The Wrap: Hezbollah in Caracas? AI Flameout? Trump Buys MBS? Really?

This week in “The Wrap,” we ponder the events of the past week and give our readers some thoughts about the future. The Trump Administration celebrated January 6, 2021 by invading Venezuela in 2026, an adept move from a media perspective, but is all of this just for show? We also ask whether 2026 is the year when the tech trade fueled by AI hype fizzles out. And don’t miss our weekly podcast with Julia La Roche, which is released on most every Saturday.

Iranian Weapons in Venezuela

With the US military intervention in Venezuela, President Donald Trump now claims to control that nation’s oil sector, but much of the narco government of Nicolás Maduro remains in place. More, the growing crowd of bad actors from Iran, Cuba, China and Russia is still on the ground in Venezuela, actively planning the demise of the United States.   The Heritage Foundation reported in 2023 about Iran's extensive activities in Latin America.

Strangely, the Trump Administration has said almost nothing about the presence of offensive military weapons in Venezuela, weapons controlled by Iran that can target the US. The US has a blockade of Venezuela, but why has President Trump made so little policial hay out of the presence of Iranian Revolutionary Guards in Caracas? What would Jack Kennedy or Ronald Reagan have said? 

Revolutionary Guards

The Credit Strategist wrote earlier this week:

“Iran has been manufacturing weapons in VZ for years. Iran’s Revolutionary Guards maintain an operational infrastructure in the country with Hezbollah operating alongside. Since 2020, Iran has transferred weapons capable of reaching any target in the state of Florida (weapons that were publicly displayed in VZ military parades in 2021 through 2023 - talk about painting a target on your back!). An Israeli defense research organization, the Alma Research and Education Center, catalogued the specific systems and notes they can target both American soil and ships in the Caribbean.”

Meanwhile, a number of observers have noted that Venezuela holds as much as 600,000 bitcoin tokens, an offshore pile potentially worth tens of billions of glorious fiat dollars, though experts question the feasibility of such a large stash given local corruption and infrastructure issues.  The whole point of "socialism," after all, is theft, right?

Why does Venezuela need such a large pile of crypto? To evade US sanctions and support global allies like Iran, of course. Never forget that the primary use case for bitcoin outside the US is to support criminal and terrorist organizations and evade US sanctions. But if the crypto stash ever existed, you can be pretty sure that Maduro and his Cuban protectors have already stolen much of it over the past several decades since Hugo Chavez seized power in 1999. Just check the blockchain...

In the Washington Post years ago, we noted that the phrase "the perfect dictatorship" was coined by Peruvian novelist and Nobel laureate Mario Vargas Llosa to describe Mexico's long-ruling party, the Institutional Revolutionary Party (PRI). His comment, made in 1990 during a televised roundtable discussion with intellectuals, stunned the Mexican political class and their American sponsors. All of the major "leftist" political parties in Latin America used the PRI as a model and made looting their respective patrimonies into an artform, but none more than Venezuela.

Oracle Gone a Bridge Too Far? 

Most analysts are optimistic about the US equity markets, as one might expect, with some talking heads even predicting stock price multiple expansion due to short-term interest rate cuts by the FOMC. But some are starting to question still inflated valuations in technology and financials.

Top of the list of negative outlooks is Oracle Corporation (ORCL), which depending on who you ask is either profoundly insolvent due to its massive bet on the marketing con known as AI or the greatest trade since Nvidia (NVDA) in 2020. We wuz there (h/t Jim Cramer). Likewise, analysts are starting to fashion constructive rationales for higher bank valuations based upon lower interest rates.

“As of November 30, 2025,” notes MTS Observer, “Oracle announced almost $250 billion in planned new lease commitments, substantially all of which are related to data centers and ‘cloud capacity’ arrangements. In simple terms, what this means is that Oracle has contractually committed to incurring and paying $250 billion in rent expense over the next 20 years in support of its push into the artificial intelligence space.”

What this means is that ORCL is on the hook for many more hundreds of billions of dollars in lease expenses and related costs for its imprudent dive into “artificial intelligence” or “AI.” Like earlier AI cons going back to IBM Watson, AI is the acronym for one of the greatest value destroyers in history.

Could ORCL possibly generate enough revenue and earnings to justify such an investment? Or is CEO Larry Ellison chasing the same shiny object as everyone else in techland? Call it reflexivity or smallmouth bass in June, same difference.

Recent massive capital expenditures for AI infrastructure have indeed impacted near-term free cash flow at ORCL, yet supporters of the stock say that its cloud growth is strong, with a huge backlog of contracted future revenue, signaling long-term profitability, even as it invests heavily in data centers. But are large language models (LLMs) really going to produce LT profitability for ORCL and other tech giants betting the farm on AI?  We are a large seller of that notion.

