Goldman Just Named an AI Capex Super Cycle. Wall Street Is Selling Into It.

Goldman (GS) CEO David Solomon gave the AI boom a name this week: AI capex super cycle. It runs three to five years and touches every financing instrument, every region, every industry.

Equity issuance is already running ahead of the last four years combined. Hyperscaler capex tops $800 billion this year, $1 trillion next year.

The UAE just got coveted AI chip access as a reward for Iran war support. G42 can now buy from Nvidia (NVDA) freely.

CoreWeave (CRWV) is exploring financial derivatives to hedge memory chip price risk.

I know because I've seen the evidence firsthand.

On March 2nd — three days after the first missiles hit — I sat across from two U.S. Congressmen in back-to-back private meetings.

Those meetings pointed me toward something I spent weeks verifying.

The real purpose behind the strikes. The real objective. And the single company at the dead center of all of it.

This isn't random. It's a calculated Two-Front Economic War.

And there's one company positioned right at the heart of it.

See the strategy behind the Iran war — and the company at the center of it

The sooner you understand what's really happening — the better positioned you'll be before August 12th.

— Dylan Jovine, Founder, Behind the Markets

The Super-Cycle Commitment

A name for the boom Goldman CEO David Solomon said it plainly this week: "We are in the middle of an AI capex super cycle where there are demands on financing in every single financing instrument, in every region of the world and across every single industry." He says Goldman is gearing up for a three-to-five-year investment cycle that's still in its early innings. Wells Fargo's Mike Mayo agrees the boom hit "a tipping point" last quarter. So far, so bullish.

The Equity Issuance Counter

But look who's selling. The Wall Street Journal reports $344.7 billion of new shares have been sold this year per Dealogic, more than 2025, 2024, 2023, and 2022 combined. US companies will issue a net $500 billion of equities and debt over the next year, versus a net reduction of $1 trillion in recent years.

Hyperscaler capex tops $800 billion this year and $1 trillion next year per Janus Henderson. Amazon raised $85 billion in H1 2026. Oracle now has negative cash flow. Janus Henderson's John Lloyd: "Many of the hyperscalers are beginning to undo years of carefully manicured capital allocation, with share buybacks now making way for share issues."

The Bank Financing Framework

BofA has helped raise nearly $500 billion for AI-related companies since 2025, accounting for 60% of such fundraising. Bank of America (BAC) extended a $520 million credit line to OpenAI, its first loan to the AI company. Meta (META) is working with Morgan Stanley (MS) and JPMorgan (JPM) on a roughly $13 billion financing package for a data center in El Paso. Anthropic and OpenAI are both preparing IPO filings. Goldman (GS) and Morgan Stanley are poised to lead both. DeepSeek is preparing an IPO filing at $71 billion pre-money.

The Bull Market Threat

The S&P 500 dividend yield is 1.05%, a record low. Comparable issuance took place in late 1999 and the first half of 2000 before the dot-com collapse. Comparable issuance took place in 2021 with SPACs. 

The Super-Cycle Signal

Watch whether Warsh calls out AI capex as a persistent inflation contributor today. Watch whether Morgan Stanley Wednesday earnings confirm the Iran-plus-SpaceX volatility framework at bank-wide scale. Watch whether Anthropic or OpenAI publish a formal IPO filing timeline before Q3 earnings.

Goldman is calling a three-to-five year cycle. The equity issuance data is running the sprint.

Trump has signed 220 Executive Orders in one year…more than almost every U.S. president in history.

Now, on July 24th… He’s preparing to sign what sources say will be his final one.

A White House leak suggests this won’t just erase Biden’s legacy…

It will trigger a $2 trillion initiative to radically reshape America forever.

While making fortunes for those who are prepared for what’s coming.

The details are shocking. But you can’t miss this.

See what Ian King uncovered.

This one's a fascinating mix of geopolitics and silicon. The Wall Street Journal reports the UAE has won expanded access to top AI chips after helping the US against Iran, flying dozens of airstrikes, intercepting hundreds of missiles, and keeping oil flowing through Hormuz. For buying sensitive tech, the UAE is now treated like a close ally (think South Korea or India) rather than being lumped in with China. In practice, that lets its AI champion G42 buy freely from Nvidia for at least nine months and clears the way for Microsoft (MSFT) and OpenAI data centers there.

There's a wrinkle worth knowing. G42 is controlled by the UAE's national-security adviser, Sheikh Tahnoon bin Zayed Al Nahyan, who, just before Trump's second inauguration, was part of a group that put $500 million into World Liberty Financial, the Trump family's crypto venture. As the Hudson Institute's Michael Sobolik cautions, being a useful wartime partner isn't the same as being able "to keep a data center secure."

The Signal to Watch

Any second Gulf state receiving expanded AI chip access before Q3 earnings converts the framework from UAE-specific to Gulf-military-alliance-wide.

Here's a subtle but telling move. Reuters reports CoreWeave is exploring financial derivatives to protect itself if memory-chip prices drop. 

The backstory: cloud operators locked in long-term deals with Micron (MU) and SanDisk (SNDK) that guarantee suppliers a price floor. Great when chips are scarce, painful if prices fall and you're stuck overpaying. So CoreWeave is eyeing put options (a bet that pays off if prices drop) to cover the downside. Talks are early. New capacity from SK Hynix (SKHY) and Micron doesn't fully ramp until early 2028.

The Signal to Watch

Any AI cloud company actually executing a memory hedge. That flips this from CoreWeave testing an idea to the whole industry bracing for a price drop.

OpenAI's hardware dream starts small and strange: a mobile, screen-free smart speaker with a camera, sensors, moving mechanical parts, and a battery, per Bloomberg. The plan is to show it this year and ship in 2027. Standing in the way is Apple, which sued OpenAI last week claiming stolen trade secrets, no surprise given OpenAI has hired 400-plus Apple people and put Apple's former design chief Evans Hankey in charge. Apple wants an injunction that could delay sales entirely. (OpenAI spent $6.5 billion buying Jony Ive's hardware startup to get here.) Sonos (SONO) shares dropped more than 10% on the news.

The Signal to Watch

A court siding with Apple on an injunction. That would push OpenAI's launch from late 2026 into 2027 or beyond.

Jim Rickards believes the Trump administration is about to take a direct stake in a tiny $2 stock. 

The Trump administration has taken a direct stake in MP Materials, Lithium America, Trilogy Metals, and USA Rare Earth.

Each time, shares sprinted higher. Click here to see why Jim believes this one is next.

CAPITAL DISCIPLINE

Here's where we stand. Goldman named a three-to-five-year AI super cycle, even as new stock sales blew past the last four years combined, a mixed signal if there ever was one. The UAE traded war support for chip access. CoreWeave is hedging its memory bets. And OpenAI's first gadget is a speaker Apple is suing to stop.

Three things this week will tell you which way the wind is blowing: whether Warsh flags AI spending as a real inflation threat, whether Morgan Stanley's trading desk had a blowout quarter, and whether the big AI labs put IPO dates on the board. 

Watch those, and you'll know if the super cycle is a durable boom or a late-cycle scramble.

If you track just three things this week: Warsh’s Senate testimony today, Morgan Stanley’s earnings report this morning, and any IPO timeline from Anthropic or OpenAI.

Together they answer the question under everything today: is this the early stage of a years-long build, like Goldman says, or the late-cycle rush the issuance data hints at, with four giant AI labs all reaching for the public markets at once?