Kyle Johnson
Heads I win, tails the taxpayers lose—that’s how financial markets operate at the highest levels.
Right now, many of the wealthiest people in tech and finance are lobbying the federal government for various forms of support. Historical precedents justify fears that insiders will be allowed privatized gains and socialized losses.
Speaking at a November Wall Street Journal event, OpenAI CFO Sarah Friar denied rumors of an IPO, claiming the company was looking to create an “ecosystem of banks, private equity, maybe even governmental” to take on more debt. She floated the idea that the US government could “backstop the guarantee that allows the financing to happen.”
Friar’s comments caused such a stir that she took to LinkedIn to explain that “OpenAI is not seeking a government backstop for our infrastructure commitments.”
OpenAI CEO Sam Altman explained on X that “we do not have or want government guarantees for OpenAI data centers.”
There’s just one problem.
On October 27 this year, OpenAI published a letter from Christopher Lehane (OpenAI’s chief global affairs officer) to Michael Kratsios (executive director of the Office of Science and Technology Policy) on its website. Arguing that AI infrastructure will strengthen America’s industrial base, Lehane wrote:
“To provide manufacturers with the certainty and capital they need to scale production quickly, the federal government should also deploy grants, cost-sharing agreements, loans, or loan guarantees to expand industrial base capacity and resilience.”
Lehane notes that government support could be made using existing authorities, “such as the Defense Production Act Title III and the Department of Energy’s Loan Program Office.”
Are we to believe that OpenAI’s CIO and CEO are unaware of the company’s lobbying efforts?
The backpedaling seems insincere and motivated by public relations—banking on most people not knowing about Lehane’s letter.
There’s another angle to the AI jockeying that is much more compelling: national security.
In 2017, Vladimir Putin claimed that the nation leading in AI will “be the ruler of the world.”
Recently, NVIDIA CEO Jensen Huang said, “China is going to win the AI race.”
Does he actually believe that? Or is he just trying to motivate prompt action in Western countries?
I’m not sure.
But his words are an excellent way to scare politicians and bureaucrats into coughing up billions, if not trillions of tax dollars.
To be clear, I think national security concerns are valid, even if so-called superintelligence or artificial general intelligence (AGI) is never achieved. Digital security is critically important.
One look at the Pentagon’s yearly budget proves America is willing to spend top dollar to secure military dominance. The US outspends other developed countries by orders of magnitude.
You’d be hard-pressed to design something better positioned for government support than AI.
Do officials want America and the West to have technological superiority?
Yes.
Are there valid national security concerns?
Yes.
Can GDP fanatics make wild growth claims?
Trillions of times over, yes!
I have no doubt that many politicians and bureaucrats believe that building the AI infrastructure grants them carte blanche to spend with reckless abandon.
The sector is ripe for government involvement. But under today’s economic conditions, there might be a perfect storm. Despite the best efforts of government economists, inflation remains a problem for most Americans. The job market is weak and crumbling.
White collar.
Blue collar.
Jobs for everyone!
What do mainstream economists love most?
Spending.
You should be aware that JPMorgan Chase analysts estimate that over the next five years, the gap between capital expenditures on AI infrastructure and expected revenue is $1.5 trillion.
The recent comments from AI insiders must be viewed in light of historical precedent. There’s reason for investors to anticipate significant government involvement in building the AI infrastructure…
This month, the Energy Department announced that nuclear power will be getting the bulk of financing aid from its Loan Programs Office (LPO)… yes, the exact mechanism lobbied for by OpenAI.
Commenting on the matter, Energy Department Secretary Chris Wright said he expects electricity demand from AI to attract billions in capital from “very creditworthy providers.” Under the new initiative, the Energy Department could match private investment dollars by as much as four to one with low-cost debt financing.
By the time Trump’s presidency ends, Wright claims he wants to see “hopefully dozens of nuclear plants under construction.” Won’t that cost a bundle? Well, the LPO has over $400 billion at its disposal.
