This piece contains a brief Hims earnings snapshot, a Mercado Libre earnings review & a Palantir earnings review. The Hims review (with far more detail) will come tomorrow with a Lemonade review. At first glance, the Hims quarter looks like a weaker performance than they’ve delivered in a while. Not terrible and also not as good as we’ve grown to expect from them. As most of you know, I’m not a bull here, but I need to read the transcript and letter before commenting further. Stay tuned.

In case you missed it:

Hims trades for 58x forward EPS. EPS is expected to grow by 32% this year and by 43% next year. Estimates should be stable following the annual guidance reiteration.

MELI is the e-commerce and logistics king of most of Latin America. It has a thriving marketplace and fulfillment business, with support for 3rd party merchants. It also features a rapidly growing financial services suite and payments platform, entertainment offerings through partnerships, a rapidly growing ads business and a loyalty program called MELI+ where it laces a plethora of product utility into one unique consumer bundle. The business model resembles Amazon without cloud computing and with financial services. Here are the names of its various products:

There’s a lot to unpack for margins. As Mercado Libre continues to build out its credit book, costs from funding that business, as well as front-loaded credit loss provisions, significantly weigh on margins. Last year, new credit was a modestly larger portion of the overall book than it is this year. Meaning? The immediate cost penalty associated with this segment’s growth eased Y/Y and the net impact was less severe for overall margins. Still an impact via continued growth in provisions and credit costs, but an easing impact in the Y/Y comp.

When contemplating results vs. expectations, credit was still a large headwind. Meli’s credit portfolio rose by $1B more than analysts expected it to. They do not offer any forward guidance, and so modeling expectations is quite hard. This means that while the credit business may have been neutral or very slightly helpful in Y/Y margin comps, analysts thought the impact would be even more positive to its overall cash flow & income statement margin profile.

So… to summarize… its credit portfolio was positive for its overall margins in the Y/Y comp, but still drove the profit misses vs. expectations. Later in the piece, we will walk through how the continued margin drag is the correct present decision and should pave the way for incrementally profitable MELI growth in the years to come.

In terms of the margin contraction, there are a few things contributing. First, net income reflected higher foreign exchange losses that fostered about 25% of the Y/Y margin contraction. Next, it greatly leaned into marketing spending, with 50% Y/Y growth for that line item. It wanted to support the lowered shipping fees and more Mercado Pago growth with well-funded marketing campaigns. This led to another point of Y/Y margin contraction and fostered “very positive user growth.” It also enjoyed immediate spikes in assets under management, improved buyer frequency and a “drastic improvement” in Q/Q app downloads. Sounds like a great decision to make in exchange for some modest near-term margin pressure. Most of the remaining Y/Y margin pressure sprang from MELI lowering some logistics fees in Brazil in June (much more later). There was a month of impact from this change for Q2, and offset some fixed cost leverage.

“We are investing to capture the many growth opportunities ahead, while maintaining our commitment to doing so profitably and sustainably. In Q2'25, we believe we struck a good balance.”

Shareholder Letter

Definition: Net Interest Margin After losses (NIMAL) measures credit revenue - credit provisions - funding costs. It’s similar to NU’s risk-adjusted net interest margin (NIM). Higher is better. 15-90 day non-performing loan (NPL) rate measures the proportion of loans 15-90 days past due. Lower is better. 23% NIMAL was a full point better than expected.

15-90 day NPL rate fell below 7% for the first time ever. This is the leading indicator for future credit repayment health and looks excellent. This is what is giving MELI the confidence to keep accelerating credit card issuance. Not hope. Data. As a reminder, they pulled back on growth two quarters ago because some competition was seeing early signs of deteriorating credit book performance. That did not manifest for MELI, as its preemptive caution surely helped. More structurally speaking, its extensive, shopping-history-rich customer data profiles that give it a unique understanding of repayment ability helped as well.

Now back to the credit business and how it’s showing clear signs of being a future profit driver. In Brazil, cohorts are uniformly reaching positive NIMAL on schedule and 51% of its portfolio there is now profitable and Brazilian NIMAL overall is breakeven. This should rise, while the percentage of the book from new customers keeps falling and eventually eliminates the profit headwind. That will create a compelling tailwind with a massive addressable market. Mexico should be a few quarters behind.

