Important Credit Lingo:

Results:

Revenue missed estimates by 0.9%. It also missed $4.91 EPS estimates by $0.28 and exactly met 2.54% net interest margin (NIM) expectations. Finally, it missed 1.20% return on asset (ROA) estimates by 6 basis points.

Credit Data:

Outlook & Call Commentary:

The charge-off rates look pretty good. Things are mostly stable on a sequential basis and improved Y/Y (which excludes seasonality). The same is true for delinquency rate data (typically a meaningful Q3 to Q4 step-up in auto delinquencies). So with that in mind, and based on the positive economic commentary below, why did provisions sharply spike higher? What are they seeing? In a word, Apple. The company had to add $2.2B in front-loaded provisions to add the Apple Card portfolio to its balance sheet. Portfolio growth also led to provisioning growth. Excluding that Apple item, which is entirely based on accounting regulation and not concern over Apple Card repayment trends, provision levels look great.

And for 2026, they see card services net charge-off rate remaining wonderfully resilient. This is expected to come in at 3.40% vs. 3.31% Y/Y and 3.34% two years ago. Great stability in their outlook, which is supported by the optimism in the call quotes below:

"Consumers and small businesses remain resilient. We continue to monitor leading indicators for any signs of stress. And despite weak consumer sentiment, trends in our data are largely consistent with historical norms and we are not currently seeing deterioration. Across income groups, debit and credit sales volume continued to perform well, up 7% year-on-year. " – CFO Jeremy Barnum 

"Oveer the next 6 months and even a year, the backdrop is pretty positive. Consumers have money. There's still jobs, even though things have weakened a little bit. There is a lot of stimulus coming. Deregulation is a plus in general... I do think the rising tide is lifting all boats a little bit… Geopolitics carries a big amount of risk...deficits are large... but we're comfortable we can build our business." – CEO Jamie Dimon

Results:

Bank of America beat revenue estimates by 2%, beat $0.96 EPS estimates by $0.02 and beat 2.00% net interest margin (NIM) estimates by 8 bps.

Credit Data:

Outlook & Call Commentary:

Just like JP Morgan, the data below looks quite resilient. Delinquency and NPL data looks solid on a sequential basis and a bit more solid on a Y/Y basis. NCO data looks good, and provisions (without JPM’s Apple headwind) were also good. There’s nothing at all alarming in this data or in the company commentary.

“It was a pretty good environment as we moved through 2025. Consumer spending grew 5%. Account balances for that broad base of U.S. consumers were stable through the year. Delinquencies and charge-offs improved in 2025 for consumer credit. Unemployment in the market remains stable… This strong consumer health bodes well for the continued improvement in growth in 2026.” – CEO Brian Moynihan

“For corporate commercial customers, they had a good profit year as the tax law settled in, the tariffs appeared to be manageable and deregulation kicked in. They had good credit quality and good money movement activity as we moved through the year. Our world-class research team has the global growth rate for GDP at 3.4% in 2026 and U.S. at 2.6%.” – CEO Brian Moynihan

Moynihan responding to an analyst pushing back against their rosy 2026 outlook:

“Is there a parade of horrible things that you can start rattling off? Yes. But you sat there and asked last year too. Those same dimensions were in place 2 years ago, 3 years ago, 4 years ago. What we see differently is the momentum in the market, the AI capital markets activity and the consumer spending is 150 basis points higher than it was back then in our customer base. Credit losses are among the lowest in our history going back 50 years. These are good setups, but we're always worried about what could happen next, as you say, and that's why we did the stress testing.” – CEO Brian Moynihan

“Provision expense in the quarter was $1.3 billion and mostly matched net charge-offs. Focusing on total net charge-offs looking forward in the near term, we expect continued stability in total net charge-offs, given the mostly benign consumer delinquency trends and low unemployment data, the continued stability of C&I and reductions in our commercial real estate exposures… in addition to the consumer delinquency statistics, note the modest changes in other stats for both our consumer and commercial portfolios.” – CFO Alastair Borthwick

“Credit performance remains strong… Consumers continue to be resilient as income growth has generally kept pace with increases in inflation and debt levels. Our nonperforming asset ratio declined modestly from a year ago and increased 3 basis points from the third quarter driven by higher commercial real estate and commercial and industrial nonaccrual loans. The drivers of this increase were borrower specific, and we did not see any signs of systemic weakness across the portfolio… at this point, what we're seeing in what's happening across the consumer base is just very consistent activity. “ – Wells Fargo CFO Michael Santomassimo

