors. Paulson’s first answer is to define the job broadly: remain within legal and ethical limits, but change tactics, timelines or organization when circumstances demand it.

He says strong leaders keep core values while being flexible about how they execute. Even junior employees can show this habit by asking why an assignment matters and what else is needed to achieve its purpose.

Paulson also emphasizes building a team that compensates for a leader’s weaknesses. A decisive person may need colleagues who encourage more reflection; candidness can become indiscretion. The leader must listen to those colleagues.

Finally, he calls for a culture where people can ask questions and raise concerns instead of hiding problems. In uncertain times, he favors liquidity and a strong balance sheet so an organization can absorb shocks and respond to opportunities.

On AI, Paulson says adopting a technology is not by itself a winning strategy. Businesses must decide how to use it to differentiate themselves. Esposito closes by thanking him for the conversation and his service.