Table of Contents:

Earnings Reviews from this Season:

& my current portfolio/performance.

GitLab trades for 52x EPS. EPS is expected to compound at a 15% clip over the next two years following a couple years of triple-digit growth.

Lululemon lowered annual revenue guidance by 2.7%, with the new target missing estimates by 2.5%. New revenue growth guidance represents 3% Y/Y growth excluding the extra week. They now think U.S. revenue will fall by 1.5% Y/Y in 2025 vs. prior expectations calling for modest growth. It also now sees 22.5% Y/Y China growth vs. 27.5% previously. This is due to some macro weakness in their large cities. It continues to forecast 20% Y/Y Rest of World revenue growth. A lot more on this revenue disappointment later. It also lowered annual EPS guidance by $1.81 (or 13.4%) to $12.87, which missed estimates by $1.74. This new guide represents a 12% Y/Y EPS decline for 2025. A lot more on this right now. Guidance now includes a larger tariff impact of $240M in gross profit or 220 bps for GPM vs. 40 bps previously. All in all, they see GPM falling by 300 bps Y/Y vs. 110 bps previously.

Virtually that entire change (170/180 of the bps) is related to most of its shipments from Canada to the USA being eligible for de minimis tariff exemptions. Other weakness is due to a 50 bps markdown headwind vs. 15 bps previously guided to. Not great. Beyond input costs, guidance includes 35 additional bps of SG&A deleveraging, due to lower sales growth assumptions, investments and FX headwinds. All in all, it now expects operating margin to fall by nearly 4 points Y/Y.

Lulu trades for 15x forward EPS (with vulnerable forward estimates). EPS is expected to fall by 9% this year and grow by 5% next year.

More Weakness in the USA:

The profit revisions are almost entirely tariff-related and not something that I harshly blame on leadership. While you could argue a better team may have been able to claw back some of these expenses more quickly, they’re doing a reasonable job in doing so. The issue is the demand guidance and bad execution in the USA. Throughout 2024, we heard about a lack of newness in their inventory assortment. They needed to introduce fresh products with seasonal colors and patterns that guests would surely respond positively to. Newness was expected to be back to historical levels by this past spring. That’s what we were told. So what happened this quarter? A few things. 

First, a lot of the new inventory didn’t move. Some of the new styles they introduced sold well, but a lot of the seasonal colors and patterns they thought were key to a demand recovery didn’t. They will need to clear out excess inventory throughout the end of the year, which is why markdown rate guidance worsened. I can’t help but think of other leadership teams in different parts of consumer discretionary like Cava. They don’t introduce a new menu item until it has been slowly, carefully tested and it’s painfully obvious that item will work. Is Lululemon not using a data-driven and evidence-based approach to their product line? This makes me think so.

Second, management believes that they’ve let some of their core franchises stick around for too long. Just like last year, they’re saying that the “newness” isn’t enough. They need many additional product launches to replace these fading product lines. Candidly, this sounds like Nike’s journey a few quarters ago when it sunsetted legacy franchises and ate a large revenue headwind as a result. Not great for a company considered a perennial double-digit grower and fundamental darling just a couple of years ago.

Thirdly, they didn’t have enough of the stuff that did work. We’ve been hearing things like that for over a year and it’s candidly getting very tiresome. The issues that forced me out of this great brand and investments a few quarters ago have not gotten any better. They’re dragging on.

Finally, they’re blaming fragile macro, a “challenging premium athletic wear market in the U.S.” and a “different competitive landscape.” The team was quick to say that no single company is materially impacting them, but it’s clear that competition is having some degree of impact. Still, most of this is just Lulu not executing.

The rare positives from this report include growth and market share gains in performance apparel, rising total guests (for all age groups), and higher guest retention rates. Conversion rates also didn’t worsen (but average order value did). And while these things are nice, they’re clearly not nearly enough to offset all of the softness. That’s mainly related to disappointing spending levels on the loungewear and social apparel sides of the business. That’s where fresh colors and patterns in the dated core franchises just aren’t working.

Lululemon is not early enough in its U.S. growth journey where it can simply rely on rapid store expansion to overcome some macro weakness and a lot more micro-level weakness. They’ve needed to do better for over a year. And unfortunately, momentum is not on their side. Traffic trends in this highly important market got progressively worse as the quarter dragged on.

USA Fixes?