“The global economy has powered through many shocks over the past few years, creating optimism and confidence that economic growth is poised to continue. With inflation now at normal levels globally, almost every central bank is becoming more accommodating. And while the labor market in the U.S. has softened, capital investment remains strong, especially in tech. And it's the combination of that CapEx, the health of the consumer, the tax bill benefits and anticipated rate cuts that should be enough to sustain growth.” – Citi CEO Jane Fraser

“It's worth noting that across our U.S. cards portfolios, delinquency and NCL rates continue to perform in line with our expectations.” – Citi CEO Jane Fraser

The data and commentary were quite positive across the board. It’s easy and data-driven to feel optimistic about 2026 economic health, but we must stay laser-focused on unemployment trends to see if that rains on the parade. Not happening yet. Will keep obsessively monitoring. For now, the backdrop is constructive and supportive of strong 2026 growth.

It has been an ugly few months for most software names. We’ve seen considerable multiple contraction as product releases from OpenAI push some to believe these businesses are becoming antiquated and past their prime. Anthropic's “Claude Cowork”, which launched this week as a user-friendly agent that can automate digital workflows for consumers, amplified those fears. I’d like to split my opinion on the matter into two separate buckets.

Bucket #1 – Platforms:

First, are the companies I view as scaled platforms. These firms have ample access to proprietary data, strong R&D engines and an ability to lace AI throughout a broad suite of products in a way that’s cohesive, value creating and easy-to-use. I think some great examples include ServiceNow, CrowdStrike, Microsoft & Palo Alto.

At the end of the day, AI agents and models are a highly innovative version of software. Folks speak about this being some brand new and unfamiliar thing for these companies. But? Many of them have been AI native for many years. It just used to be called machine learning instead. At the end of the day, AI is more rapid data processing and more work automation. That’s why models seem so magical. They do the exact same work we’ve been doing for years… using a tiny, tiny fraction of the time to do it. That is exactly what these firms have been trying to do since their inception, and I view AI as an accelerant in their pursuit of doing so… not a death sentence.

These enterprises can tap into world-class models and build highly valuable agents themselves (and they are). More importantly, they can train these agents with more access to high-fidelity and perfectly relevant data than an AI generalist can. And most importantly, they can use this data and decades of experience to teach agents all of the unique edge use cases prevalent across every little nook and cranny in the space.

It’s like the GPU vs. TPU debate. GPUs are amazing at way more things than TPUs. But TPUs can do a much smaller subsection of work better than GPUs can. Well? Claude Cowork is really good at more things than something like CrowdStrike’s Charlotte AI. But? I am very confident that Charlotte will continue to be leaps and bounds better than Anthropic for the specific, granular use cases that CrowdStrike has spent a generation mastering. Maybe Anthropic and OpenAI can do 20% of that work really well. But they can’t come close to 100% of it. Incumbents can. To me, as long as these incumbents are innovating at a solid clip (which everything I own I think is) the same companies that led the software boom in 2020 will benefit from AI too. That hasn’t happened yet and the stocks tell us investors don’t have confidence that it will. I believe they are wrong and that patience will be rewarded. I continue to lean in.

But what about the seat-based risk? What about companies that sell licenses on a per customer basis such as ServiceNow? Won’t agents lead to a lot less seat-based demand as autonomy displaces human workforces?  

Two things in response to those questions. First, that’s not happening. All of these companies are constantly asked about it and they don’t see any material headwinds. We see all of the layoff news from megacap tech companies and that has not been a material seat-based growth headwind. It could come... but I find this encouraging.

Second is something I talk about all the time; agents are just another asset or machine-based identity to protect. These agents are exponentially growing in size and that tailwind should comfortably surpass the non-existing seat-growth headwind everyone is worried about. The AI-based assets collect data, access applications and complete complex work a lot faster than we can. They can do more. They can build more. They can act faster. That all supports demand for software and applications that these agents are interacting with on a daily basis. It doesn’t remove the demand… it amplifies it. And not just that, but it gives them the opportunity to build their agents (which again they are) to monetize and to create overarching dashboards to help customers track, monitor, orchestrate and deploy these intimidatingly capable assets.

