Short Thoughts: February 2, 2026
Earnings are in bloom. Market developments left and right deserve comment.
I cannot cover all the interesting topics out there.
24-hour trading markets and the volatility at present make it difficult to write with any depth before the situation significantly changes.
I do have fully-formed thoughts in some depth, and I will run with those.
Some quick hits will have to suffice for the remainder. First, the quick hits.
Fannie Mae & Freddie Mac, my thesis has not changed. I am a buyer on Fannie in the $7s and Freddie in the $6s. I continue to stand behind what I wrote in Fannie & Freddie: Toxic Twins No More No More. I will publish a Recurrence post on this article sooner than later.
Warsh is Trump’s choice for Federal Reserve Chairman. Powell has not agreed to leave yet. The Supreme Court still has a decision to make. I have little to say on this matter.
Lululemon Athletica (LULU) has pulled back and I am adding to my position a bit. The load up the truck price remains down in the $150 range. I have been meaning to publish a post concerning LULU, but other topics have taken precedence. I continue to believe at current prices, the company is a tempting target for a founder-led or other PE buyout.
Now, let’s address in much greater depth crypto, precious metals, the AI buildout’s accounting bugaboos, Molina Healthcare, and GameStop’s potential M&A strategy, including possible targets.
Bitcoin’s “Come to Jesus” Moment
Much is made of Bitcoin falling to levels that now stress various Bitcoin treasuries and especially Michael Saylor’s Strategy (MSTR). At this writing, Bitcoin is falling, having touched lows hit during the Liberation Day tariff scare before bouncing a little bit.
Bitcoin’s fall is different than most securities or other assets in that a sustained fall in price could itself set in motion a death spiral leading to massive value destruction.
Bitcoin has been exposed as a purely speculative asset, and is not near the debasement trade hedge that gold and other precious metals are.
There is a reason for this. Increasingly, investigation of Bitcoin’s public ledger by law enforcement has dramatically reduced Bitcoin’s share of illicit trade globally. Bitcoin is also not amenable to high-velocity criminal uses, which has reduced its criminal usefulness, removing a floor on its utility.
Analysts and law enforcement still find dark corners likely holding Bitcoin that cannot be penetrated. But the total use case in illicit trade by most accounts is minimal at present.
This decline in illicit use has somewhat paralleled Bitcoin’s rising acceptance as a portion of corporate treasuries – nearly 200 public companies now hold Bitcoin. Many Bitcoin bulls point to this as a new floor on utility and price.
As well, merchants and charities have begun accepting Bitcoin in lieu of cash/credit. Unfortunately, the rising value of Bitcoin makes spenders less likely to spend it. There is a bit of a paradox here, in that falling prices will make merchants less likely to adopt it. The dynamic here kills velocity/liquidity for market transactions.
So, Bitcoin bulls estimate that over 70% of Bitcoin supply is now held by long-term holders, generally as a reserve asset or for settlement purposes rather than trade. A good part of the reason for this is that Bitcoin has been seen as an idiosyncratic asset rising in value over time but theoretically uncorrelated with broader market moves.
However, there is nothing permanent about treasury assets. What’s more, treasury assets generally must be marked to market and included in financial reporting. If an asset falls in value, risk managers will bring pressure to sell the asset.
Correlation with the securities markets tends to rise, especially when securities markets fall. Newer Spot ETFs involving crypto are speculative assets by their nature and only increase speculation. Equitization of crypto through ETFs also increases correlation with stock markets. The correlation of Bitcoin with the S&P 500 lately has approached 0.50.
Some say the mean Treasury cost basis in Bitcoin is around $80,000, and the average Spot ETF cost basis is around $84,000.
Theoretically, risk-mitigation/liquidations will kick in aggressively when loss positions start to grow. Empirically, liquidations have already risen.
For instance, Blackrock’s Bitcoin ETF (IBIT) on January 30th set a new record for largest single-day outflow - $528.3 million - from a single fund since Bitcoin ETFs were allowed. Institutions liquidated $1.1 billion worth of Bitcoin over three straight days to end the month.
In fact, the four largest single day Bitcoin ETF outflows have occurred since November 20, 2025. Three of them occurred in the last 10 days of January 2026.
On January 29th in particular, there were $1.75 billion in crypto-wide liquidations, with up to 95% of the liquidations being long positions.