Beyond yet another quarter of talking about fixing assortment and pushing back the timeline for resolving this issue, it has a few other fixes planned. They plan to accelerate the pace of new product testing and invest in more rapid inventory response to actual demand levels. Why this wasn’t already a priority in an age where data is king and not using it is flying without a net, I don’t know. Maybe the new Chief AI and Technology Officer can help them in this regard. McDonald is also excited about the work of Jonathan Cheung (Global Creative Director). He feels Cheung’s team is now in pace and that their product innovation will begin to bear material financial fruit next year. At that time, they expect product newness levels to rise from 23% of total to 35% of total. Again… I feel like I’ve been hearing the same things for over a year. Now they’ll surely fix it in 2026 instead of 2025. I will believe it when I see it.

International:

International results were actually fine. They added new stores in Italy and franchise-owned stores in Turkey and Belgium. Lulu also announced an exciting franchise store partnership to expand into India next year. The team sees a very long runway for unaided brand awareness and market share gains across the globe. It also thinks some of the fixes in the USA can be relevant for international markets.

This quarter does nothing to make me want to re-enter the position I exited earlier in the year around $250/share. It makes me less interested – even at this admittedly dirt-cheap valuation. I almost expect guidance to be cut again this year and for estimates to keep falling. The forward multiples we are all looking at are likely not accurate and will end up being higher in reality.

This team no longer has a good feel for their business. They don’t seem to know what inventory will work and what won’t. They can’t move quickly enough when they actually determine what is selling. They haven’t inspired much confidence that yet another round of operational changes will be fruitful in  2026. We’re left with a strong brand that has completely lost its way and is showing zero signs of turning things around. Sure, there are some international bright spots… But if the USA is going to look this bad, that doesn’t really matter right now.

I do not think this is Alo or Vuori preventing Lulu’s success. Both are small fish in a giant industry and can grow much, much larger before having a major impact. Lululemon needs to look no further than a mirror to understand what is going wrong at that company. Apologies if this note is harsh… but so was the quarter. It’s time for a new leadership team.

Salesforce is one of the largest enterprise software firms on the planet. It provides a broad suite of products to help clients optimize customer interactions. The overarching niche is called Customer Resource Management (CRM). Salesforce offers a variety of cloud services for its customers. There’s a sales cloud, which perfects consumer touch-points. There’s a commerce and marketing cloud to build online storefronts and augment promotional activity. There’s a service cloud to handle customer issues and inquiries. There’s also a platform cloud, which includes Slack.

It also features a data cloud. This is an aggregated analytics service to ingest, organize & glean insight from 1st-party data. It conjoins siloed context and “unlocks” previously disparate sources for client value creation. It’s similar to Snowflake but more purpose-built for managing customer relationships. MuleSoft and Tableau are both key pieces of this data cloud. MuleSoft integrates apps and data to enable management of these products within Salesforce. Tableau is a data visualization tool to create automated progress reports and suggestions to leverage findings. Finally, it offers industry-specific clouds for sectors like healthcare. These are customized to meet specific regulatory and operational needs. All of these clouds and products make up the firm’s subscription & support revenue, which represents 93% of its total business. Professional services make up the rest.

Separately, Salesforce offers a product called Einstein One. This is a full set of AI tools, including outcome prediction, chatbots, image recognition, sentiment analytics and more. It’s considered a general-purpose AI platform infused into all Salesforce products. Most recently, through an OpenAI partnership, it debuted Einstein GPT. Einstein existed before the GenAI wave, but is now getting an upgrade thanks to it. Einstein GPT allows Salesforce clients to plug into language models (including OpenAI, Anthropic and Cohere) to make workflows more productive, intuitive, conversational and automated. It features a low code tool set to reduce the barrier for non-experts to build applications; it also boasts expert-level tools to build more complex apps. 

The newest Salesforce product is called Agentforce. This is (shockingly) another AI-powered platform with a diverse set of agents automating tasks across all of its other clouds. It’s worth noting that the Data Cloud was a vital prerequisite for this launch, as it ensures agents are trained with unified and cohesive context. Agentforce is a full-service, out-of-the-box suite freeing companies to turbocharge GenAI adoption by making the embrace of this technology easier and safer. It brings together all of Salesforce’s work in Einstein AI and Customer360 (data unification across disparate sources), creating an end-to-end GenAI platform that actually drives value. Founder/CEO Marc Benioff sees Agentforce as an immediate ROI driver across a boatload of use cases. These agents do not replace humans, they simply assist them within things like drug discovery, deal pipeline building, customer service etc. And? Knowing Salesforce cannot possibly create every agent a customer could ever want, it offers model and agent builders for customers to seamlessly create those specific agents themselves.

The Salesforce platform, which enables native app development, was included in all of its 10 largest deals, while 70 of its 100 largest included 5+ clouds. These trends are great for lifetime value and retention.