All of this is to explain why I’m leaning into my software holdings so aggressively right now. I see deals. I see great companies.

Bucket #2 – Point Solutions:

This is where the conversation gets less positive. The advantages scaled platforms have is in better data and more ways to use that data in innovative ways across a large base of interoperable products. That greatly favors larger incumbents in my mind. There are some smaller companies successfully cross-selling enough to eventually be a scaled platform (I think Rubrik is a good example), but there are many more that aren't. Okta is a good example of a company that has struggled to expand beyond its core and has left itself more vulnerable to AI disruption. SentinelOne, Sprout Social and Docusign are more examples.

Anthropic will have an easier time closing edge use case gaps if there aren’t as many of those use cases to address and if there’s not as much proprietary data boosting the competing point solutions. That’s where to worry, in my opinion. But? Mr. Market has decided to worry about pretty much every software company and has thrown the babies out with the bath water. That creates opportunity.

Thoughts on the Last Few Months:

I think the last few months have been among the most frustrating for software-dominant investors (like most of us are) in a long time. I know that sounds weird with benchmarks at all time highs. But that's part of the reason why. For competitive people (like I am), the absolute losses are less irksome than the relative underperformance vs. benchmarks. It's less frustrating when things are falling alongside benchmarks, but software is nearing a 20% correction while profit estimates haven't really moved and the S&P hasn't budged at all. Since late October, IGV is down 17% and the S&P 500 is up 1%. 

Chip designers led S&P higher for a long time and then essentially passed the torch to networking winners who then passed the torch to memory darlings, dragging the benchmark higher and higher. Pretty much everything else besides a few pockets of high-beta have sold off a lot. We're human. It's so easy and natural for those highly exposed to software (like myself) to be at least a little annoyed. We know we're going about stock picking the right way and expect to be rewarded for it. Profit estimates aren't falling so why are stock prices? It's a fair question to ask. These are fair feelings. I have them too. And while I have them, I don't let them control my actions. I zoom out. I remain confident that I will be rewarded over the long haul (like I have been) if I do this the right way by fixating on premium, durable profit growth at a fair price.

Decades of outperformance will entail strings of underperformance along the way. Look at Apple, Microsoft or any other megacap and you see periods of 25-50% drawdowns along the way to massive growth.  I don’t think my holdings are selling off being at risk of heavy negative profit revisions. That means this software selloff is driving a ton of multiple contraction and boosting risk/reward. It means the spring is coiling and will uncoil at some point – timing entirely uncertain. I will let myself be frustrated. It's just so important not to get too caught up in those short-term swings, as they feel monumental in real-time despite being irrelevant over the long term. And while I need to let myself feel frustrated (because I'm not a robot), I also need to force myself to be excited and take advantage where I know I should. And I carefully, slowly will.

Gemini was selected to power Apple’s Next-Gen Siri. This simply adds to the positive fundamental momentum for Google's AI division. In the span of months, we've gone from fearing Google can't be Apple's default search partner to now knowing they can be while also learning they'll be the model partner for this highly important release. This will simply amplify Alphabet's distribution leadership, fortify its data leadership and accelerate the innovation engine of its leading models. That's a lot of leadership, and this news merely supports it.

Doug Thompson has been the COO of Texas Roadhouse for nearly 25 years. He is now joining Cava as the company's new COO. Quite the headline. Cava is about 2/3 the size of Texas Roadhouse and Thompson keeps the same title. We don't know what kind of raise he's getting (if any), but this is a good hire. It almost reminds me of when Lemonade hired USAA's Chief Claims Officer as their own about 4 years ago. The gap between Cava and Texas Roadhouse in terms of market cap and reputation is certainly smaller than LMND vs. USAA in 2021, but this is still encouraging news in my mind. Thompson is a highly respected executive in the industry and clearly sees Cava as a good opportunity.

There's an article from the Korea Times about Coupang's Harold Rogers "leaving Korea" to avoid police questioning. I think that's pretty click bait-y. He just wrapped up several days of hearings and has meetings in other countries to attend. These are pre-scheduled. The show must go on. I don't love this, but I'm not upset about it either. They can't just spend the rest of their lives accepting endless requests to be asked the same questions over & over again. Need to push the puck forward and he has participated in plenty of hearings already. The government also told Coupang to stop publishing their own internal findings on the investigation while they verify claims.

