In case you missed it, I published my portfolio earnings season preview this week. That piece also included my updated holdings, performance and some more detail on this week’s digital advertising news.

I also published detailed Netflix and Taiwan Semi Earnings Reviews.

Needed Definitions:

Brief commentary on the data below:

“Near term, we don't expect much change in net charge-offs, as you can see improvement in both early and late-stage delinquencies from the fourth quarter. That tells us that net charge-offs could even be a touch lower next quarter on the consumer side.”

CFO Alastair Borthwick

Considering commercial NCO is typically well below consumer for BofA, the shift to commercial loans since the Great Financial Crisis (GFC) puts it in much better shape to endure any economic downturn. Generally speaking, its non-performing loan (NPL) rate was 0.55% vs. 3.75% in Q4 2009, while NCOs are $1.5B today vs. $11B then.

Despite expectations for more rate cuts than it assumed 3 months ago, Bank of America reiterated a quarterly net interest income ramp to $15.6B by the end of the year. It continues to expect about 6.5% Y/Y net interest income growth.

“Our research team, led by Candace Browning, does not currently believe we'll see a recession in 2025. However, they've lowered their GDP growth rates for 2025 and continue to see no rate cuts during 2025. As inflation gets under control, they expect you may see them in 2026.”

CFO Alastair Borthwick

“Consumer spending has been consistently growing Y/Y. Last year, it slowed a bit and then picked up in the fall. That resulted in Q4 growth of around 4%. That pace has continued through the first part of April. We note that some retailers may say that their sales are slower and others are picking up, and it really reflects the change in consumer spending behavior. But in the aggregate, the consumer keeps pushing money into the economy.”

CFO Alastair Borthwick

“Heading into Q4 2027, our average credit card consumer FICO was 50 points lower than today.”

CFO Alastair Borthwick

“Our strong position allows us to better serve our clients in times of street, which may come ahead according to our projections.”

CEO Brian Moynihan

The U.S. government is extending export bans for Nvidia. The Hopper 20 (H20) GPU that they purpose-built for that market is now banned for export. Whether we like it or not, Nvidia has the most valuable export this nation has to offer. They have the largest trade chip for this administration, and it looks like that chip will be used.  

As part of the announcement, Nvidia took a $5.5 billion inventory charge, which will have a one-off impact on GAAP profits next quarter. In terms of the revenue hit, Q1 is already over. For Q2, China was set to represent about 10% of total Nvidia revenue, with the H20 likely a big piece of that. Nvidia will work to build new chips that are less powerful and eligible for eventual export, and trade developments are also highly fluid at the moment. Nvidia CEO Jensen Huang quickly flew to China to communicate his commitment to building those chips, which I found to be a bold move amid the geopolitical tension.  With all of that said, I think it’s reasonable to assume this represents a high single-digit impact to Q2 revenue. The Q2 margin hit will be modestly smaller, as economies of scale for H20 are not as powerful as for its mainstream Hopper and Blackwell chips.

In other news, Nvidia committed $500 billion to build AI infrastructure in the USA and manufacture its supercomputers domestically.

As a reminder, SoFi has a lending segment, SoFi Lantern and a Loan Platform Business (LPB). Lending segment revenue comes from originating loans for its own balance sheet and either holding them, or selling them to capital market buyers. When borrowers are rejected, SoFi can send applicants to its loan marketplace, called Lantern, to match them with other creditors in exchange for a referral fee. That’s financial service revenue. More recently, as of Q3 2024, it now signs agreements with partners like Fortress Capital to originate loans on their behalf – right from SoFi’s app or site. This is for borrowers within SoFi’s credit bands and outside of them too. Here, servicing revenue is part of its lending bucket, with everything else (origination & other up-front fees) part of financial services revenue.

LPB gives SoFi yet another outlet for originating more loans without more balance sheet pressure, and allows it to expand to riskier borrowers without more unwanted credit risk. Some of these subprime borrowers will also eventually be great candidates for a loan. This wider top-of-funnel means SoFi can say yes to them today rather than directing them to a competitor. In turn, that makes cross-selling other products much easier and gives SoFi a better chance of securing repeat loan business down the road.