Its acquisitions of Convergence AI (makes AI agents) and Bluebirds (for sales cloud automation) both closed. It is working on closing acquisitions of Regrello (AI workflow automation) and Informatica (data management cloud). Informatica could close a bit earlier than expected, but it’s not included in the guide.

Salesforce trades for 21x forward EPS. EPS is expected to compound at an 11% clip for the next two years.

Agentforce:

In the world of Agentic AI, most companies building tools for other enterprises remain in “product-fit” mode as they figure out how to create real value. Considering this, I think it’s helpful to walk through how Agentforce is actually impacting its product suite and customer experience. Notably, Salesforce calls itself “customer zero” for all of this innovation. It has a massive organization where this agentic technology can be readily inserted, driving cost savings and better outcomes. It can use its own business to offer proof of concept for customers exploring new products. This allows it to pitch them with “see, it works,” rather than “take my word for it.” And it also makes Salesforce more efficient and profitable. 

For example, a customer service agent can already cover 94% of its global case volume and boasts a 77% case resolution rate. Furthermore, they’re now finally responding to all sales prospects, which was simply impossible to manually do. There was too much scale, and AI is perfectly suited to help with that, while sending high-quality leads (with ample context) to salespeople. Across its sales, service and other clouds, its agentic technology is automating tedious, mundane work and upleveling the productivity of labor forces:

Data Cloud:

As with other enterprise software firms, CRM’s data cloud is a vital complement to every other cloud it offers. It’s how these clouds and agents are trained with the context and understanding needed to actually be valuable. It’s where AI learns. It’s why CRM thinks their agents are “probably the most accurate in the industry.” These agents aren’t fully replacing the need for support staff, but they are helping a ton. CRM cutting 40% of that team and greatly shifting resources to more sales talent to accelerate growth. Better accuracy within sales agents means hallucination rates are lower than competing products. That means these agents can handle a larger volume of overall inquiries, which is how Salesforce and its customers have been able to conduct this team reshuffling more aggressively than others.

Data Cloud and AI/Agentforce annual recurring revenue (ARR) rose 120% Y/Y to $1.2B. The convincing majority of these $1.2B comes from data cloud, as last quarter they disclosed $1B in data cloud ARR and $100M in Agentforce. Data Cloud specifically enjoyed 140% Y/Y customer growth as well a 326% Y/Y growth in row access traffic. In Q2, Salesforce added 60 $1M+ contracts that had both Data Cloud and Agentforce in it, while 40% of overall bookings for those two products came from existing clients. This is turning into arguably its strongest cross-selling and top-of-funnel customer growth tool. 

“If a business customer isn't actively shipping, our own marketing cloud campaign is automatically triggered and sales reps are alerted and it's all happening through our Data Cloud. – Co-Founder/CEO Marc Benioff

Public Sector:

Benioff was excited to announce that they’ve added the U.S. Army to their customer list. This joins large contracts with Veterans Affairs and the U.S. Coast Guard. The contract includes a “digital front door” for non-stop “service and support to all soldiers and millions of veterans.” They expect to do a lot more with the U.S. Army in the years to come. And with Agentforce receiving FedRAMP high status, the door is open for this new product to be part of these contracts.

IT Service Management (ITSM):

Salesforce sees ITSM as a compelling source of future growth. They will soon launch an “agentic IT service platform” and expect this to perfectly blend into its existing Slack user interface to eliminate disruption and enjoy “zero learning curve.”

Competitive Landscape:

Benioff ripped into other enterprise software companies promising artificial general intelligence (AGI) and an all-encompassing software layer that displaces the need for all other tools. He finds that unrealistic and egregious. There has been a growing narrative that AI-native tools are displacing the need for software as a service (SaaS). Benioff argued (and I fully agree) that AI isn’t replacing these needs… it’s simply making the products better at addressing them. He poked at “AGI coming tomorrow” and talked about feeling well ahead of the competition in their product positioning – including in AI.

“When I look at the other large enterprise software companies and I look at their websites and I look at the capabilities they're providing… I'd say we're way ahead.” – Co-Founder/CEO Marc Benioff

This was an average quarter. I will say that positive bookings and pipeline commentary does point to accelerating growth if they can execute. I also think the Informatica purchase is a good one and that they should keep using M&A as a source of growth in the years to come. They’re sitting on a mountain of cash, buybacks are ramping up, cash flows are robust and if they need to keep buying demand to continue more profitable revenue growth, they should. At 20x forward EPS for a blue-chip like this, I do find this somewhat interesting. Still, I gravitate towards watch list names like ServiceNow and its 20% compounded top-line growth (at a higher multiple) over this one. I don’t love or hate the investment case.