This is music to the ears of all companies involved in building the AI infrastructure… ducks in a row.
I appreciate the difference between a loan and a bailout. But because the government is not motivated by profit like private-sector actors are, the incentive to control costs is diminished. All government involvement should be analyzed with greater suspicion.
Fannie Mae was created as a government agency as part of the New Deal in 1938. Ostensibly, it was privatized in 1968. Freddie Mac was created as a “private” competitor in 1970. Both were officially recognized as “government sponsored enterprises” (GSEs). This designation came with a wink-wink, nudge-nudge guarantee.
Officially, the Fannie had a $2.25 billion line of credit with the US Treasury. But everybody behaved as if there would be more during a crisis.
According to a 2008 article in the New York Times, two Fannie Mae employees claim that former Chief Executive Daniel Mudd told employees to “get aggressive on risk-taking, or get out of the company.”
People at Fannie and Freddie certainly behaved like those were their marching orders. In August 2008, while commercial banks were capped at 15:1, Fannie was leveraged 20:1. That September, Freddie was leveraged about 70:1. For reference, at that time Morgan Stanley’s leverage was 34:1, and Goldman Sachs’ was 27:1.
And what did our dear overlords have to say about Fannie and Freddie while this was taking place?
Speaking before the House Financial Services Committee hearing in September 2003, self-appointed financial expert Rep. Barney Frank trumpeted:
“I worry, frankly, that there’s a tension here. The more people, in my judgment, exaggerate a threat of safety and soundness, the more people conjure up the possibility of serious financial losses to the Treasury, which I do not see. I think we see entities that are fundamentally sound financially and [able to] withstand some of the disaster scenarios.”
At the same hearing, Rep. Maxine Waters asked, “if it ain’t broke, why do you want to fix it? Have GSEs ever missed their housing goals?”
Speaking for the Senate Banking Committee in February 2004, Sen. Christopher Dodd called the GSEs “one of the great success stories of all time.”
Speaking before the House Committee on Financial Services in July 2008, then-Treasury Secretary Henry Paulson stated, “Their regulator has made clear that they are adequately capitalized.” He did not warn of imminent doom.
Fannie and Freddie were placed into conservatorship on September 6, 2008.
On September 8, Paulson explained, “Let me say this, I never said the company was well-capitalized. What I said is the regulator said they are adequately capitalized and the market has concerns about that, because I had no basis to say it. I had never been inside. It was only after we had gone inside with the appropriate people and the appropriate teams that it was clear to me that the layer of capital was thin, and it was not as substantial as you would like. The statutory requirements for capital in these companies was set at a very low level, and was counted as capital, what was defined to be capital, in many instances, in my judgment, was perfectly appropriate under the accounting standards, was not economic capital.”
Rather irritating, despite being factually true.
According to the Congressional Budget Office, Fannie and Freddie had owned or guaranteed roughly half of all outstanding mortgages in the US by 2009. Before going into conservatorship, Fannie and Freddie held a combined $4.45 trillion in mortgage-related obligations. For reference, total US debt was only $10.6 trillion in Q4 2008.
It’s so unfortunate that Paulson didn’t think to look until the brown stuff had already hit the fan.
We were not then, nor are we now, governed by the best and brightest.
Can I interest you in a potential AI-related kicker?
Both Freddie and Fannie are still in conservatorship. Nevertheless, Director of the Federal Housing Finance Agency (and chairman of Freddie and Fannie) Bill Pulte announced that he’d like them to take stakes in publicly traded tech companies.
The tech industry has been around for decades. Why now? What’s changed? Which companies specifically?
Is there a single tech company that is entirely removed from AI?
Founded in 1994, Long-Term Capital Management (LTCM) was a highly leveraged hedge fund that relied on quantitative models to exploit inefficiencies in the market. Its board of directors included Myron Scholes and Robert Merton, who shared the 1997 Nobel Prize in economics.
The firm’s models worked… for a while. But in early 1998, LTCM lost $4.4 billion of its $4.7 billion in capital due to high leverage and exposure to the Asian financial crisis.
Later that year, the Federal Reserve Bank of New York orchestrated a $3.65 billion bailout. A group of 14 private-sector banks purchased 90% of the firm; LTCM partners retained 10%.
According to Roger Lowenstein’s book When Genius Failed: The Rise and Fall of Long-Term Capital Management, the firm began 1998 with a debt-to-equity ratio of 25:1.
LTCM deployed various strategies in the derivatives markets. Lowenstein claims that LTCM’s off-balance derivative positions had a notional value of $1.25 trillion.
What lesson did Wall Street learn from LTCM?
Might as well swing for the fences. If you have the right connections, the Fed will bail out your firm, and the executives might even walk away with generational wealth.
This is known as “moral hazard.”
Given the potential upside, I’d bet that Wall Street and tech insiders are currently using high leverage and derivatives with AI-related financial instruments.
Another lesson is that the Fed fears contagion—the idea that allowing a firm to fail will have a domino effect, bringing other firms down as well. So, everything’s fine as long as you’re too big to fail.
Can you think of a single industry that is isolated from AI?
As part of the Emergency Economic Stabilization Act of 2008 (EESA), Congress authorized up to $700 billion to be pumped into the financial system under the Troubled Asset Relief Program (TARP)—approximately $443.5 billion was eventually distributed.
A rehashing of the housing bubble and ensuing global financial crisis (GFC) is beyond our scope here. But there are some important takeaways from this episode.
Congress justified its actions by claiming bailouts would limit contagion.
Why wouldn’t politicians and bureaucrats use the same reasoning if the AI gets into trouble?
This episode also teaches us that the government will engage in favoritism.
Paulson served as CEO of Goldman Sachs until June 2006, leaving to join the Bush administration as Treasury secretary. Paulson then quarterbacked the 2008 bailout bill that ultimately sent funds to his former employer.
Institutions like Wells Fargo, Bank of America, and Morgan Stanley received TARP funds. JP Morgan received TARP funds and government assistance to acquire Bear Stearns. Lehman Brothers was allowed to fail. Only a fool would believe the final outcome was random or strictly meritocratic.
Suppose that the AI network falters. Who gets bailed out? Who’s allowed fail?
Your guess is as good as any. I’d look for individuals who have gone from Wall Street firms to key positions at government agencies like the Securities and Exchange Commission, Treasury Department, Department of Energy, and the like.
Only one thing is certain: taxpayers will lose.
There’s one more interesting bit about the TARP bailout: American Insurance Group (AIG) was the largest recipient of assistance from the federal government and Federal Reserve ($161 billion). The Fed offered AIG an $85 billion line of credit, on which $72 billion was drawn.
The list of companies begging for AI-related subsidies is longer than many would suspect. It could easily include every company that plays a part in the construction and powering of data centers: insurance companies, energy companies, water companies, construction companies, miners, refiners, etc.
You’ve probably noticed how many Fortune 500 companies are attempting to replace thousands of workers with AI.
What happens if the power goes out?
Expect companies in retail, shipping, logistics, etc. to also rattle cups.
During the Global Financial Crisis, Cyprian banks introduced the world to a new type of emergency financial action: the bail-in.
Put simply, distressed banks pilfered customer accounts to satisfy their obligations relating to investments made in the markets.
This move was celebrated and adopted at highest levels of global governance.
The Financial Stability Board (FSB) was established in 2009 and quickly became hosted and funded by the Bank for International Settlements (BIS)—what many refer to as the “central bank for central banks.”
In 2011, the FSB began publishing a list of Globally Systemically Important Financial Institutions (G-SIFIs), which are subcategorized into globally systemically important banks (G-SIBs) and globally systemically important insurers (G-SIIs).
An FSB report published on November 10, 2014, defines “bail-in” power as:
“the power to write down and convert into equity all or part of the firm’s unsecured and uninsured liabilities of the firm under resolution or any successor in a manner that respects the creditors hierarchy and the extent necessary to absorb losses.”
In plain English, bank deposits will be used to settle the debts of the bank in exchange for equities (which will most certainly be illiquid). Look for yourself: many US financial institutions are listed as G-SIFIs.
The BIS/FSB’s guidance is in line with a report published by the International Monetary Fund on April 24, 2012: From Bail-out to Bail-in: Mandatory Debt Restructuring of Systemic Financial Institutions.
The report defines a “bail-in” as a:
“Statutory power of a resolution authority (as opposed to contractual agreements, such as contingent capital requirements) to restructure the liabilities of a distressed financial institution by writing down its unsecured debt and/or converting it to equity. The statutory bail-in power is intended to achieve a prompt recapitalization and restructuring of the distressed institution.”
What does this have to do with AI?
Pre-approved bail-ins are offered as a fast-acting mechanism to stop contagion. And the power brokers’ favoritism is for all to see in black and white.
If AI lives up to a tiny fraction of the hype, wouldn’t many key players in AI be “globally systemically important” on par with various banks and insurers?
Big-cap CEOs are prone to bluster. It’s generally wise to assume their public comments are offered for posturing rather than truth.
But I found a few statements from Mark Zuckerberg to be extremely revealing. This September, Zuckerberg went on the ACCESS podcast and explained Meta’s commitment to being at the frontier of superintelligence. He admitted that an AI capital expenditure bubble was possible.
Zuckerberg: “From Metas’s perspective, I think the strategy is actually pretty simple … at least in terms of building out the infrastructure. No one knows when superintelligence is going to be possible. Is it going to be three years? It could be five years… it could be eight years… whatever… Let’s say you weren’t sure if it’s going to be three or five years. In a conservative business situation, maybe you’d hedge building out your infrastructure. Because you’re worried that you build it out assuming it’s going to be three years and it takes five, then you’ve lost maybe a couple hundred billions of dollars or something.”
Interviewer: “That’s a lot of money.”
Zuckerberg: “[I]t is objectively a huge amount of money.”
Interviewer: “Didn’t you just tell Trump that you were going to spend $600 billion?”
Zuckerberg: “I did … through 2028.”
Interviewer: “That’s a lot of money.”
Zuckerberg: “It is. And if we end up misspending a couple of hundred billion dollars… I think that is going to be very unfortunate, obviously. But… I actually think that the risk is higher on the other side. If you build too slowly and super intelligence is possible in three years but you build it out assuming it will be there in five years, then you’re just out of position on what I think is going to be the most important technology that enables the most new products, innovation, and value creation in history.”
I’ve previously argued that AI seems poised to create more trillion-dollar companies and possibly the first US dollar-denominated trillionaire. Judging by his comments and actions, Zuckerberg seems to agree.
I suspect Sam Altman, Elon Musk, Jensen Huang, et al. feel the same.
Even if the governments and central banks allow industry players to rise and fall on merit, hundreds of billions (if not trillions) will be spent building the AI infrastructure.
You can read about rent-seeking behavior in any introductory economics book.
Private actors have used various schemes for their own gain since the first government was formed. It’s no surprise that such behavior continues today.
AI-related government loans have already been made. The industry is well positioned for “emergency actions” should it or the economy falter.
What to do? Well, resource investors should keep in mind that the AI infrastructure simply cannot exist without copper. And nuclear is the insiders’ preferred energy source.
Even without the AI boom, copper seems destined for a bull market in the coming years. And the bull market in uranium has a long way to run.
But the potential for governments and central banks to spend like drunken sailors on AI infrastructure adds sparks to an already exciting time for copper and uranium.
KJ
P.S. Copper was Lobo’s highest-confidence play for 2025 and will be again for 2026. Learn more about how he’s playing copper, uranium, and other metals by subscribing to our free, no-hype, no-spam newsletter: theThe Digest.