This week, SoFi announced a pair of new funding deals for its rapidly growing loan platform business (LPB). First, it inked a $2 billion contract with Fortress. Second, it finalized another 2-year $1.2 billion deal with Fortress and Edge Focus. With the news, Fortress’s total investment in LPB has crossed $5 billion.

I love these deals in all environments. It gives SoFi a way to grow its lending business through fee-based, asset-light revenue. It allows it to untether loan origination volume from balance sheet risk. The potential issue here is that capital market participation is cyclical. When macro sours, credit spreads widen, liquidity wanes and funding dries up. That’s why the news coming this week is so encouraging to me. Bond markets have already shown warning signs in recent weeks… credit spreads have widened and become volatile.… geopolitical chaos has already struck. And? SoFi is getting more than a doubling in commitments from its most important LPB partner.

Where does this resilience come from? It goes back to an idea we cover constantly. SoFi is not your average fintech; it does not wade in the waters of subprime credit. Instead, it fixates on ultra-prime customers with FICO scores over 770 and income over $150,000/year across loan types. Affluent borrowers are always more resilient borrowers across cycles, which is why SoFi stayed consistently below its life of loan loss rate target through the most aggressive hiking cycle of the last few generations. It’s also why capital market loan deals have uniformly occurred at gain on sale margin in excess of the fair value markings everyone loves to bicker about. And that’s likely why capital markets are remaining open for this lender. Fortress would not be pouring billions of dollars into this platform if repayment trends were bad. They have better, more granular cohort data to study than you, me or any sell-side analyst. They are explicitly telling you credit trends look great… and they’re telling you this with the most important thing: their wallet.

The $1.2 billion deal over the next two years is worth $600 million in annual origination volume alone. I think a revenue take rate assumption of 4% is conservative, based on past gain on sale margin trends for previous personal loan sales. Based on this, the deal is likely worth around $25 million in annual revenue for the next two years. The $2 billion deal doesn’t have a finite timeline and will function more like a funding revolver. If we assume it’s used up over the next two years, that would add another $40 million a year, or $65 million combined, in high-margin, asset-light revenue.

I loved seeing this piece of news.

Other SoFi News:

SoFi will soon likely reintroduce its crypto offering with needed regulatory clearances in place. An account on X (@gertrud_yadon) dug up code from a SoFi app update that includes SoFi Crypto.

During the first quarter, Tesla registrations in California fell 15% Y/Y compared to roughly 8% growth for the sector as a whole. As a result, market share fell from 55.5% to 43.9% Y/Y. This is either related to the ongoing Model Y refresh or political blowback from the left-leaning state against Musk’s work with the Trump administration. It’s likely a combination of both. If politics are the main source of weakness here, then it would be reasonable to assume market share trends look somewhat better in most other states outside of liberal states like Illinois, New York and a few others. But when looking at country-level data, things don’t look ideal. Overall, market share trends have flattened or even modestly declined in markets for Tesla over the last several quarters. They stopped providing the chart below in Q2 2024, but 3rd-party data says the same thing. 

Macro is absolutely impacting growth, but it’s impacting growth for everyone, and we all expect Tesla to be the relative outperformer in all environments. Market share fragility doesn’t depict that. 

In other news, Tesla is delaying the launch of its $25,000 mass-market model from the first half of 2025 to somewhere between Q3 2025 and 2026. The plan remains to enter the USA first, then China and Europe. It’s unclear if tariffs had anything to do with this news, although I find that somewhat likely.

To me, both pieces of news place more pressure on the thriving energy business and the future pieces of this firm (Optimus Robots and Robotaxis). In my view, while I deeply admire this company, I am not willing to invest in a large cap where most of the future value is based on products that aren’t yet in mass production. That is too speculative for my taste. Those willing to bet on this name will be fantastically rewarded if bold promises of market size and production ramps for these products play out. I just find risk/reward to be more compelling elsewhere, but do find myself rooting for the bulls in this case. Tesla the company is a great American growth story and I love when great American growth stories win.

Finally, there was news on Tesla delaying robotaxi part imports. The delay to the $25,000 model is rumored to be related to focusing more on Robotaxi. So this news could just mean they’re shifting the still-nascent supply chain.

