CNBC Fast Money
2026-08-12 · Hosted by Melissa Lee · CNBC
Executive Summary
A day after Nvidia's CEO Jensen Huang assembled leaders from Goldman Sachs, Blackstone, KKR, Brookfield, Apollo, and BlackRock to announce a $500 billion AI infrastructure financing push, the panel debated whether this represents dangerous circular financing or sound securitization. Nvidia itself closed roughly flat, but stocks of the financial backers rallied sharply. Dan Nathan warned of a "single point of failure" around Nvidia, while Karen Finerman compared the structure to decades-old vendor financing models like GM Financial.
Key Stories & Changes
1. Nvidia's $500 Billion Financing Consortium: Genius or Risk?
Nvidia CEO Jensen Huang convened leaders representing over $23 trillion in assets under management (Goldman Sachs, Blackstone, KKR, Brookfield, Apollo, BlackRock) to back a $500 billion AI infrastructure financing push
Nvidia stock closed roughly flat; financial-backer stocks (KKR, Apollo, and peers) were among the S&P's top performers on the day
Dan Nathan warned the arrangement creates a single point of failure around Nvidia, which must keep backstopping the ecosystem to sustain the trade
Karen Finerman noted Jamie Dimon/JPMorgan was notably absent from the table and drew a parallel to GM Financial's decades-old vendor financing model
Julie Biel argued the real AI bottleneck isn't financing but physical constraints (data center buildout, power/turbine availability) and unclear unit economics
2. Steve Eisman: Not a New Asset Class, But Watch the Concentration Risk
Eisman, the "Big Short" investor, said the Nvidia deal is "not a new asset class" — it's a securitization of AI data center revenue, comparable to 50 years of securitizing other revenue streams
He noted hyperscalers' free cash flow has deteriorated from funding AI capex with debt and (increasingly) equity; the $500 billion facility could help relieve that pressure
Key risk flagged: roughly 70% of AI revenue at Microsoft, Amazon, Google, and Oracle comes from just two customers — Anthropic and OpenAI
Those two companies represent 25-35% of hyperscaler cloud revenue, yet remain private with financials "hidden to almost everybody"
Eisman's flagged catalyst for concern: cheaper Chinese open-weight models taking share and triggering a price war
He declined to make a directional call, comparing his approach to subscribing to Moody's Securitization Database before shorting mortgages pre-2008 — he lacks an equivalent real-time dataset here yet
3. Core Weave Surges on Earnings Beat
Shares up 13.5% after hours; revenue doubled year-over-year
Backlog grew to $104 billion, with more than $25 billion in new commitments added early in Q3
Net loss of $626 million, narrower than Wall Street feared, driven by heavy debt service
Active power capacity reached 1.5 gigawatts, more than triple a year ago, but still less than half of contracted capacity
Nvidia owns roughly 12% of Core Weave; the company could benefit as a middleman in Nvidia's new $500 billion financing pool
Julie Biel: encouraged by backlog strength but flagged debt-dependent economics as a lingering question mark
4. On Holdings Craters, Drags Sneaker Sector
ONON: On Holdings — -20%+ (worst day ever) — Mixed results, weak guidance; margin pressure from refusing to discount
NKE: Nike — Lower — Down from $179 high to ~$41; part of broader sector downtrend
CAVA: Cava — +11.5% — Same-store sales beat estimates; strength in lower-income and 25-34 demographic
Karen Finerman: On wanted to avoid appearing as a "lesser brand" by discounting, which hurt margins without preserving premium positioning
Julie Biel: the "Apple of sneakers" premium-brand thesis is struggling against U.S. wholesale discounting dynamics
Carter Worth (technical view): "don't buy stocks in downtrends" — no relationship between low valuation and near-term price direction
5. CPI Preview and Fed Rate Path
July CPI due Tuesday 8:30am ET; consensus expects core CPI to dip slightly, per Rick Santelli-style setup discussed by Joe LaVorgna
LaVorgna (SMBC Americas chief economist): the Fed should have already hiked, framing it as removing "insurance" cuts from last year now that growth looks strong (potentially near 6% in Q3)
He still expects a hike before year-end, but says a soft CPI print would take a September hike off the table, pushing the decision toward December given midterm-election timing sensitivities in November
Core PCE (Fed's preferred gauge) still running "well over 3%"; PPI due the following day expected to remain above 4%
Trends Identified
1. Securitization of AI Compute Enters the Mainstream Debate
The panel's extended debate over whether Nvidia's financing consortium constitutes a "new asset class" or plain-vanilla securitization reflects growing Wall Street focus on how AI capex gets funded as hyperscaler balance sheets strain. Eisman's framing — that this is old wine in a new bottle — pushes back against more alarmist comparisons to 1929-style leverage, while still flagging genuine customer-concentration risk downstream.
2. Consumer Discretionary Bifurcation in Sneakers/Retail
On Holdings' historic drop, alongside weakness across Nike, Under Armour, and Dick's Sporting Goods, signals that premium athletic brands are struggling to sustain direct-to-consumer, no-discount strategies in a value-conscious U.S. market. Cava's strength among lower-income consumers suggests spending remains resilient at the value end even as premium discretionary categories wobble.
3. AI Capital Expenditure Still Expanding Despite Debt Concerns
Core Weave's record backlog and Super Micro-style guidance beats (referenced via the broader AI infrastructure trade) show demand for compute remains "effectively sold out" even as debt service costs weigh on near-term profitability — a tension likely to persist as neo-clouds scale. ---
Sentiment Analysis
Overall Market Sentiment: Divided — Bullish on AI Infrastructure, Wary of Financing Structure
Panelists broadly endorsed near-term AI infrastructure demand while expressing structural unease about how that demand is being financed and its dependence on two private AI labs.
Risk Factors Highlighted
AI customer concentration: ~70% of AI revenue at major hyperscalers traces to just Anthropic and OpenAI, both private and financially opaque.
Chinese open-weight model competition: Cheaper Chinese AI models could trigger a price war that undermines the economics underpinning the financing deal.
Circular/vendor financing risk: Critics argue Nvidia is effectively backstopping demand for its own chips across the ecosystem.
Core Weave's debt burden: ~$35 billion in debt with roughly $1 billion in annual debt service, requiring continued economic improvement.
Sneaker/athletic retail margin pressure: On Holdings' guidance cut and broader sector weakness reflect a difficult wholesale/discounting environment.
Hot CPI print risk: A hotter-than-expected inflation read could accelerate the timeline for a Fed rate hike, pressuring equities.
Leverage/margin dynamics reminiscent of history: Karen Finerman explicitly invoked 1929-era margin dynamics as a cautionary parallel, though contested by other panelists.
This episode was covered in today's [The Market Signal — 2026-08-12](https://marketsignal.beehiiv.com/p/the-market-signal-2026-08-12), a cross-source synthesis of multiple podcast reports.