The AI systemic-risk debate: Bessent sees no problem behind closed doors, Barclays shifts quotas to keep lending, Blackstone calls fears overblown

The debate over whether AI debt poses systemic risk produced three contradictory signals on October 21: Treasury Secretary Bessent told a closed-door Wall Street gathering "I don't see a systemic problem"; Bloomberg revealed that after Barclays hit internal financing limits, it used an internal auction to shift capital quotas from its rates desk to prime brokerage to keep funding Asian AI equity trades; and a senior Blackstone executive told a tech conference that fears over AI-buildout debt are "overblown"^[1][2][3].

Bessent: the smartest people have 99% of their net worth on it

Per the Financial Times, citing three attendees, Bessent told the invite-only New York event co-hosted by JPMorgan and the Robin Hood foundation, in an interview with Bridgewater CEO Nir Bar Dea: "I'm good at telling when something's wrong in the system. I don't see a systemic problem." He also explicitly rejected the bubble framing: "The smartest people in the world — who have 99% of their net worth in these companies and have built the most profitable businesses in the world — believe in this"^[1]. The event's tickets started at $7,500 and was closed to media, with other speakers including Stanley Druckenmiller and Steve Cohen; the Treasury spokesperson's response to the transparency flap was punchy: "Secretary Bessent has 35 years of experience in finance in the public and private sectors and knows well what constitutes material non-public information. Everything he said at the event he has said publicly before... The media has become so inaccurate that he can no longer read newspapers"^[1]. Context matters: the remarks came as markets churn and some investors grow anxious over the enormous debt financing datacenters and chips — the Treasury choosing this moment to declare "no systemic risk" is itself a policy signal.

Wall Street's "moving goalposts": what to do when leverage can't grow

The banks' maneuvers make an odd counterpoint to the Treasury's calm. Bloomberg disclosed two workarounds. First, after hitting internal financing caps amid Asia's AI equity boom earlier this year, Barclays used an internal auction to transfer capital quotas from its macro rates desk to prime brokerage — keeping loans to Asian hedge funds expanding; the bank calls it "dynamically deploying the balance sheet," while insiders' version is that the "strict controls" amounted to redrawing the ceiling after hitting it. Second, traders at Goldman and others began pressing hedge-fund clients to hold more short positions — which shrinks balance-sheet usage on paper without reducing actual market leverage^[2]. The data capture the trick: in the week to October 18, US long-short funds' net leverage fell 3.2 points to 43.7%, a five-year low — but nearly all from added shorts; since July 31, gross long leverage slipped just 0.5 points to 128.8% while short leverage rose 8.0 points to a five-year-high 85.1%, leaving total leverage at roughly 214%, up rather than down^[2].

The lenders' view: Blackstone says fears are "overblown"

The buy side added its own weight. Blackstone veteran Jas Khaira said on stage at TechCrunch Disrupt last week that fears over AI-buildout debt are overblown — the private-equity giant is placing itself at the center of the boom, writing everything from massive datacenter loans to frontier-model and application investments, and this year opened Blackstone N1 (BXN1), a San Francisco AI hub led by Khaira himself^[3]. His argument echoes Bessent: unlike earlier cycles such as mobile or SaaS that let companies reach scale with moderate capital, AI's capital intensity is inherently higher — measuring a new cycle with an old cycle's leverage yardstick invites overestimating the risk^[3].

The shared facts beneath three positions

Line the three signals up and the debating sides actually share one set of facts: AI buildout's financing needs have grown large enough to keep bumping into banks' risk limits, to require the Treasury to set a tone, and to require the biggest lenders to actively cool the worries. The disagreement is purely interpretive — "normal credit expansion" versus "classic late-cycle leverage symptoms." For investors, three operable tests: the actual size and pricing of hyperscalers' Q4 bond issuance (widening spreads are the market's vote); whether Barclays-style internal shuffling draws regulatory inquiries; and whether the Fed's October language on financial stability turns softer in step with Bessent's framing^[1][2][3].