Broadcom creates & manufactures a slew of semiconductor-related equipment within data center, networking and industry-specific use cases. Chips and high-performance compute (HPC) can’t all be packed into the same corner of a data center. GPUs must be able to connect to one another to drive better bandwidth and performance, with faster, more efficient model training and inference to cut costs. This is where Broadcom thrives. 

It also offers a range of software tools, which significantly broadened out with its VMWare acquisition. VMware offers virtual, localized layers of software that sit on top of hardware. This allows the centralized hardware to run several different operating systems from the same place. The company, which is now a Broadcom unit, calls these “virtual machines” or virtual private clouds. By reducing hardware requirements, VMWare saves its clients money. 

This company does not compete with Nvidia in terms of designing GPUs. It does, however, create application-specific integrated circuits (ASICs) for more specialized workloads. It also makes variable processing units (XPUs), which are the high-performance accelerator subsection of ASICs. All XPUs are ASICs but not all ASICs are XPUs. These are often used to optimize data center, networking and GPU performance. In some cases, this can replace various needs for more generalized chips like GPUs. Furthermore, its core niche focuses on networking and connectivity, which competes with Nvidia’s switches and its SpectrumX networking product.

AVGO trades for 40x forward EPS. EPS is expected to grow by 39% this year and by 34% the following year.

Semiconductor Solutions – AI & XPUs:

XPU outperformance drove a material portion of the successful quarter. Y/Y growth for this bucket accelerated from 46% Y/Y to 63% Y/Y to reach $5.2B. Its three core customers (Alphabet, Meta & Bytedance) did contribute to above-consensus results, but a 4th large customer had an even bigger impact. This 4th customer (Anthropic) signed a $10B XPU order with Broadcom this quarter. The revenue contribution has been immediate and sizable and will continue well into next year. And as a result, the company now expects AI revenue overall to exceed the 60% Y/Y growth target it most recently set for 2025. In 2026, they expect AI revenue growth to be meaningfully faster than 60% Y/Y. No slowdown in sight.

The company was asked about the possibility of adding more customers to the overall list (4 large customers in production; 3 in experimentation mode). It’s open to that, but will remain extremely picky about who it will work with. It only wants to sign contracts that entail vast sums of volume.

Finally, the GPU vs. XPU debate rages on. GPUs are excellent for training/inference and general use cases within high-performance computing. XPUs are more so specialists and masters of very specific kinds of workflows. They do very few things better than GPUs can. Leadership is confident that over time, XPU share will surpass GPU share and AVGO will benefit.

Semiconductor Solutions – AI Networking:

Broadcom is leveling-up scaling-out (using networking equipment to connect more server racks) capabilities with its performance-optimizing Tomahawk Six switch and its Jericho Four switch (both Ethernet-based). The Jericho offering is expected to handle connection capacity of up to 200,000 compute nodes across data centers. That cross-data center networking connectivity for the Jericho switch is essentially scaling-out on steroids and is called scaling across. They are adamant that they’re delivering Ethernet equipment that can handle more data processing and more chip connections than Nvidia’s competing offerings.

Semiconductor Solutions – Non-AI:

The demand recovery was called “slow for another quarter. Revenue growth was 0% Q/Q, as enterprise networking and service storage weakness offset promising broadband strength. As you can see from the guidance commentary, that trend is expected to continue and should create Q/Q growth next period. They’re seeing this play out in 20%+ forward bookings growth and it’s helping them gain more comfort in this segment bottoming for the cycle. Hock Tan cautioned that they’ve been “tricked before” with promising forward-looking demand signals that did not manifest, but they’re confident nonetheless.

Infrastructure Solutions:

VMWare’s 17% Y/Y growth was another source of company outperformance this quarter. The company booked $8.4B in total contract value for VMWare this quarter alone, and now feels like the multi-year integration project is finally complete. This should enable more interoperability and seamless cross-selling for the VMWare Cloud Foundation (VCF) offering. As a reminder, VCF “enables the entire data center to be virtualized and customers to create their own private cloud environment on-premise.” This frees them to run their apps in reliable, scalable places other than in a public cloud.  And that's a timely addition to AVGO's value prop. With data leakage a heightened risk and focus area in the age of GenAI, many customers are re-thinking how much of their infrastructure they'd like to maintain privately.

All in all for VMWare, cross-selling to its existing customer base continues to go well; OpEx efficiency gains continue to be realized. Well-executed M&A.