One more note on NIMAL. During the period of hyperinflation in Argentina, MELI was able to hike the rates it received from borrowers faster than its lenders could hike the rates on its own debt. This led to a large (unhealthy) NIMAL tailwind last year, with disinflation since then leading to the Y/Y decline. This is healthy. Furthermore, the credit card keeps growing as a % of the overall credit portfolio. This has a lower NIMAL than personal loans and represents another headwind.

MELI trades for 45x forward EPS and 16x forward FCF entering this report. EPS is currently expected to grow by 32% this year and 38% next year. FCF is expected to fall by 24% this year and rise by 46% next year. I expect profit estimates to fall after this report. At the current stock price, I think assuming a forward EPS multiple closer to 50x is fair.

Commerce:

MELI debuted a new search interface on its marketplace. Shockingly, it infused new partner AI models and a lot of its 1st-party data into this upgrade. The impact so far has been great. It directly supported Brazilian items-sold growth acceleration in the month of June, as that metric jumped from 26% for the full quarter to 34% in June specifically. GMV growth also accelerated accordingly. Overall, UAB growth was as fast as it has been in over 4 years, thanks to broad-based strength, and items sold per unique buyer match a record high for the company. 

Execution in Mexico also improved significantly Q/Q. If you’ll recall, during Q1, technology sales held back overall results. Competition cut prices and added compelling finance terms, while MELI thinks it was too slow to react and did not have optimal assortment. All of that has since been addressed, and so it’s no surprise to see FXN Y/Y GMV growth there accelerate 9 points vs. last quarter. The company is also working harder on competing more effectively in Mexican cross-border, and notched a material acceleration in that business as a result.

1st-party commerce also enjoyed great 103% Y/Y FXN growth. MELI will also be a predominantly 3rd-party operation and will fixate on supporting other merchants. Still, there are gaps in the product assortment that it knows it can fill (non-perishable groceries, smartphones etc.) and it will continue to do just that.

Logistics:

We really could combine this with the commerce section. Logistics revenue is part of that bucket and the network is a key enabler of MELI’s fantastic marketplace success. It’s how it matches great assortment with great service and fulfillment times. Just like for Amazon, that directly bolsters conversion rates, net promoter scores, retention order frequency and everything else that investors should care about. And it’s how leadership thinks it has built a defensible moat and turned this business into a juggernaut with a massive runway. Logistics and commerce - salt and pepper.

While we all want companies to deliver maximum profit every single quarter, that’s not always consistent with the right decision for maximizing long-term profit ceilings. Sometimes, investments in a better customer experience and motivating more delight can have a more powerful impact on the long-term bottom line than forgoing those near-term expenses. MELI has been phenomenal at this since its inception, and it plans to keep embracing this philosophy where it makes sense. As they say every single quarter, “the best is yet to come.” And to ensure they take a commanding piece of a competitive market, spending is necessary and temporary margin pressure is occasionally inevitable. That is what played out this quarter. We’ve already discussed the margin impacts of more marketing spend and the continued credit growth. Now, for the two briefly mentioned shipping fee reductions.

First, it again lowered its free shipping threshold in Brazil from 79 Brazilian Reals to 19. This is the third time it has made this move in five years, and is the result of continued investments in its logistics network to enable the kind of localization and economies of scale that make this rationally possible… while making MELI much harder to compete with. In the past, these decisions have directly fostered faster e-commerce penetration gains, rising order frequency and higher customer lifetime value. Leadership is convinced that the same will be true following extensive testing.

“We would not be a $50+ billion GMV company today if it were not because of building our logistics infrastructure and launching our free shipping program back in 2017. And we think this is the same case now.”

Incoming CEO Ariel Szarfsztejn

Second, it cut merchant shipping fees for goods priced between 79 and 200 Brazilian Reals. There was a take rate “cliff” at the 79 Brazilian Real threshold that it thought deterred some assortment and some GMV growth. It has smoothed out that “cliff” and the results have been positive. Merchants are lowering their product prices across the board, and assortment growth is picking up. Importantly, these two changes will lead to some negative contribution margin orders, but vitally, the positive observed impact for everything else makes this positive for overall profit; that impact is expected to amplify in the coming quarters and years; Again… near-term profit concession… to keep distancing itself from the competition and raise the long-term profit ceiling. 