There was a chart that went viral on ChatGPT racing from 19 million to 428 million monthly active users (MAUs) vs. Gemini reached 27 million during that time. It’s important to note that Gemini 2.5 Pro really flexes its muscle in code writing, while AI overviews (within Google not Gemini) are the most popular form of AI search for Alphabet, as Google has billions of MAUs. For Gemini specifically, the new 2.5 Pro Model really shines in code writing. Key performance indicators (KPIs) for code writing are not MAUs, but premium user revenue and API calls. ChatGPT is a search app, where free usage friction is non-existent and MAUs are a more relevant metric.

In other news, a district judge ruled that the mega-cap’s ad-tech business is an illegal monopoly. The company will likely appeal, but (as a bull) I do think this ruling is somewhat fair. Google’s ad network routinely routes demand to its own supply. They are present in too many pieces of the value chain, which inherently creates conflict of interest. I do think they’ll have to change some part of this business and won’t be able to actively manage campaigns for business, while routing most of their impressions to Search and YouTube. This is ok with me. Leadership has been pretty open about the fact that they’ve been actively deprioritizing the ad network and campaign-building business long before this news came. It is not a vital part of the growth engine.

The news could potentially create a more level playing field, which is good news for all open internet advertising players like Trade Desk, Magnite, Pubmatic etc.

North Carolina will likely raise its gambling tax rate from 18% to 36%. That will impact profits for the sportsbooks. I think it’s inevitable that more states will raise taxes. But for companies like DraftKIngs, at 18x forward FCF, a triple-digit, multi-year FCF CAGR and great trade war insulation, I think the margin of safety is so massive that this doesn’t matter. Furthermore, sportsbooks will probably all combat this with lower marketing or promotional spend. And as that happens, the books with the largest scale and highest unaided brand awareness should relatively benefit – FanDuel and DraftKings. The tax news I fear is progressive tax rate policy like we see in Illinois. That penalizes higher volumes with increasingly higher tax rates and punishes the big boys for succeeding. That continues to look like an anomaly, which is what I care about most here.

Per Yipit (to the Yipit employees reading this I waited until this was all over social media to share this), North American sales growth for Amazon accelerated from 8% Y/Y to 12% Y/Y. It’s possible this is related to antsy consumers making purchases before tariffs are implemented.

Uber is exploring a purchase of Trendyol Go. This is a delivery subsidiary of the Turkey Trendyol. According to investor materials, it had 2.3 million weekly active users as of last year.

Mercado Libre is apparently set to grow headcount by 33% Y/Y in 2025. That doesn’t sound like a company worried about trade wars.

Snowflake released research showing that 92% of early AI adopters are seeing compelling return on investment (ROI). 

Meta was accused of being a monopoly by the FTC. While I think the Google Ad Tech case is strong, this one is a joke in my mind. X, YouTube, LinkedIn, Snapchat, Pinterest and TikTok are all scaled competitors.

PayPal added OpenAI and a few other software development kit integrations to its platform to help build agentic workflows through its Commerce API. PayPal and TerraPay formed a new partnership for real-time transfers in the Middle East and Africa. Seaport downgraded PayPal to sell due to the trade war.

Palantir inked a deal with NATO for its Maven product. Anthropic is also using the company’s FedRamp accreditation program.

According to Counterpoint Research, Apple led in market share for the global smartphone market.

The SentinelOne employee that led to colleagues losing some federal clearances has resigned. SentinelOne had already called the clearance losses immaterial, but this likely raises the odds of the firm getting them back. Any other analysis on this would be inherently political, so I’ll leave it there.

In a talk this week, Fed Chair Jerome Powell essentially said they have no idea what impact the current trade war will have on the economy. He said it’s likely to slow growth and boost inflation, at least in the near-term, and that the Fed needs more time before doing any kind of policy easing. Considering potential rate cuts would be a response to policy that can change rapidly, I think that’s prudent. At the same time, I do think a little QE would help calm some of the issues we’ve seen in bond markets and credit spread widening. 

Smartphones, semiconductors and other items are exempt from the 145%-245% China tariff. Instead, they’ll pay a 20% tariff for now, although the administration did say potential semiconductor tariffs are coming soon.

We continued to hear about positive negotiations and talks (including with China) this week. I continue to think this will all be very loud, ugly, and chaotic – before eventually being resolved. I’m still cautiously optimistic that it will happen before the 90-day tariff pause expires, although that could easily change with future headlines. We shall see.

Output data:

Inflation data:

Consumer & Employment Data: