Winning the AI Race Part 4: Scott Bessent, Howard Lutnick, Chris Wright, and Doug Burgum

The session opens with the moderators addressing "Secretary Besson" and asking for a high-level update on the "333 plan," a framework first set out during the campaign. After a brief pause to switch on a microphone, the plan is laid out in three targets: cutting the budget deficit from about 6.7% of GDP, described as the highest ever recorded outside of war or recession, down to 3%; sustaining economic growth of 3% or more; and adding 3 million more barrels of oil-and-gas energy equivalent before President Trump leaves office.

As evidence of progress, the speaker says this June was the Treasury's first positive June since 2015, with an actual surplus. They credit higher revenues, partly from tariffs, together with lower spending.

The main reason for optimism is AI, which the speaker hopes could bring back the 1990s pattern, when Alan Greenspan was able to "run the economy very hot" during the IT boom while growth stayed non-inflationary. The speaker calls a repeat "highly likely" and argues that growth of that kind would shrink the deficit quickly.

Asked about a recent post on an inflection point in capital spending, the speaker calls the current investment picture a "barbell." On one side is traditional industry. They cite a trip to Pittsburgh with President Trump for the announcement of the US Steel–Nippon Steel deal, which involves substantial investment by Nippon Steel. On the other side is AI, illustrated by an AI summit in Pittsburgh the following week. The speaker calls the city a natural AI location because of cheap energy, Carnegie Mellon and Pitt.

The hyperscalers are described as being in an arms race. Their AI capex is estimated at roughly 1% of GDP a year, which the speaker puts at about $300 billion. In the speaker's "perfect world, which never happens," this capex boom would hand off to a productivity boom sometime in 2026. A questioner notes that the boom seems to defy economic theory because it has not produced the usual negative effects, and asks whether that is temporary or a sign of a structurally different economy.

The answer draws on economic history, which the speaker says they used to teach. In the 1880s and 1890s, railroads made crossing the United States 10 times faster, producing a productivity boom with very large GDP growth that was also disinflationary. The speaker describes double-digit GDP numbers alongside inflation of minus 2%, minus 3% and minus 4% as costs fell.

The 1980s under Reagan are described as a "deregulatory boom," a time when airline ticket prices, telephone bills and banking services had all been regulated. The speaker credits Paul Volcker with bringing inflation down but argues that deregulation also mattered. In the 1990s, a long electronics buildup finally paid off, especially in office work, producing a productivity boom and a surplus that paid down national debt. The speaker recalls a paper from that era asking what would happen if government bonds ran out, and remarks drily that this has since been "fixed."

The speaker returns to the 3% target: growth acceleration, combined with lower interest rates made possible by non-inflationary growth, would change the trajectory of the debt. They add that the Fed "is gonna have to be open to this idea."

A two-part question follows. It asks how much tariffs from the recently negotiated trade deals might dampen growth, and how to read a Chinese report about selling half of their US Treasury holdings.

On Treasuries, the speaker expects China to divest slowly. They argue that the GENIUS legislation passed the previous week could create several trillion dollars of demand for T-bills, because the law involves maturities under 90 days. They predict this will "lock in the US dollar" for ordinary people from Nigeria to Kuala Lumpur who use US-backed stablecoins.

The speaker sets this against central bank digital currencies in China, the eurozone, or even Canada. They recall that during COVID the Canadian government seized and froze the bank accounts of truckers whose actions it disliked. They argue that with a government-backed digital currency, someone could be shut down for "a mean tweet," whereas consumers would have "unbridled choice" with US dollar stablecoins. The panel jokes about its own reputation for mean tweets.