| Finance Studio Advisors |
The Ledger Letter |
|
|
A $134 Billion Trial Starts Monday. The Real Money Is Already Three Layers Down.
Jury selection in Musk v. Altman begins Monday in federal court in Oakland, with $134 billion in claimed damages on the table, per Fortune and CNBC. The headlines will frame it as a billionaire grudge match. The capital flows say something else. The four largest U.S. hyperscalers have guided to nearly $700 billion in 2026 capex, per disclosures compiled by Futurum and Introl. Roughly 75% of that is AI infrastructure. The court will decide who owned a charter. The market has already decided who owns the stack.
|
|
|
The Breakdown
|
| 01 |
The Lawsuit Is the Headline
Musk's complaint, narrowed to unjust enrichment and breach of charitable trust, alleges OpenAI abandoned its 2015 nonprofit mission after taking Microsoft funding and restructuring as a for-profit, per Fortune. OpenAI, Altman, Brockman, and Microsoft deny the allegations and have called the case baseless. Satya Nadella is on the witness list. The jury verdict will be advisory only.
|
|
|
| 02 |
The Stack Is the Money
Microsoft alone spent $11.1 billion leasing data center space in Q1 2026 and is tracking toward $120 billion or more in fiscal 2026 capex, per Introl. Its disclosed Azure backlog of unfulfilled orders sits near $80 billion — gated by power, not demand. The OpenAI partnership is now an infrastructure problem before it is a software one.
|
|
|
| 03 |
The Market Already Knows
Goldman Sachs projects total hyperscaler capex from 2025 through 2027 at $1.15 trillion, more than double the $477 billion spent over the prior three years. Capital intensity at the top five cloud providers has reached 45–57% of revenue, per CreditSights — a profile that resembles utilities and railroads more than software. Investors are rewarding the builders of the stack, not the wrappers on top.
|
|
|
|
|
By the Numbers
2026 AI Infrastructure Capex, by Hyperscaler
|
| Amazon |
$200B |
| 2026 capex guidance, predominantly AWS — per Network World |
|
| Alphabet |
$175–185B |
| Pichai-stated 2026 budget — per Network World |
|
| Microsoft |
$120B+ |
| Fiscal 2026 — Azure backlog ~$80B, per Introl |
|
| Meta |
$115–135B |
| 2026 guidance — per Futurum |
|
| Oracle |
$50B |
| 2026 target — Stargate JV anchor, per CreditSights |
|
|
|
Combined 2026 commitment: roughly $660–720 billion, or about 2.2% of U.S. GDP, per IEEE ComSoc analysis. Roughly 75% targets AI infrastructure directly.
|
|
Partner Perspective
|
|
The traditional playbook treats lawsuits as a sideshow to fundamentals and AI as a winner-take-all software contest. Both assumptions are now misleading. The Musk v. Altman trial will produce months of headlines about a charter, a board, and a $134 billion damages claim, per CNBC. None of it changes the underlying capex commitments already made by Microsoft, Amazon, Alphabet, Meta, and Oracle. UBS estimates 2026 hyperscaler capex will consume nearly 100% of operating cash flows, against a ten-year average of 40%. Bank of America puts the post-buyback figure at 94%. Big tech has issued more than $100 billion in bonds in 2026 alone to fund the buildout, per IEEE ComSoc. These are not the metrics of a software business. They are the metrics of a utility cycle wearing a software brand.
|
|
Three forces explain why infrastructure is now where the money is consolidating. First, the model layer is getting commoditized faster than expected. OpenAI, Anthropic, Google, Meta, and several open-weight competitors are converging on similar capability levels, which compresses pricing power for any single lab. Second, the inputs above the model — accelerators, memory, fabrication capacity, power, cooling, fiber, real estate — are concentrated among a small number of suppliers with multi-year lead times. Nvidia controls the dominant accelerator. TSMC and SK Hynix control the silicon and the high-bandwidth memory. The hyperscalers control the racks. Power utilities and grid operators control whether any of it can run at scale. Third, the value is now being captured before public listings exist. SpaceX has reportedly filed confidentially for an IPO at a targeted $1.75 trillion valuation, per CNBC and Reuters. The Stargate joint venture is targeting $500 billion in AI infrastructure investment by 2029. Anthropic, xAI, Databricks, and Anduril are all in late-stage private rounds at multi-hundred-billion-dollar valuations. By the time a ticker exists, much of the re-rating has already happened in private markets.
|
|
What This Means for Positioning
|
|
The trial outcome is binary, the capital cycle is not. Musk could win, lose, or settle, and the underlying buildout would still proceed. The more useful framework for long-term investors is to think in layers, not tickers — and to recognize that the companies most exposed to AI value capture are not always the ones with "AI" in their pitch deck. A short list of the layers attracting institutional capital today:
|
|
Compute — Nvidia, AMD, Broadcom, custom-silicon programs at hyperscalers.
Power — utilities, IPPs, nuclear restarts, gas peakers tied to data-center load.
Real estate & networks — Equinix, Digital Realty, dark fiber, dedicated long-haul.
Distribution — Microsoft, Amazon, and Google as the contracts gateway to enterprise demand.
Edge & orbit — satellite broadband, direct-to-cell, defense-tech dual-use, mostly still private.
|
|
|
Goldman's $1.15 trillion three-year capex forecast is roughly the inflation-adjusted cost of the U.S. interstate highway system. Dell'Oro Group reported global data center capex grew 57% in 2025 to $726 billion — the fastest growth the firm has tracked since it began measuring. Amazon's contracted backlog alone now sits at $244 billion, up 40% year-over-year, per Network World. The pattern these numbers describe is consolidation: a small number of vertically integrated operators absorbing decades of forward demand before competitors can build comparable capacity. Whatever the verdict in Oakland, that consolidation is the story investors are pricing.
|
|
The trial decides who controlled the past. The capital cycle is already deciding who controls the next decade.
|
|
|
|
Finance Studio Advisors
Precision in every paragraph. Financial communication built for clarity, credibility, and action.
|
|
This newsletter is for informational purposes only and does not constitute investment advice. Past performance is not indicative of future results.
|
|