Advertising:

Its Google Ad Manager and AdMob integration was completed at the beginning of the quarter. This gives ad buyers far more access to non-MELI impressions while maintaining access to lucrative MELI shipping data. It also fosters the ability to manage campaigns that seamlessly mix-and-match on-platform and off-platform impressions (meaning not on a MELI property). It thinks this off-platform initiative puts MELI directly in the path of becoming the largest digital advertiser in LatAm – and that’s the overall goal. Overall, advertising revenue rose by 59% FXN Y/Y (38% on a nominal basis), with nearly 100% Y/Y display and video growth. That newer bucket is picking up steam. In Argentina, slowing inflation is unlocking more budget and more margin for ad spend. They continue to improve ad-buying experiences, impression access, targeting, reporting and other tools… and advertisers continue to respond quite well.

Fintech:

The fintech flywheel effect is working exactly as planned. As the buyer base grows, items sold per user in the marketplace grows right with it. MELI hoped that getting more people to embrace its fintech suite (mostly credit cards, loans and savings accounts) would drive intuitive cross-selling opportunities, supporting its core business segment. That is happening.

The Mercado Pago marketing campaign fixated on communicating MELI’s superior savings account yield, which it thinks is driving engagement just like free shipping does on the commerce side. It’s amazing what happens when you take good care of your customer, and MELI is fortunate that most of the legacy competition just simply doesn’t. Continued product adoption and more marketing led to the largest Q/Q rise in assets under management in Brazil and Mexico during the quarter.

Acquiring TPV:

Acquiring TPV refers to the piece of MercadoPago that lets merchants seamlessly tap into these easier payment options, augmenting conversion rates (through software and point of sale hardware). Acquiring TPV happens whenever MercadoPago is used as the main payment facilitator/processor on the marketplace or on a merchant’s own site (“off-platform” acquiring TPV). 

The company continued to gain share in this category for yet another consecutive quarter. It continues to successfully move upmarket in Argentina and also significantly cut Chilean payment processing time for its point-of-sale systems (thanks to moving the processing onto Mercado Pago). That performance improvement immediately led to an acceleration in growth. In Brazil, FXN growth was faster than it has been in over two years and it added Mexican bill pay integrations to expand use cases. Finally, it added new automated currency conversion to showcase U.S. dollar prices on a merchant’s site. It plans to charge for this in the future, representing a compelling margin and take rate opportunity.

This was not a perfect quarter. Despite the fact that MELI doesn’t guide to profitability, they still missed expectations and we can’t call that positive.

On the other hand, the reasons for the miss are about as encouraging as they can be. The company is pursuing highly compelling growth opportunities and will not sacrifice those for quarterly results. Good for them. The leadership team continues to be world-class allocators of capital, which makes us highly confident in this decision and leaves us thinking “please, go for it.” The profits from more revenue growth in the years to come will follow. But for now? It’s not time to optimize profitability. It’s time to greatly fortify the competitive moat and raise the revenue ceiling. 

If Mr. Market decides to overly push this decision, I will say thank you and buy more shares. I’m eyeing roughly the $2,000/share level to do that, which represents a roughly 42x earnings multiple and a 1.4X PEG ratio (following what I expect for negative earnings revisions).

Palantir is a software company that helps customers get the most out of their structured and unstructured data. Like many others, it pulls from years of AI/ML work to automate insight-gleaning. It utilizes complex neural networks to power anomaly detection, trend forecasting and natural language processing. All with the aim of making enterprises more successful in cultivating customer relationships and driving financial success.  It works openly with many database vendors for scalable, interoperable storage and low-latency querying, and offers its own tools there too.

Overall, it frees clients to conjoin disparate data sources while utilizing its software to uncover ideas that manual analytics and legacy competition cannot derive. It gives customers a bird’s-eye view of their operations, with detailed suggestions to help optimize products and workflows. This happens in a zero-stakes-or-risk environment via a process called ontology. Ontology enables clients to freely test massive digital twins to actually observe what works and what doesn’t. It’s like split-testing on steroids.

It rips on other software firms for (it says) building “slide decks” rather than matching this real-world utility. It pounds its chest for being superior here, with countless tangible case studies and its last two years of results providing compelling evidence that they’re right.

Revenue is neatly split into two buckets – “government” and “commercial.” Government clients predominantly use its Gotham product platform, while commercial clients mainly use its Foundry product platform. With Gotham, Palantir routinely builds custom use cases for individual government clients. Foundry was built to be more malleable, with far more pre-built app integrations and developer kits available. That diminishes the need to conduct custom builds for every single enterprise. Still, it does materially more custom building than a typical B2B software firm will. That’s more expensive, but has also led to wonderfully sticky client relationships with ample opportunity for up-selling. Like a client purchasing more modules from a vendor raises retention thanks to more unique value enjoyed, this accomplishes that in a related way.