Now that AVGO has debuted VCF and moved nearly 100% of VMWare customers to that new offering, it sees a new priority over the next 2-3 years. It needs to help customers drive use cases and value from VCF to show them how impactful it can be. That should drive workload and overall revenue growth for this segment.

Great quarter. The AI revenue ramp remains explosive and should keep accelerating for at least another year. Broadcom continues to win large contracts with some of the most important companies in the world and keeps proving how big of a piece of the pie XPUs can take over time. They flawlessly transitioned VMWare and have positioned it for structural growth, while a potential bottoming in non-AI semiconductor buckets could add fuel to the overall fire. The team is clicking and yet another strong performance shows exactly that.

This and a few of the other sections were shared in the Discord room during the week. Max readers, sign up! It's part of your existing plan. Non-Max readers, become Max readers. You get 40+ earnings reviews per season. Other big event coverage. Timely, contextualized news and analysis. My own portfolio/performance. And a community to go for noise-free collaboration. Sign up.

How are things going?

Branded checkout growth continues to be stable and resilient. The tariff impacts cited on the Q2 call have not changed much, and they’ve seen no material impact from the removal of global de minimis tariff exemptions. As a reminder, China-to-USA represents about 2% of their volume, while the tariff exemption removal impacts less than 3% of their overall volume. Leadership still expects to exit 2025 with branded checkout growing above the rate they entered the year with (so faster than 6% Y/Y). While this is a Y/Y acceleration, it still includes slight deceleration during Q3 and Q4. That’s partially related to assuming consumer spending levels will slow, but it doesn’t sound like they’ve seen that yet. Could be a source of upside.

When looking out further, CFO Jamie Miller feels confident in the company’s trajectory through 2027 that calls for transaction margin dollar and EPS growth comfortably ahead of consensus estimates. She didn’t want to give a finite timeline on the path to 8%-10% branded checkout growth by 2027, as it will not be linear. Miller still did call that goal “fully intact.”

Germany Service Disruption:

The impact from the resolved system update error and Germany outage will not have a material impact on payment volume. There will be a small transaction loss impact, but they do not see that preventing them from meeting guidance.

Branded Checkout:

PayPal continues to pocket expected conversion rate and market share uplifts for merchants with its latest branded checkout flow. They’re paying close attention to cohorts with this new process and like what they’re seeing. Miller expects to keep making steady progress with U.S. merchant adoption and even faster progress in Europe, where more of its merchants are on the newest checkout integration. That makes adding the upgraded checkout flow easy.

Miller also repeated previously shared disclosures, including debit card users on PayPal delivering a 20%-30% branded transaction uplift. These omnichannel customers enjoy margins that are the same or better than pure online branded checkout.

Continue to expect 20% Venmo growth over the coming years. It’s “just the beginning” of this acceleration, with Big 10/12 partnerships a compelling new way to ensure it’s a ubiquitous part of younger consumer lives.

Take Rate:

As a reminder, they don’t optimize for take rate. Products like payouts and debit cards have low take rates despite also being its highest-margin offerings. She talked about this in the interview. And while they aren’t focused on boosting this metric, they still expect it to be healthy in the coming quarters. Branded checkout pricing stability is the most encouraging reason why. Braintree contributions and better small and medium business (SMB) growth are the other two reasons.

Braintree:

As previously stated, they maintained more market share than expected following the contracting pricing renegotiations. Braintree has now contributed to profitability for 5 quarters, and revenue growth is set to turn positive next quarter (and accelerate from there). Great turnaround here.

Quick Take:

Nothing in here is at all concerning to me. CEO Alex Chriss and company continue to execute this gradual turnaround and morph PayPal back into a healthy, margin-accretive grower. I will remain patient with this name as long as the company is performing well. The underperforming stock will not deter my conviction as long as fundamentals remain solid.

This was shared in the Discord room during the week.

Changing Search:

Hoffmeister was asked about the evolution of search’s impact on Shopify’s business. As GenAI changes the way people shop and as search becomes a more popular shopping channel, the company needs to make sure they’re putting merchant products where eyeballs are moving. Leadership is on it. With this team, I never worry about them falling behind in the innovation race. As discussed in the earnings review, they’ve already debuted product catalog integrations with chatbots. They’ve also debuted a new product enabling consumers to use one cart to checkout with multiple merchants through a chatbot. Again… always innovating quickly… always putting their merchants in the best position to succeed and grow. This is how Shopify is accelerating past 30% Y/Y top-line growth while the Salesforce Commerce Cloud grows at a low single-digit clip. This is how it keeps taking market share every single quarter.

“If you're not using the Shopify platform, you're making life harder than it should be. That’s what we’re trying to do.” – CFO Jeff Hoffmeister

Winning in Large Enterprise:

We often attribute Shopify’s product breadth and quality as the main reason why they’re taking share with larger enterprises. And while that’s true, there’s another (related) tailwind forming. Shopify has effectively proven, across a wide array of massive customers, that it can handle their traffic with optimal performance, great up-time and lower total cost of ownership. The big boys no longer need to bet on Shopify’s platform superiorly scaling with their needs. They can observe that happening for their competition, which is creating a compelling domino effect and another demand accelerant. Want to keep up with those competitors? Use Shopify.

Hoffmeister told investors that Shopify has possessed these capabilities for years. They just really haven’t focused on communicating that in go-to-market. They also haven’t packaged products in a way that appeals to large enterprises. Through Shopify Plus, Commerce Components by Shopify (à la carte buying), headless commerce (front-and-back-end separation for more customization) and go-to-market prioritization, they’ve done the work to win going forward.

Leverage:

Hoffmeister said he “feels good about what Shopify can do with this headcount.” It sounds like revenue growth will greatly outpace headcount growth going forward, which should be a material source of incremental operating leverage.

This was shared in the Discord room during the week.

Growth Focus:

There has been considerable debate surrounding the recent slowing in DraftKings handle (volume) growth. As indicated on the earnings call, the team can generate materially more handle growth right now if they want to – that’s just not the current focus. Instead, the past year+ has been spent optimizing promotional spend and driving higher hold rate (take rate) through mix-shift towards parlays. These focus areas (especially the promo item) have held back handle growth; they’re excited to make that current trade-off because they get “so much value” from doing so.

Margin Puts & Takes:

AI has given Robbins a “rare” degree of excitement about their fixed cost structure. They think AI agents are creating considerable opportunity to do more with less. This could minimize future headcount-related growth expenses. They’ll still hire some engineers and in a few other areas, but less aggressively than in the past.

Next, Robbins flat out told investors to bake some level of rising tax rates into their multi-year forecasts. He sees this as all but inevitable, but thinks it will be with lower tax rate states playing catch-up with others. As long as there aren’t more progressive tax rate states (punishing more scale) like Illinois, this isn’t alarming and it affects all competitors equally. They’ve shown an ability to recover a lot of the cost headwind and operate at strong margins in high tax rate states like New York. And while this will be a margin headwind, he believes there’s upside to the long-term 30% EBITDA margin goal, if anything. They expect the headwind to be consistently present; they expect to overcome it through other sources of operating leverage. 

And as a reminder, there are also meaningful regulatory tailwinds to look forward to. iGaming legalization is in only 7 states with DKNG having the largest presence in the space; sports betting is still not legal in 12 states.

Outcome Luck & Parlays:

Skepticism regarding past “bad outcome luck” is fading following the very good luck they enjoyed this past quarter. Outcome-related volatility will inevitably continue and DKNG has changed its guidance methodology to reflect this. It will no longer offer a specific revenue target, but instead a range to reflect the unpredictability of sporting results. Probably should have already been doing that, but good change nonetheless.

Robbins was also asked about parlay mix raising bet concentration and adding to outcome-related risk. They’re happy to accept this in exchange for higher parlay bet share, as that is objectively positive for lifetime value. That’s why DKNG continues to round out the parlay offering with products like “Stacks.” This is a “narrative-based” parlay builder that suggests legs to add to parlays based on other pieces of the ticket. For example, if I bet on Jared Goff to throw for 300 yards and 4 touchdowns while Jahmyr Gibbs and David Montgomery each run for 100+, it might recommend adding the Lions money line to the mix.

Prediction Markets:

Robbins does not feel pressure to be the first mover in prediction markets and has chosen a wait-and-see approach. When/if they do offer this, initial focus would be on introducing this in the non-legal sports betting states. That’s because he thinks sports betting will take the vast majority of overall market share in states with both. Robbins pointed to the UK, where both exist and where prediction markets are a single-digit % of overall bookmaker revenue. When sports betting is available, that’s where people flock. When it isn’t, they’re more likely to try prediction markets. The patience is also to avoid ruining regulatory relationships they’ve worked so hard to build.

The DraftKings CEO also thinks that online sports betting will be a much better product than peer-to-peer prediction markets over the long haul. He doesn’t envision these companies ever having the full feature set that DKNG has because of risk management issues when you’re an exchange instead of the market-maker. DKNG can block and limit world-class betters to manage risk whenever they feel like it, which means they can offer a lot more bet assortment to players while still protecting the company. For prediction markets, it’s much more of a free-for-all, without control over who takes the action and with a requirement to allocate funding for every single possible bet outcome that their uncontrolled user base makes. 

Interestingly, Robbins eventually sees DKNG entering other financial services to compete with vendors like Robinhood. I’d like to see them focus on sports, iGaming and prediction markets. If they do enter the prediction market space however, that will require licensing that would allow them to offer other financial services. He doesn’t see downside to experimenting and that’s fine as long as  it isn’t a distraction to core operations.

iGaming:

They’re doing a much better job attracting slots-first customers in recent months. Table games are doing very well as is sports customer cross-selling, but slots market share greatly lags those two categories. It thinks that’s because of marketing, not product, and has gotten much more vocal in messaging. Robbins is seeing trends improve accordingly. While he didn’t want to say when this would begin to materially benefit financials, he indicated that benefit would eventually (and materially) come.

Jackpocket:

Jackpocket is enjoying “crazy high customer acquisition numbers” with a $1B+ jackpot. They’re finally seeing the M&A thesis come true and expect this to be a great source of cross-selling to and from its core product. 

A variation of this was shared in the Discord room during the week.

Throughout the latest SBUX earnings review, we talked about the long list of subtle signs pointing to Niccol turning this troubled bluechip around. We got another encouraging data point this week. CEO Brian Niccol publicly shared that the return of SBUX Pumpkin Spice drove its largest sales week ever… in North America. That has been its most troubled market aside from maybe China, and is starting to show real signs of a comeback.

I continue to think Niccol is the perfect person for this job. Things were done so unfathomably poorly that the list of high-value/easy fixes was miles long. Just like he did with Chipotle, I expect him to turn this company back into a fundamental darling and a margin-accretive compounder. Whether it’s in-store improvements, data-driven menu decisions (what a concept), revamped marketing and loyalty programs, throughput improvements and more, he’s a fat kid in a candy store in terms of everything he can do to drive value. We just need to give it some time.

Bids for a stake in the Starbucks China business are coming in around $5B – or 10x forward EBITDA. They're fielding 10 non-binding offers and could choose one in the coming weeks.

A variation of this was shared in the Discord room during the week.

Well… we finally got the anti-trust ruling we’ve all been waiting for. And thankfully, a forced Chrome sale is not happening. This is great news. It preserves the fortress, diverse product distribution ecosystem and the data advantage that comes with it. It’s this proprietary, scaled data that enables Gemini models to top leaderboards and enjoy faster pace of improvement than the competition. It’s hard to overstate how important this dynamic is for their AI value proposition, and this ruling keeps that intact. They will need to share some level of search data with competition, but that concession is considered quite modest, and the most feared outcome was avoided.

The forced Chrome sale was a risk that wasn’t seen as overly likely. On the other hand, cutting Alphabet’s default Google Search contract with Siri was seen as probable. This is also not happening. Alphabet and Apple will no longer be able to sign long-term exclusive default search deals that block competitors. But? Alphabet will not be forced to halt current payments and the pre-loading of Google onto Safari can continue. All that needs to happen now is these contracts being negotiated on an annual basis. A slap on the wrist… if that.

But wait, there’s more. Rumors that Apple will use Gemini to power its upgraded Siri product are advancing. Apple will test these models in its future product, with an expectation that it will be selected if things go well.

Between these positive headlines, it’s easy to see why Alphabet enjoyed some multiple expansion this week. But at 20x forward EPS for a name like this, I think that expansion could easily continue.

This was shared in the Discord room during the week.

SemiAnalysis came out with a new AWS note this week. They talked about an "AWS AI resurgence." This is thanks to the successful innovation from its Trainium2 training/inference chip and a near-future explosion in Anthropic revenue. As a reminder, Amazon has invested $8 billion in Anthropic while its Trainium2 chip is being heavily used by that model builder. The article points out (as Amazon many others have) that Trainium2 is not on par with Nvidia Blackwell from a gross performance perspective. But? It delivers memory bandwidth efficiency gains and lower total cost of ownership (TCO) for Anthropic. SemiAnalysis sees a large chunk (1+ gigawatt) of AWS data center capacity coming online by the end of the year and supporting a revenue growth ramp above 20% Y/Y for AWS. Consensus is right around 18% for Q4. They see Anthropic taking a lot of that capacity and greatly contributing to overall revenue as some training/inference demand goes to AWS. While Anthropic is behind OpenAI in the revenue ramp, that lead is expected to close as this model builder foresees a period of hyper-growth. That will benefit Amazon (and Google). So far, most Anthropic demand has gone to Google Cloud as that was the first partner. 

A variation of this was shared in the Discord room during the week.

DA Davidson downgraded Duolingo this week. While I don’t pay much attention to the price targets, I do pay close attention to the data supporting these opinions. For this specific note, DA Davidson sees daily active user (DAU) growth tracking below consensus estimates for a second straight quarter. This led to considerably more share price weakness.

The glass-half-empty view is that the AI-backlash leadership thought they were past is still raging. Skeptics argue competition from GenAI chatbots is passing this product and teaching better than it can. They’d finally say the social media team leaving means that consistent source of growth is ending. That’s what they say.

The glass-half-full view, which is where my mindset lies, is that DAU growth is powered by their social media marketing engine. The AI-backlash forced them to pullback on viral social media marketing content while they recovered negative sentiment. That process was expected to wrap up during this quarter, which is when DAU growth was expected to recover. This social media engine is not tied to one person. It’s tied to a beloved brand with a mountain of data and a world-class ability to use it. Zaria Parvez was a star… but she’s replaceable. For this reason, the note on DAU growth tracking below consensus for the first part of the quarter makes sense. They’re operating with a hand tied behind their back and that will soon end. Intuitively, back-half DAU growth should be better. The other item is related to a product change DUOL recently conducted. They effectively blocked users who were excellent at using the product daily and gaming the system to avoid payment. So? That obviously led to some disgruntled users and some top-of-funnel weakness. But that’s also expected to be a positive for bookings and overall revenue. 

As I’ve said recently, this has fully morphed into a battleground stock. There are high-profile bears calling for continued weakness and polarizing debate surrounding the future investment case. 

And while all of this noise rages, I will stay calm. This team has admirably executed for several years and features backgrounds that would impress anyone. Duolingo is a fantastically run company with a core business that centers around fun and competition that is also productive. Chatbots are arguably more productive for people learning a language. But they don’t come close to Duolingo’s entertainment factor and massive base of competing users. That’s the core niche, while DUOL will also lean on preferred partnerships with players like OpenAI to also ensure the latest and greatest tech is in its product too (like the AI FaceTime tool). Furthermore, its other segments are growing like weeds and have massive runways – just like language learning still does. While I understand the skeptics, the business quality, execution and opportunity all lead me to stick with this name. I’d need to see structural decay in actual numbers to grow sour… not some alternative data (that has been wrong in the past) looking underwhelming. They have more than earned my trust and my patience and I expect that to continue through more strong quarterly results. 

My bullishness is always earned and never unconditional. I’m always open to cutting ties with holdings as my mind is wide open to being wrong. For now, I think panicking due to bad price action… while DUOL grows at 30%+, expands margins and sports a sub-30x FCF multiple… is not the right decision for me. I’ve made a lot of money in this name since the IPO and it’s still my expectation that I make a lot more.

Meta is reportedly exploring Gemini and OpenAI models to complement Llama for its Meta AI product. There are also mumblings about some AI talent leaving. To me, that is inevitable when you’re creating this much organizational change. Some people are bound to no longer be a cultural fit. That’s what we’re getting through.

Taiwan Semiconductor’s Validated End User (VEU) permit was revoked by the U.S. administration for its Nanjing Plant in China. The company no longer has open permission to get equipment from the USA for that facility, but it can seek permits on a shipment-by-shipment basis (which creates timing & regulatory risk). This facility makes up less than 2% of TSM's total production. It only makes 16nm and 12nm chips there – not advanced tech chips that are vital for GenAI. And it's confident it can weather this change without disruption. Intimidating headline., but this will not sharply impact its results.

SentinelOne partnered with Schwarz Digits to extend European go-to-market.

Redburn upgraded Chipotle to a buy. They think current result softness is macro-driven and think the brand and business model are in great shape.

Uber closed its $300M Lucid investment.

Output data:

Employment & Wage Data:

Employment data from the week is a bit concerning. It shows cracks in the labor force becoming a tad less subtle. 4.3% unemployment is still quite good, but the payroll numbers have not been for a few months. With consumers dominating economic spend in the USA, this is vital to keep an eye on. Yes... it is all but ensuring rate cuts come next month, which is good for liquidity and risk asset valuations. But rate cuts are a lot more positive when they're due to disinflation. Not disinflation paired with rising unemployment. This data isn't alarming to a point of me feeling a need to raise a lot of cash. But it is not great and needs to be closely watched.