In a fascinating interview with the FT over the weekend, former Facebook (FB) AI guru and admitted wine connoisseur Jan LeCun argues that LLMs are useful but fundamentally limited and constrained by language. An electronic parrot, like we said. To achieve human-level intelligence, you have to understand how our physical world works too, LeCun argues, and thus LLMs are ultimately a dead end. “Intelligence is really the thing that we should have more of,” he argues. “We suffer from stupidity.”

Regardless of how you feel about the investment prospects of AI, what we can say is that the AI-induced hype surrounding stocks like ORCL and Nvidia is starting to wear thin with many investors. But perhaps more concerning is the fact that there are growing numbers of examples of where AI is enabling fraud and costing investors money – a lot of money.

AI & Mortgage Fraud

President Donald Trump said on social media Thursday that he is directing the federal government to buy $200 billion in mortgage bonds, a move he said would help reduce mortgage rates at a time when Americans are worried about home prices.

In fact, the housing GSEs or even the Treasury must hedge their portfolios, thus the net impact on mortgage rates of MBS purchases will be zero while increasing taxpayer risk. We discussed this last year in National Mortgage News ("Reviving GSE MBS purchases would repeat the Fed's mistake.") Also, neither of the GSEs has the cash liquidity to fund the purchases (h/t Dennis H.). We remember when former FHFA Director Sandra Thompson retired the term debt of the GSEs. We asked her: Don't you think they'll need the funding? She said "Nope, don't need it." Somewhere Thompson is laughing.

Only the Fed under Janet Yellen was dumb enough to buy $3 trillion in MBS, costing the taxpayer hundreds of billions in losses and without really affecting mortgage rates at all. Yellen's "operation twist" was another fiasco, where the Fed bought longer-term Treasury bonds and simultaneously sold shorter-term ones to lower long-term interest rates and stimulate borrowing. LT interest rates did not fall and the Fed lost tens of billions because of the duration mismatch.

A mere $200 billion from Donald Trump is a rounding error, pure populist political pulp that will not impact home affordability at all. In fact, if the folks in the White House asked Treasury Secretary Scott Bessent, he'd tell them that mortgage spreads are  already pretty tight. But all that the White House cares about is the November midterms, thus look for the level of silliness to increase.

Getting Fannie Mae and Freddie Mac to repurchase their own debt is a truly idiotic idea that has been circulating around Washington for some time, but President Trump is getting more progressive by the day. In fact, the childish suggestions coming from the Trump White House on housing may continue to push LT interest rates higher. Yet the carnage caused by AI in the world of mortgage lending may be worse than any yowling in Washington. Consider the rising incidence of fraud in US mortgage lending.

AI is being used by dishonest borrowers, and sometimes even crooked loan officers and whole lenders as well, to alter documents and create completely fictitious borrower identities. These techniques help criminals to bypass traditional security measures and automated bank verification systems for income and employment. In essence, the entire traditional process for income verification has been rendered obsolete, putting lenders, the GSEs and investors at risk. Even documents like the IRS Form 4506-C are being falsified by unscrupulous lenders using AI.

When Plaid was founded a decade ago, we initially thought the idea of giving vendors access to your bank data was outrageous. But today, in the world of AI and loan fraud, it may be essential. WGA just invested in a private firm that supports non-QM lending with income and asset analysis using direct source data (Plaid, Finicity, MX), not PDFs. In fact, the poor old PDF is muerto as a source document in lending. Consider a short list of some of the threats facing US banks and lenders.

Synthetic Identities: Criminals use AI to generate convincing, yet fake, personal identities by combining real and fabricated information (e.g., mixing a real Social Security number with a fake name and address). These "synthetic identities" are then used to open fraudulent accounts or apply for loans with no intention of repayment, making them difficult to track as there is no real victim to report the fraud.

Forgery of Documents: Advanced AI tools can produce authentic-looking fake documents such as pay stubs, bank statements, tax returns, and identification documents (passports, driver's licenses). These forged documents are used to inflate income information, hide existing financial obligations, or verify the fake identities during the loan application process, bypassing verification systems.

Deepfakes: Fraudsters use AI to create highly realistic audio and video impersonations ("deepfakes") of legitimate individuals, such as loan officers, real estate agents, or even the borrowers themselves. This technology can trick bank employees or automated systems relying on voice or video verification into making unauthorized changes to loan terms or misdirecting funds.