It also seamlessly leverages the commercial platform to cater to industry-specific needs. By-industry models are intuitively named “micro-models.” These boast sector-specific use cases with granular, relevant regulatory compliance help. A financial services model from Palantir, for example, may specialize in assessing credit risk or fraud detection.

Palantir Apollo provides continuous integration and continuous delivery (CI/CD) to automate software package building and deployment. It’s a foundational piece of the firm’s ability to collect, utilize and drive value from broad data ingestion. It’s also how Palantir can help operationalize these learnings to introduce valuable products. Apollo ties very closely into Foundry and Gotham as a software enabler for both platforms.

AIP 101:

In the realm of GenAI and Agentic AI, Palantir is not playing the game of building the biggest model or buying the most GPUs to have the largest infrastructure footprint. It simply gives clients the tools and integrations needed to build apps for their own needs. It also has some products for model customization, along with a large roster of 3rd-party large language model (LLM) integrations. 

And while it's no slouch on the model menu for its client base, where it shines brightest in GenAI is as a leader in application monetization.

Its most exciting product is called Artificial Intelligence Platform (AIP). This is a highly intricate automation and AI app-building layer that complements Foundry and Gotham perfectly. The company compares AIP to what public cloud vendors did for compute and workload modernization. AWS, Azure, Google and Oracle provided the environment, tools, storage, security and maintenance needed to grow compute capacity without managing it yourself. This made migrations and adoption the rational decision. AIP attempts to do the same thing in terms of pushing enterprises to build and use GenAI applications. This fully manages expedited product creation for client deployment. It allows for open collaboration between software developers, data scientists, models and project managers to ensure effective work. It directly supports Foundry and Gotham by uplifting and augmenting potential use cases to “extract value from GenAI models.” 

Considering the lack of finite and structured end products stemming from AIP, I think it helps to hear about some examples of what clients are doing with it:

As leadership will tell you, AIP isn’t just another dime-per-dozen chatbot. It’s an aggregator of data, tools and services needed to actually build valuable apps and to embrace GenAI. It’s how Lowe’s cut overdue task rate by 75% and how General Mills saves $14 million a year in expenses. The list goes on and on. AIP is where jumbled data, processes and ideas turn into the operationalized, actionable creation of GenAI products.

Initial go-to-market for AIP has been its “bootcamps” where it hosts events to provide hands-on support and “get clients from 0 to use case in 5 days. It has more recently begun to build out an external sales team (slow ramp) to support this segment’s momentum. AIP progress is most noticeable in its impressive U.S. Commercial results.

“We are delivering the operating system for the modern enterprise in the era of AI.” – Co-Founder/CEO Alex Karp

More Products to Know:

“The growth rate of our business has accelerated radically, after years of investment on our part and derision by some. The skeptics are admittedly fewer now, having been defanged and bent into a kind of submission. Yet we see no reason to pause, to relent, here.”

CEO Alex Karp

“I've been cautioned to be a little modest about our bombastic numbers, but there's no authentic way to have anything but enormous pride and gratefulness about these extraordinary numbers. We're sorry that our haters are disappointed, but there are many more quarters to be disappointed, and we're working on that too.”

CEO Alex Karp

Forward-Looking Demand Looks Awesome:

PLTR trades for 250x forward EPS. EPS is expected to grow by 42% this year and by 27% next year. Estimates will sharply rise following this report.

Just like last quarter, this was a very short call with not much newness. And when you’re posting numbers like they’re posting, that’s entirely fine. They’ve earned the right to do what they want and to, within reason, say as much as they want to say. The review below is lighter than most that we write because of this.

AIP:

AIP remains the crown jewel of this company and the software layer of global AI monetization. This product and this company are one of one in that regard – and I do not think that’s a controversial take. Why are they winning? For the same reason they’ve been winning since the launch. While most others struggle to extract tangible value and return on investment from AI models, PLTR is helping clients rapidly build software to do just that. They unlock vast digital twin split-testing through ontology that closely ties models and software together and allows companies to freely test ideas without operational risk or disruption. They give customers the keys and tools to know what will actually work, before they build and deploy it. And in turn, they rapidly accelerate the learning curve and time-to-market on impactful product updates. Simply put, they’re winning because their customers have been so wildly successful with this product. 

“LLMs, on their own, are at best a jagged intelligence, divorced from even basic understanding… LLMs simply don't work in the real world without Palantir.”

Chief Revenue Officer Ryan Taylor

And while many other companies may claim to emulate that utility, PLTR backs it up… quarter after quarter… with a long list of compelling case studies. Fannie Mae is using AIP to cut mortgage fraud detection time from months to seconds and save the U.S. government millions. Lear Corporation expanded their usage of AIP in a new 5-year deal, following Palantir’s unmatched ability to automate tariff exposure mitigation and tedious administrative work. Citibank is using AIP to cut their know-your-customer (KYC) security checks from nine days to seconds. Nebraska Medicine raised discharge lounge utilization by 2100% to effectively “add another unit to their hospital” in terms of patient throughput. Extra unit… with very little cost vs. building one. They now use the term “Palantir unit of time” thanks to the hours upon hours saved with AIP across several workflows. The U.S. Department of State is cutting foreign service candidate approval from 60 days to 12, while Land O-Frost finishes production schedules in 30 minutes instead of previously 40 hours. An American telecom company 10Xed their PLTR commitment and expects to save hundreds of millions in expenses. This wasn’t even the full list of examples from solely Q2. Other software companies would struggle to come up with a list this long since the AI boom began.

“Software companies are re-platforming away from the highly un-opinionated services and building blocks of the hyper-scaler stack onto AIP with its highly opinionated building blocks that get you to value 10 times faster.”

CTO Shyam Sankar

More on AIP Products:

AI Forward Deployed Engineer (AI FDE) is a newer piece of the AIP product suite. Rather than leading with a massive direct sales team and “steak dinners,” this provides customers with an AI-based equivalent of a highly-skilled engineer to manage and guide more of the app creation process. This is “accelerating the already mouth-watering time to value for customers” by managing ontology, data migration, code debugging and other tasks. In turn, this greatly reduces resources needed to deploy AIP.

AI Workbench is another AIP tool that provides a safe environment for testing and iterating on AIP work. It’s a developer playground for building and testing agents and setting up data workflows, with slick tools for checking work and debugging issues. And finally, it introduced ontology-as-Code, which integrates with a developer’s familiar tools and allows them to build ontologies with more guidance and guardrails. All of these products reduce friction associated with using AIP and Palantir’s software by adding support for 3rd-party services, expediting and managing product creation and offering a secure environment (with compelling automation) to do all of this work.

Government Business:

Palantir enjoyed a nearly $800M boost to its Maven smart system contract ahead of strong expected demand. As a reminder, this is PLTR’s Department of Defense AI and data analytics platform. And just last week, as I’m sure many of you saw, it signed a $10B Army deal to displace 75 contracts. We only talk about point solution displacement and vendor consolidation being a great tool for retention and cross-selling. I can’t remember another example of 75 of those point solutions being cut in one fell swoop.

More on International:

International commercial business fell 3% Y/Y. It’s happy to work with U.S. allies, but is far more focused on growing the U.S. business at this point in time. Europe has moved very slowly in embracing AI, and there are just more opportunities in the states than elsewhere. Still, international government revenue did rise 37% Y/Y – largely thanks to U.K. growth.

Sales Team:

Palantir has been putting up these elite numbers with a tiny sales team. It does not feel the need to lead with sellers who don’t truly understand the product suite. It wants its actual engineers to be talking to customers at boot camps (for demonstrating and helping with product creation). They know the product. They know how to deliver value with it. It also thinks organic word-of-mouth will keep becoming a more and more powerful growth vector for the company. That’s the luxury of delighting customers. They are readily and loudly talking Palantir up to all of their partners and friends. That should not only foster more growth… but more growth at sky-high margin & unmatched efficiency.

Just like the last several quarters have been, this was a historically strong report. They are putting up growth and margin at their scale that nobody else is coming close to matching. I think that all goes back to the central idea of providing real, demonstrated value while others have a hard time matching it. This is one of the highest-quality companies on the planet and the team deserves yet another round of applause.

From a valuation point-of-view, even with sharp upward profit revisions, this will still be more expensive than anything else we talk about. And the gap is quite large. If there was one company I’d be willing to pay 80x+ forward revenue and 250x+ forward EPS for, it would be this one. But? I am not willing to pay that multiple for anything. I will remain on the sidelines here and continue admiring their execution from there. I’d love to own shares at a reasonable price tag, but this isn’t even in the ballpark of reasonable. Iconically impressive company but too expensive for me. The risk/reward is not there, in my opinion. But for bulls… please have fun counting your mountains of profit. You’ve earned it and this valuation opinion from me does not change that in the slightest. Congrats to you.