That same day, Bitcoin ETFs saw a large one-day reversal in fund flows. BlackRock (IBIT), Fidelity (FBTC) and Grayscale (GBTC) accounted for 92% of the exodus. Such bitcoin ETF concentration is a liquidity trap. Rather than a brake or stabilizing force, ETFs are the new Bitcoin panic button.
Without a black market “floor,” and with both treasury and merchant use not providing an effective floor, Bitcoin has been left to speculative flows at a time when it has never been easier to speculate on Bitcoin.
There is no organic use case reason for Bitcoin to slow or stop its descent. The blind spot in the bull case was that new uses of Bitcoin as treasury assets or ETF assets somehow meant those positions were exempt from risk-management and market forces.
The last 48 hours of January saw 270,000 traders and $2.5 billion worth of leveraged bitcoin positions liquidated. Coinglass.com provides a helpful chart.
As February gets underway, volatility is continuing.
Gold & Silver Catch a Crypto Cold
As Bitcoin is held by the paper hands of treasuries and speculators, when Bitcoin fell below key levels, speculative algorithms kicked in, de-risking by selling profitable speculative positions in gold and silver and contributing to the precious metals’ fall.
These speculative positions are in tokenized silver and gold futures that trade on crypto exchanges instead of traditional venues like COMEX. This type of futures contract is perpetual and allows leveraged exposure to precious metals 24/7 using crypto assets as collateral. On these exchanges, a trader's crypto and metals positions often share the same crypto collateral pool. This is called a portfolio margin account.
Gold is the most tokenized metal. Platinum is tokenized too. These are called “on-chain” derivatives as they use Blockchain.
The problem is tokenized metals futures are not backed by actual physical metals - and trade with far higher leverage and no position limits - so can overwhelm trading in physical metals, as happened in late January.
Surprise, surprise, sky high leverage on these crypto exchanges due to rising metals prices meant that as the crypto positions fell, margin calls meant the tokenized metals had to be sold. This is a collateral death spiral. It was reported that tokenized silver futures liquidations actually exceeded Bitcoin liquidations on one crypto market called, ironically, Hyperliquid.
This in turn forces liquidation of traditional speculative positions in metals. Arbitrageurs who keep prices aligned across venues transmit the leveraged synthetic asset liquidations into selling on traditional exchanges.
The collateral death spiral continues as, back in the real paper world, the Chicago Mercantile Exchange saw what was happening and increased margin requirements on silver and gold by as much as 50%. The exchange implemented this because it saw the stress, but in doing so, it caused more downward pressure in metals prices.
This forced liquidation of more collateral. Then the algorithms kicked in. It looks like up to $1 billion in precious metals were liquidated at month’s very end as a result of falling crypto prices.
The effect was dramatic with silver off 35% from its peak, gold off 12% in a day.
On January 30th, regulators shut down a small bank, Metropolitan Capital Bank & Trust. This is the first bank failure of 2026. Some are looking into whether it was caught up in the dash for cash and forced collateral liquidations described above. If so, it would be the first traditional bank failure linked, if only indirectly, to a crypto collateral death spiral.
Sickening scenarios have now come within reach. If Bitcoin falls another 10% under $70,000, MSTR will be over $4 billion in the red and would find capital markets essentially closed. Institutions would be down 15-20% on their Bitcoin holdings. Risk managers would get more aggressive.
As I write this Sunday evening, gold futures are down 6%. Silver is down 17% off Friday’s close.
These moves if sustained could result in Bitcoin breaking $70,000 because of this interconnected world of portfolio margin accounts, tokenized metals futures, and crypto collateral.
With Bitcoin at $60,000, MSTR would face an existential crisis. MSTR owns about 17% of all Bitcoin held by public companies, governments and ETFs combined.
At $50,000, miners would go bankrupt and be forced to sell their BTC reserves, tokenized metals futures would collapse into a black hole with no buyer. Physical metals may break from the trend on safe haven demand.
Or, Bitcoin can bounce off that Liberation Day low, and we can live the charade a while longer.
Blessed Fraud Recurrence Revisited
We are in the middle of earnings season, and with Tesla, Microsoft, and Meta reporting this past week, there is no shortage of news. What stood out, so far, to most was Tesla’s pivot, Microsoft’s reticence, and Meta’s spending. Readers can easily find unending coverage of these events.
I will zero in on what makes this all tick.
The depreciation thesis we discussed in Unicorns & Cockroaches: Blessed Fraud was featured in a Financial Times article recently, “Big tech’s $680bn buy-now-book-later problem.”