Social Engineering: Generative AI allows scammers to create large volumes of highly personalized and grammatically perfect phishing emails, texts, or social media messages that mimic the tone of legitimate companies or individuals. These messages are designed to trick victims into revealing sensitive personal and financial information needed for loan applications or account takeovers.

Robot Attacks: AI bots can automate large-scale credential stuffing attacks, rapidly testing stolen login information across multiple websites to gain unauthorized access to accounts that can then be used for fraudulent loan applications.

During 2026, we think that the cost of fraud in the consumer lending process is going to become an urgent priority for lenders and also regulators and government-sponsored entities that insure loans like Fannie Mae and Freddie Mac. The prospect of loan repurchase demands by the GSEs or private investors as loan default and also loan defect rates rise is a significant risk factor for banks and nonbanks alike in 2026.

Annaly Capital Management (NLY)

While we are focused on potential risks from lenders, we remain very sanguine about one of our core portfolio holdings, Annaly Capital Management (NLY), even as spreads tighten. NLY invests primarily in mortgage backed securities (MBS) and also mortgage servicing assets, an area where WGA LLC is actively involved. 

Among the REITs, NLY was early recognizing the value of negative duration mortgage servicing assets which also have cash flows and valuable optionality. Imagine that. “We remain constructive on the opportunity in Agency mortgage REITs, mainly given the support for dividends if interest rate volatility remains this low, and particularly if expectations stay focused on the Fed cutting interest rates further this year,” notes Eric Hagen at BTIG. He continues:

“But on a relative basis, with MBS spreads versus Treasuries near multi-year tights around 115 bps, we're preparing for a little more upside in NLY if spreads grind tighter, and we expect it could show a more stable stock valuation if spreads widen back out if interest rate volatility is resurfacing. We think the higher risk scenario for NLY would involve a plummet in long-term interest rates, which could expose the prepayment sensitivity in its MSR and non-QM portfolios. It's not currently our expectation for MBS spreads to revisit the absolute tights near 75 bps, like we saw at the end of 2021, although our outlook embeds some belief that the Trump Administration will try to orchestrate lower mortgage rates, especially if it dovetails with broader policy objectives surrounding a relisting of the GSEs.”

Finally, we must close with the happy news that JPMorgan Chase (JPM) has agreed to take over the loans underpinning the Apple (AAPL) credit card portfolio from rival Goldman Sachs (GS), extricating it from one of the last businesses related to the ill-fated excursion into retail banking. In Q3 2025, GS had a net loss rate 2x the average for Peer Group 1 and orders of magnitude above Morgan Stanley (MS) and other asset gatherers as shown below.

Source: FFIEC

Hopefully JPM cut a better deal with AAPL than did GS CEO David Solomon, who paid a considerable price for the operational and financial expenses from this modest sized $20 billion portfolio. JPM's growing credit card book was about $240 billion at the end of Q3 2025. “This transaction substantially completes the narrowing of our focus in our consumer business,” said Solomon in a statement.

In our next issue of The Institutional Risk Analyst, we’ll be looking at the universal banks led by Goldman Sachs (GS) and Morgan Stanley (MS) in front of next week's earnings. We’ll also provide our thoughts on financials for 2026 and review the positioning of our portfolio for the New Year. 

The Institutional Risk Analyst (ISSN 2692-1812) is published by Whalen Global Advisors LLC and is provided for general informational purposes only and is not intended for trading purposes or financial advice. By making use of The Institutional Risk Analyst web site and content, the recipient thereof acknowledges and agrees to our copyright and the matters set forth below in this disclaimer. Whalen Global Advisors LLC makes no representation or warranty (express or implied) regarding the adequacy, accuracy or completeness of any information in The Institutional Risk Analyst. Information contained herein is obtained from public and private sources deemed reliable. Any analysis or statements contained in The Institutional Risk Analyst are preliminary and are not intended to be complete, and such information is qualified in its entirety. Any opinions or estimates contained in The Institutional Risk Analyst represent the judgment of Whalen Global Advisors LLC at this time, and is subject to change without notice. The Institutional Risk Analyst is not an offer to sell, or a solicitation of an offer to buy, any securities or instruments named or described herein. The Institutional Risk Analyst is not intended to provide, and must not be relied on for, accounting, legal, regulatory, tax, business, financial or related advice or investment recommendations. Whalen Global Advisors LLC is not acting as fiduciary or advisor with respect to the information contained herein. You must consult with your own advisors as to the legal, regulatory, tax, business, financial, investment and other aspects of the subjects addressed in The Institutional Risk Analyst. Interested parties are advised to contact Whalen Global Advisors LLC for more information.