The Ledger Letter
Finance Studio Advisors · Thursday, June 11, 2026
Market Intelligence Partner
Dear Reader,
Nvidia CEO Jensen Huang recently said something that shocked most investors…
Stating that robotics is the tech giant’s biggest opportunity after AI.
“We’re working towards a day where there will be billions of robots, hundreds of millions of autonomous vehicles and hundreds of thousands of robotic factories that can be powered by Nvidia technology,” Huang said.
The market reacted…
Helping Nvidia become the world’s first five-trillion-dollar company.
Thanks to its entry into the lucrative robotics market.
But here’s what the headlines missed…
Nvidia didn’t do this alone.
And if history is any indicator…
Nvidia’s $7 silent partner could be due for a big move up.
Michael Robinson
Michael Robinson
Weiss Ratings

AI Just Became a Balance Sheet Problem

Oracle reported record revenue Wednesday night. $19.2 billion. Cloud infrastructure up 93 percent year over year. The backlog: $638 billion. Then the company said it would spend $70 billion on AI data centers next year and raise another $40 billion in debt and equity to cover it. The stock fell nearly 9 percent after the bell, per Reuters. Hours earlier, the Wall Street Journal reported OpenAI is weighing drastic price cuts to compete with Anthropic. The AI trade just crossed a line. It stopped being priced like software. It is being priced like infrastructure somebody has to finance.
The Breakdown
Today’s disagreement: AI revenue keeps beating estimates; the balance sheets behind it keep getting heavier.
01
The Tab
Oracle spent $55.7 billion in FY2026 on capital expenditures, above its own $50 billion target, per Reuters. Interest expense climbed 29 percent to $4.6 billion. For FY2027 the company guided $70 billion in its own capex and plans to raise roughly $40 billion more through debt and equity to fund it.
02
The Price War
OpenAI is weighing drastic token-price cuts to compete with Anthropic, per the WSJ. Both companies have filed for IPOs. Combined 2026 spending: approximately $65 billion, per industry estimates. Neither has reported a profit. Revenue per unit of AI compute is falling while the cost of building that compute is rising.
03
The Cost of the Cost
May CPI hit 4.2 percent year over year, the highest reading in three years, per the BLS. The 10-year held at 4.53 percent. In this tape, AI infrastructure borrowers are competing for capital in a market where inflation keeps the cost of money elevated and the Fed is not cutting.
By the Numbers
Oracle FY2027 capex guidance$70 billion
Oracle backlog (RPO)$638 billion
Oracle FY2026 interest expense$4.6B (+29% YoY)
OpenAI + Anthropic combined 2026 spend~$65 billion
10-year Treasury yield (Jun 10)4.53%
May CPI headline (YoY)4.2% (3-yr high)
Sources: Reuters, BLS, LSEG, Oracle company filings, WSJ. Figures reflect Jun 10 close and latest available data.
Partner Perspective
Image
I’m about to do a live demonstration.
Of Elon Musk’s latest genius invention.
It’s an AI agent…
Perhaps the most powerful ever created.
Elon himself believes it could 70x your money… in a short period of time.
The Full Picture

Who Pays for the AI Buildout

The Pivot
For two years the market priced AI like a software story. High margins. Low capex. Winner-take-all platform economics. Oracle’s earnings call Wednesday was not a software company talking. It was a capital-intensive infrastructure operator describing its borrowing schedule.
Revenue beat estimates. Cloud infrastructure revenue hit $5.8 billion, up 93 percent, per Oracle’s filing. But gross margins will “step down” in FY2027, per CFO Hilary Maxson. The stock dropped because the income statement won and the balance sheet lost.
The Capital Stack
Oracle raised $43 billion in debt and $5 billion in equity in FY2026 alone, per company filings. It plans another $40 billion in FY2027. It already carries roughly $100 billion in long-term debt. Earlier this year, its five-year credit default swap cost surged from around $40 to over $150, per market data reported by Reuters. Our view: when the credit market reprices a company faster than the equity market, the credit market is usually ahead.
The Race to Discount
The timing is worth watching. OpenAI mulls price cuts the same week it files for an IPO, per the WSJ. Anthropic filed days earlier at a reported $965 billion valuation. The two companies are projected to spend $65 billion combined this year on compute, training, and operations, per industry estimates. Neither has reported a profit. OpenAI has told investors it does not expect profitability until 2030.
When the two largest AI labs start competing on price before they compete on profitability, the margin profile of the entire sector compresses. Sound familiar? It should. It is the cloud-computing price war of 2014 running again at ten times the capital intensity.
Where the Bottleneck Lives
The real question is not who builds the best model. It is who controls the physical infrastructure the models require. Nvidia. Power generators. Data center landlords. Cooling manufacturers. The AI model builders need these inputs. The input owners do not need any particular model.
Worth watching: as Oracle and its peers issue tens of billions in new debt, the companies supplying chips, power, and floor space collect the checks without shouldering the financing risk. That is the asymmetry this market has not priced in.
What Your Brokerage Statement Sees
If you own a broad tech index, you own the AI capex cycle now. You own the revenue and the borrowing. The S&P 500 dropped 1.62 percent Wednesday. The Nasdaq fell 1.98 percent. Industrials lost over 3 percent. The VIX pushed above 22. In this tape, the market is asking a question it has not had to ask for two years: what happens when AI growth costs more than AI growth earns?
Revenue beats and the stock drops nearly 9 percent. The market is no longer asking whether AI works. It is asking who pays.
The Ledger Letter
When markets disagree, the signal is in the disagreement.
This newsletter is for informational purposes only and does not constitute investment advice. Past performance is not indicative of future results.

Recommended for you