Hey, I recognize that shirt.
The article also included charts and commentary from Morgan Stanley’s crack accounting team led by gurus Todd Castagno, CFA, CPA and Kate Konetzke, CFA, CPA. Todd and Kate know what’s up. This is an excerpt from a report they put out last week.
Depreciation is challenging to forecast because it depends on the timing of data center construction completion and the useful life assumptions of both the shell and the chips. These assumptions are subject to higher uncertainty given the early stage of this emerging technology. Furthermore, major players, Alphabet [GOOGL], Meta Platforms [META], Microsoft [MSFT], and Oracle [ORCL] are increasingly turning to finance leases to build out their AI infrastructure, also driving expenses higher. Our software team recently reduced EPS targets for ORCL due to rising depreciation from GPUs and finance leases, read more here.
Because these companies have historically been asset-light businesses, their financial disclosures are not adequately set up for investors to see where depreciation is reported in the income statement, and most consensus models do not directly forecast adequate impact on earnings. We estimate depreciation expense using our analysts’ latest traditional capex and finance lease capex assumptions and compare it with consensus expectations for operating expense growth to assess whether consensus may be underestimating total expenses. Based on this analysis, we expect expense forecasts will need to revise higher to reflect capex spending, potentially driving margin expectations lower if revenue revisions do not keep pace.
In its recent earnings report, Microsoft showed its narrowest gross margin in three years, and the stock plummeted over the last two days of the month, wiping out $381 billion in market cap.
I do not believe this will be a blip.
Below, one sees how horrifically far depreciation expense runs behind capital expenditure. The widening chasm at each company is clear.
There is good reason for this. I had mentioned in the Unicorns & Cockroaches: Blessed Fraud post that capital expenditure categorized as construction-in-progress (CiP) does not begin to depreciate but rather sits on the balance sheet as an undepreciated asset until it is placed into service.
I expected that these companies would increasingly delay depreciation expense by categorizing finished assets as CiP.
According to Todd & Kate, Google’s CiP is 96% of Capital Expenditure. This is remarkable. These are GPU and servers and memory and cooling equipment and more. All of these very much depreciate in value every second due to the inexorable advance of technology.
When you hear stories of Microsoft and others sitting on warehouses full of Nvidia GPUs, please sleep well knowing that those are not expensed against Microsoft’s earnings at all.
As well, sleep well knowing that Nvidia has booked those entirely into revenue and earnings.
Speak of the devil, Microsoft’s CiP is not disclosed. Neither is Amazon’s. I imagine theirs are not worse than Google’s 96%. One must wonder however, if their CiP numbers would be any better given they would rather not disclose!
Oracle’s CiP is 78% of its Capital Expenditure. Meta’s is 72% of its Capital Expenditure.
I have claimed that AI buildout assets are not being depreciated enough, boosting earnings. Well, this capital expenditure into CiP is not being depreciated at all.
The corollary here is that this massive spending is running ahead of the market’s understanding of how to account for it. Management teams are well aware of this. The opportunity for fraud and manipulations is certainly enhanced by these conditions.
Accounting issues tend not to cause problems while a bubble is expanding because growing wealth papers over any concerns.
Perhaps times are changing. Just this past September, markets rallied on news that Nvidia would invest $100 billion in OpenAI. This was widely reported.
As the New York Times put it:
The Nvidia investment is the latest example of OpenAI raising money from the companies it relies on for products and services. Microsoft, which invested $10 billion in OpenAI in 2023, has made billions of dollars after the start-up paid it for computing power from its Azure data centers.
Nvidia invested in the start-up last year as well. During an appearance on CNBC on Monday, Jensen Huang, the chief executive of Nvidia, said his company would support the largest data center build out in history, which will result in the sale of four to five million Nvidia chips to OpenAI.
Nvidia shares rallied, adding $750 billion in market capitalization over the next month.
Now, Bloomberg reports that Jen-Hsun backtracked while talking to reporters on February 1st:
“It was never a commitment,” Huang told reporters in Taipei on Sunday. “They invited us to invest up to $100 billion and of course, we were, we were very happy and honored that they invited us, but we will invest one step at a time.”
Jen-Hsun let the stock market’s $100 billion interpretation stand for nearly 5 months. There is a CEO lie in here somewhere.
At the same time, The Register wrote of a report that Oracle is having trouble finding the capital to fund its data center buildout dreams. As you know, I own puts on Oracle and Nvidia.
Oracle could cut up to 30,000 jobs and sell health tech unit Cerner to ease its AI datacenter financing challenges, investment banker TD Cown has claimed, amid changing sentiment on Big Red’s massive build-out plans.
A research note from TD Cowen states that [it is] finding equity and debt investors are increasingly questioning how Oracle will finance its datacenter building program to support its $300 billion, five-year contract with OpenAI.
This year, “both equity and debt investors have raised questions about Oracle’s ability to finance this build-out as demonstrated by widening of Oracle credit default swap (CDS) spreads and pressure on Oracle stock/bonds,” the research note adds.
The research note says US banks had pulled back from Oracle-linked datacenter project lending while private operators leasing to Oracle were also struggling to secure financing, impeding the leasing option in its build-out plan.
We are seeing hesitancy, which has not been a feature of the AI bubble thus far.
That is not all. These companies – and Amazon too – are increasingly using finance leases to build AI infrastructure. This hides the expense from traditional capital expenditures as represented in the cash flow statement. These expenses are rather in the Finance Cash Flows and often not included in calculations of free cash flow or owners’ earnings.
For instance, including finance leases, Microsoft’s capital expenditure-to-sales ratio rises from 28% to 38%. These businesses are even more capital intensive now than they or their analysts admit, and the whole group’s earnings are dangerously inflated. This will be revealed to all only after the fall, if history is any guide.
Molina Healthcare
Recent news dramatically impacted share prices of Managed Care Organizations (MCO). United Healthcare, Humana, and CVS (Aetna) took the brunt of the selling and despite their Medicaid focus, Molina, Centene and Elevance all saw big drops as well. Molina, with the 90% pure focus on Medicaid, fell the least.
CNBC reported January 27, 2026:
Shares of several big-name health-care companies plunged Tuesday after the Trump administration proposed nearly flat rates for Medicare Advantage insurers.
Medicare Advantage plan provider Humana dropped more than 20% in early trading, while CVS Health shed 13%. UnitedHealth Group lost more than 19% following the Medicare rate news and after it posted 2026 revenue guidance that was worse than expected. Elevance Health tumbled about 13%, while Centene dropped more than 10%.
The proposal entails a net average payment increase of 0.09% for Medicare Advantage plans in 2027, according to a release from the Centers for Medicare & Medicaid Services, or CMS, on Monday. That number is significantly less than Wall Street analysts’ expectations that the agency would propose a rate increase of between 4% and 6% for next year.
I do not believe this news is significant for the long-term Molina Healthcare long thesis. This is also an industry that will find plenty of use for cost reduction with small language models and other innovation on the way to actual AI.
Subscriber Steven Lombardi brought to my attention layoffs at a local hospital.
Mercy Hospital in Des Moines, Iowa is not a small rural hospital. And it is already anticipating what is coming. These layoffs are just the beginning. The political pressure to raise the reimbursement rates will begin soon enough….Rural Iowa is served by approximately 82 to 95 hospitals, with 82 of them designated as Critical Access Hospitals (CAHs) designed to reduce financial vulnerability in rural areas. These facilities are part of a broader network, including 187 rural health clinics.
And I will reiterate what I have said in the past – if fraud in the system is a big part of the expense increase, Congress is going to be more aggressive finding that fraud than cutting health care for the poorest and most vulnerable.
Less fraud in Medicaid means less medical expenses and a lower medical expense ratio for insurers like Molina.
A recent Mizuho’s physician survey indicated healthcare utilization growth trends decelerated sequentially despite easier year-over-year comps, which could indicate the brutal upward march in healthcare expenses is peaking.
This will happen as pricing relief starts to get priced into annual contracts this year.
There will be plenty of political and fiscal expenditures noise – it is shaping up to be a brutal election year. Long-term, Molina is a great business with great management at a great price today.
I continue to believe if Molina’s price remains this low as catalysts start to fire, the company will be scooped up by a private equity buyer.
GameStopped
CEO Ryan Cohen has given two interviews with CNBC and the WSJ. Now, for the third consecutive business day, Ryan Cohen will give a live interview with Charles Payne on FoxBusiness at 2PM EST Monday, February 2nd.
Last week, I published Final Stop GameStop, which followed December’s The Big Short Squeeze historical piece.
Together the two posts synthesize both my experience and my expectations for GameStop under Ryan Cohen. I, however stopped short of specific recommendations for GameStop’s “transformative” M&A.
That changes now.