On the analyst side of things, Deutsche Bank upgraded Coupang to a buy with a $25 price target, while Bank of America reiterated their buy rating and set a $32 price target. Both institutions think this company will weather the blowback from the data breach and lean on their unmatched product breadth and service to normalize the GMV and customer growth curves over time. They think the risks from the event are priced in. Nomura downgraded it to neutral due to risks from the data breach and profit headwinds from the customer concessions.

Leadership reported today that the company crossed 3M customers. That means they've added 131K customers since the end of Q3. At the same time, this announcement came on January 14th and it's unclear if the milestone was reached on that day or earlier in the month. If it was reached on January 14th, then customer growth for Q4 likely slightly missed customer growth estimates. If the achievement happened closer to the New Year, customer growth will be roughly in-line with expectations. Regardless, this represents slower net new Y/Y customer growth than we've grown used to over the past year, which makes sense. Lemonade is now lapping the period when it aggressively leaned back into growth spending, which is why customer additions nearly tripled Y/Y during Q4 2024.

While the news could be taken as a very modest disappointment, the extremely volatile stock reacted positively to it during an ugly day for markets and high-beta. I don't like paying too much attention to or assigning too much meaning to short-term price action, but I did find that comforting. 

DraftKings is deepening its presence in Brazil through a new partnership announced this past week with Minas Gerais Football Federation. That is the global football governing agency for Minas Gerais, a Brazilian state where 10% of their population lives. DKNG is now the exclusive 2026 data streaming and fan engagement partner for their soccer league's 2nd division. To me, this is a precursor to their deeper entry into the market whenever they finally get a license. They're making sure their brand and pieces of their product are deeply embedded into the highly important Brazil market so they don't have to work so hard for market share when they're allowed to operate. The company is currently in the "final backlog" of the regulatory bodies in Brazil and is optimistic about its chances. I expect more partnerships like this to be announced and for traction to come as soon as DKNG gets the green light to launch its product. For context, firms like Wells Fargo see a path to DKNG having a top three market share position there. Speaking of Wells Fargo, that firm upgraded DKNG to outperform after a few months on the sidelines.

Bloomberg published an article on Friday that I found extremely noisy. In it, they talked about sports books losing to prediction markets. Their two big pieces of evidence lacked context and a basic understanding of the reasoning. First, they said Kalshi enjoyed more volume during the NFL playoffs. Well... so do sportsbooks. NFL playoff games are among the highest-volume events for these companies. Considering prediction markets are vehicles for sports gambling, this is not at all surprising. And DKNG has delivered 9% Y/Y volume growth in New York over the last 6 weeks.

Second, they talked about revenue falling 40% Y/Y this past week in New York as a cause for concern. Wrong again. Timing of the biggest volume events with the most enticing matchups and especially outcome luck can have big impact on weekly revenue. On event timing, the Buffalo Bills were heavily home favorites in last year's wildcard round against the Broncos. That always naturally attracts a ton of volume from its home state of New York (where this data is from). They were road underdogs this year. That alone is a big headwind. On the second note, if there's worse outcome luck, that means volume leads to less revenue and amplifies the headwind. It had a 5% hold rate (revenue/bet volume) last week vs. an expected hold rate of 11%. Calling this a red flag is as silly as saying "the Rams beat the Panthers so the investment case has changed." Seriously. This is why it's so important not to focus too much on a single week.

What will actually concern me? If Y/Y volume growth trends meaningfully slow on a several week basis. One week is not close to enough to concern me, especially with all the positive anecdotes we've seen about little prediction market competition customer overlap. We'll see how the coming weeks and months look.

Meta fired 1,500 people in the Metaverse division. China is also making their planned Manus acquisition more difficult with an investigation into Meta. Meta is already banned in that nation. And Manus has already moved to Singapore and shuttered all Chinese ties. I don't think China can do anything to stop this from happening. 

Next, some took the Meta compute announcement as Meta's official entry into GPU renting. Zuck did not say that and I think many on social media took a leap when they shouldn't have. He is merely organizing and standardizing their approach to data center expansion with some re-aligned talent and a combination of all related teams (data center, energy procurement etc.). They hinted at exploring business models, but to me that's tied to using partners like Blue Owl for financing... not renting GPUs like AWS does. I think Meta is open to doing that down the road, but has so much more to do internally before it has excess compute to provide the field. 

Meta sunsetted some interest-based targeting variables in its ad campaigns as they move to the Andromeda AI model. That transition has come with some latency and campaign structure issues that I expect to be quickly resolved. The change will also entail tighter API-level access to its lucrative data, which some view as a way to prevent other companies from using META's valuable assets for their own campaign-building services. Companies like AppLovin (APP) have seen several short reports written on their unauthorized use of data from companies like META. APP denies it every chance they get, but I suppose we'll find out if they were reliant on META in the coming quarters, as that's no longer an option.

Finally, Meta named Dina Powell McCormick as its new President. She was a partner at Goldman Sachs and the President at BDT & MSD Partners ($50B AUM). She was also an advisor for Trump.

CFO Matt Skaruppa is out as CFO. He will be replaced by an existing DUOL board member and ex-Vimeo CFO Gillian Munson. This isn't a positive in my mind like (for example) Calvin McDonald stepping down as LULU's CEO was. This guy is not the problem for the Green Owl. Duolingo's issue isn't financial planning. It's struggling social media data, lackluster user trends and an "AI-first" messaging blunder that they haven't recovered from. It's them needing to sacrifice their profitable growth because their product isn't teaching people as well as it needs to. A new CFO doesn't change any of those things.

As part of the news, Duolingo announced preliminary results that were about in line on daily users and bookings. But again... they talked about near-term tradeoffs to "teach better" and forgo better financials. I think that's them preparing everyone for an underwhelming 2026 guide.

This news does not make me any more interested in adding this back into the portfolio. If anything, I'm less interested. Still no recovery in social data.

The Senate Committee on Banking, Housing and Urban Affairs postponed their vote on the Digital Asset Market Clarity Act. Among other things, the act was supposed to include an amendment that would have created another loophole for non-banks to emulate stablecoin yield. Currently, the Genius Act only allows chartered banks to do this “yield mimicking” through a tokenized deposit loophole. Others like Coinbase and Robinhood can't, so they were very excited about this specific piece of the legislation. The amendment was abruptly pulled this week, likely due to lobbying from the banking industry. That led to Coinbase publicly removing support for the bill and the overall delay. 

If rules continue to only allow banks to effectively imitate customer yields paid on their stablecoin holdings, that would be good news for SoFi's competitive positioning. And despite this, SoFi is actually supporting the Clarity Act and that amendment. They say they think that will accelerate stablecoin adoption and be a net tailwind, but I think there's another reason. I do not view the permanent outcome of only banks being allowed to offer yield as likely. I think SoFi agrees and sees this change as inevitable at some point. This vote was delayed, not canceled. 

I do think this will stay noisy but believe the non-banks will eventually figure out a way to gain permission to form their own loophole. I could be wrong but that’s my best guess. It’s too important for their competitive positioning to do anything but kick, scream and lobby until they figure it out. I don’t see this yield edge staying exclusive to banks forever and don’t see this as a durable source of SoFi differentiation. Fortunately, they have many other, more structural sources of competitive edges to enjoy.

There's a new trade deal between the EU and LatAm that should be a small positive for MercadoLibre's business. It will lower tariffs and trade barriers between the two nations, enabling cheaper inventory procurement. It should also support GDP growth in Brazil, Mexico, Argentina and the other markets where MELI operates. 

Rubrik announced "CXO Visionaries" as a group of Global 2000 executives to collaborate on securely accelerating AI adoption. Executives come from companies including Domino's, PLUS Malaysia, Renown Health and more.

China is forcing companies to stop using CrowdStrike, Wiz (Alphabet buying them), SentinelOne and other cybersecurity firms. These products are already heavily, heavily restricted there and I don't believe any of these companies have large established teams or businesses in China. None of them even have offices there. It's not a focus and I don't think this news will be material for any of them.

Nvidia H100 (introduced in 2022) rental prices have risen from just under $2 to $2.34 per hour since November. That's an encouraging reversal of a consistent decline and bodes very well for chip depreciation intensity.

Output Data:

Inflation Data:

Consumer & Employment Data: