It was a marathon week for earnings coverage and articles as peak earnings season raged on. In case you missed it, the following reviews were already published:
I also published a portfolio and performance update article.
Next week, I will publish Shopify, AppLovin, Spotify and Cloudflare reviews. For now, I’ve included financial overviews of each report. I will also write Cava, On Holding, Sea Limited and Hims.
1. Shopify (SHOP) – Brief Earnings Snapshot
The full SHOP earnings review will be published in the coming days.
a. Demand
- Beat gross merchandise value (GMV) estimate by 3%.
- Strength enjoyed across all merchant sizes, cohorts, sectors and geographies.
- Beat gross payment volume (GPV) estimate by 3.3%.
- Beat monthly recurring revenue (MRR) estimate by 1%.
- Beat revenue estimate by 3.8% & beat growth guide.
- Beat subscription solutions revenue estimate by 4.5%.
- Beat merchant solutions revenue by 1.5%.
b. Profits
- Met GPM estimate.
- Beat 37.8% merchant solutions GPM by 60 basis points (bps; 1 basis point = 0.01%).
- Beat 79.7% subscription solutions GPM by 120 bps.
- Beat FCF estimate by 21%.
- FCF margin expansion was 150 bps Y/Y excluding an expected benefit from merchant cash advance accounting changes.
- Beat $0.31 GAAP EPS estimate by $0.04 ex-equity investment help.
- OpEx was 33.5% of revenue vs. 37% Y/Y.
c. Balance Sheet
- $4.9B cash & equivalents.
- $525M investments.
- -0.8% Y/Y dilution.
d. Guidance & Valuation
- Low-30% revenue growth guide beat 26% growth estimate.
- ~27% gross profit dollar growth guide beat 23% growth estimate.
- ~20% FCF margin guide beat 18% margin estimate.
e. Key Call Notes
- Agentic shopping is growing like a weed for Shopify. Their catalog product is convincingly best-in-class, as it provides better product descriptions and metadata, easier integrations and broader channel access than any competitor. This is why it’s converting at twice the rate of alternative products that are scraping public data from the web to surface products. Overall, agentic commerce rose 200% Y/Y (still small base for now) and is proving to be incremental to traditional search traffic too. Search sessions are still growing at a strong and steady clip.
- Its AI assistant (Sidekick) enjoyed 260% Y/Y daily active merchant growth. Average engagement rose by 380% Y/Y as well, showing rising usage per customer. It’s helping businesses ramp revenue more quickly and make better decisions.
- Shop APP GMV rose 70% Y/Y as Shopify rapidly builds another scaled channel that none of its direct commerce competitors have free access to. This merely feeds their “sell anywhere” strength and extend the differentiation they enjoy here over the field.
- They continue to swiftly take large enterprise market share.
- Merchant solutions GPM was flat Y/Y. Interesting to see that after years of steady segment GPM declines tied to payments proliferation (carries a lower GPM). This quarter, they notched enough growth in “higher-margin revenue streams” to offset all of that ongoing pressure. It would be amazing to see that continue, although leadership didn’t commit to it. Sidekick scaling is currently pressuring subscription solutions GPM as expected.
- Shopify spoke about increasingly mixing different models for internal tasks just like so many others this quarter.
- Credit loss rates are stable.
2. AppLovin (APP) – Brief Earnings Snapshot
The full APP earnings review will be published in the coming days.
a. Demand
AppLovin missed revenue estimate by 1% & missed guidance by 0.5%. It blamed a worse-than-expected pace of model improvements for the miss. They took ownership for what was called an execution issue. The issues were fixed right after the quarter ended and began helping results as expected. Still, guidance for that period (as shown below) was still lighter than what we’re used to from them.
b. Profits
- Missed EBITDA estimate by 1.2% & missed guidance by 1.2%.
- Missed FCF estimate by 29%.
c. Balance Sheet
- $3B cash & equivalents.
- $3.5B LT debt.
- -1.5% Y/Y dilution.
d. Guidance & Valuation
- Revenue guidance met estimates.
- EBITDA guidance missed by 1.1%.
APP trades for 17x forward EPS. EPS is expected to grow by 67% this year and by 20% next year. I need to read the letter and call (and write the review) before commenting further.
3. Cloudflare (NET) – Brief Earnings Snapshot
The full NET earnings review will be published in the coming days.
a. Demand
- Beat revenue estimate by 4.5% & beat guidance by 4.7%.
- Beat remaining performance obligation (RPO) estimate by 6.6%.
b. Profits
- Beat EBIT estimate by 6.1% & beat guidance by 6.2%.
- Beat $0.27 EPS estimate by $0.02 & beat guidance by $0.02.
- Missed FCF estimate by 5%.
c. Balance Sheet
- $4.1B cash & equivalents.
- $2B notes.
- 3.7% Y/Y dilution.
d. Guidance & Valuation
- Raised annual revenue guide by 2.1% which beat by 2%.
- Raised annual EBIT guide by 5.8% which beat by 5.5%.
- Raised annual $1.195 EPS guide by $0.06 which beat by $0.055.
- Q3 was also ahead across the board.
NET trades for 200x forward EPS. EPS is expected to grow by 36% this year and by 31% next year.
4. Duolingo (DUOL) – Brief Earnings Snapshot
a. Demand
- Beat bookings estimate by 1.2% & beat guide by 2%.
- Beat daily active user (DAU) estimate by 1.2%.
- Beat revenue estimate by 1% & beat revenue guide by 1%.
b. Profits
- Beat EBITDA estimate by 8% & beat EBITDA guide by 8.9%.
- Missed FCF estimate by 1.8%.
- Beat $0.61 GAAP EPS estimate by $0.05.
c. Balance Sheet
- $1.2B cash & equivalents.
- $100M LT investments.
- No debt.
- 2% Y/Y dilution.
d. Guidance & Valuation
DUOL trades for 15x forward FCF. FCF is expected to grow by 8% this year, 20% next year and 5% the following year.
5. Spotify (SPOT) – Brief Earnings Snapshot
a. Demand
- Slightly missed revenue estimate & missed guide by 0.5%.
- The FX headwind was a few basis points larger than expected.
- Slightly missed monthly active user guide; slightly beat subscriber guide.
b. Profits
- Beat 33.1% GPM estimate by 30 bps & beat guide by 30 bps.
- Beat EBIT estimate by 3% & beat guide by 4%.
- 1.6% beat ex-payroll tax help.
- Missed $2.75 EPS estimate by $0.14; Missed FCF estimate by 6.5%.
c. Balance Sheet
- €9.4B cash & equivalents.
- No debt.
- 1.7% Y/Y dilution.
d. Guidance & Valuation
- Q3 revenue guidance beat estimates by 1.4%.
- Q3 GPM guidance exactly met 32.9% estimates.
- Q3 EBIT guidance missed estimates by 1%.
6. A Few More Notes on Earnings Season So Far
What another horrendous showing for The Trade Desk this week. They missed their “at least” revenue guide and analyst estimates by 5% each and EBITDA expectations by 8%. Their Q3 revenue guidance missed estimates by 20% despite those estimates having already fallen significantly. EBITDA guidance missed by more than 50%. Rough. They’ve had ongoing feuds with important agencies, seen their executive talent roster function like a revolving door and botched a massive platform launch. They have not executed well for about 3 years now, and the excuses don’t make it any better. This is why I sold the position a while ago. When that happened, it was with the hope that the company would turn things around so I could eventually re-enter. Instead of that happening, things have gotten steadily worse. I do not think covering this is worth the time or energy, and I will not be focusing on it going forward.
Next, the Atlassian, JFrog, Twilio, Cloudflare and Axon (yes Axon) quarters were all excellent. Figma got punished for decent numbers too. Hubspot wasn’t amazing, but most enterprise software earnings data looks quite healthy. Winners are becoming more obvious to the masses and finally getting credit for being winners.
Separately, there’s a clear earnings theme among cloud providers & software giants like ServiceNow & Palantir:
- Single-model preference is going away. Customers want the cheapest option & an easy, automated ability to mix & match models. That, rapid open source progress & Google/Meta/Microsoft undercutting frontier models on token cost all foster the price cuts we’ve seen recently.
- Using several models is the norm. Open & closed. Customers pushing back against model lock-in.
- Real value is in enterprise data (duh) & using it to harness model potential into orchestrated, reliable, enterprise-specific work. Not using a model built for a standardized benchmark. Using one that’s tuned for you.
- Customers are tired of paying for straight tokens. They want to pay for value creation & better outcomes. And they want help delivering those better outcomes.
The trend felt inevitable. But it’s now playing out in full swing. In my opinion, the implications are bearish for pure-play model companies and not really anyone else. Hyperscalers don't need to directly profit on selling inference tokens. Based on current business models, OpenAI and Anthropic do. Hardware should also benefit from deflation across the supply chain (the Jevons Paradox stuff) as lower costs have always driven enough usage in core cloud to make it a net positive. And this would also be good news for software if they get cheaper access to needed models for work.
Finally, I put together a quick snapshot of Zeta earnings for Discord members during the week (see below) This could be a name I consider adding to the coverage network to replace TTD in ad-tech.
7. Alphabet (GOOGL) – Talent Leaving
Google DeepMind CEO Demis Hassabis is changing his role. He will become its chairman and chief scientist and will lead Isomorphic Labs (drug discovery spinoff from Alphabet). Koray Kavukcuoglu was also named DeepMind's new SVP. He has been there for 13 years. Current Chief Scientist Jeff Dean and another AI employee are also leaving to create their own AI company. Google is investing in the project. This makes sense to see some changes following the rumored issues with the new Gemini release, but in this case, the Dean news is just him ready to do his own thing. Can’t blame him. Amazingly talented and driven people (like he is) generally like to see how far those talents can go. DeepMind's talent bench and Alphabet's team (with Brin involved) make me confident in them rearchitecting the teams to compete effectively in the model race. While there have been some departures, there is still a long list of world-class engineers that any other research lab would love to have. There’s also Sergey Brin, which helps a ton.
In other news, they’re raising another $25B via a 10-part bond deal.
8. Headlines & Macro
- SoFI partnered with CAZ Investment Partners to expand private market investing access.
- Meta launched a new AI coding tool with its latest Muse Spark 1.2 model. It’s a small fraction of the cost of leading models Anthropic and OpenAI.
- Amazon’s Zoox will launch in Las Vegas next week.
- July manufacturing data was strong across the board this week.
- The non-manufacturing purchasing managers index (PMI) was a tad weak.
- Jobs data wasn’t very good this week. ADP Nonfarm Employment Change was 44K vs. 68K expected. Labor unit costs rose by 1.3% for Q2 vs. 2.2% growth expected (wage growth = more consumption = more GDP growth). Non-farm payrolls were also -23K vs. 85K expected. Overall government jobs fell by 60,000 jobs, which hurt. Still there was weaker hiring across many sectors as businesses navigate higher prices and (in some cases) implement AI-based automation. I didn’t love seeing this, but I did love seeing the unemployment rate again edge down to 4.1% for July vs. 4.2% expected and 4.2% in June. That’s great to see.