1. Alphabet (GOOGL) — Earnings Review

a. Demand

• Beat revenue estimates by 0.5%. Foreign exchange neutral (FXN) revenue growth was 15% Y/Y.
• Beat search revenue estimates by a comfortable 2.0%.
• Beat cloud revenue estimates by 2.6%.
• Missed YouTube advertising revenue estimates by 3.2%.

b. Profits & Margins

• Beat GAAP EBIT estimates by 3.4%. OpEx rose by just 5% Y/Y. R&D was the source of this. G&A fell Y/Y due to fewer legal charges and S&M was flat Y/Y/
• Beat $1.85 GAAP EPS estimates by $0.04.
• EPS rose by a robust 31% Y/Y.
• Google is lapping a $10.5 billion deferred cash tax benefit from Q2 and Q3 2023, which is making FCF comps very difficult. That, paired with hefty capital expenditures (CapEx) to build its AI infrastructure led to the weakness here.
• Google Services EBIT margin was 40.1% vs. 35.4% Y/Y.

c. Balance Sheet

• $101B in cash & equivalents.
• $13.2B in debt.
• Diluted share count fell 2% Y/Y. Bought back $15.7 billion in stock vs. nearly $15 billion Y/Y.
• Paid out $2.47 billion in dividends vs. $0 Y/Y. Initiated its first dividend last quarter.

d. Guidance & Valuation

Google continues to expect EBIT margin to expand Y/Y in 2024. It is the only mega-cap tech company that continues to commit to this. Still, Q3 EBIT leverage may be more muted than the first half of the year. The added CapEx for data center capacity started in Q3 2023 and will begin to lead to higher Y/Y depreciation expense in Q3 2024. Furthermore, it moved a Pixel hardware launch and the coinciding expenses to Q3 2024, which will mean more revenue… but at lower margins. Finally, Google will begin to lap Asian merchant seller strength that began in Q3 2023, which could weigh on revenue growth and profits a tad.

• Reiterated that quarterly CapEx will remain at or above $12 billion in 2024. CapEx rose from $6.9 billion to $13.4 billion Y/Y this past quarter.
• Headcount will grow a bit next quarter as it hires new graduates.

Google trades for 23× 2024 EPS. EPS is expected to grow by 31% Y/Y this year and by 14% Y/Y next year. Here’s how its current multiple compares to historical norms:

e. Call & Release Highlights

The main theme of the call was how GenAI is being infused across its product suite to improve product utility and monetization potential. Notably, Google is already generating “billions” in revenue from this GenAI work and has 2 million developers using its tools to date. 1.5 million developers are now using Gemini tools specifically. As a reminder, Gemini is Google’s series of foundational models used across its suite of products. It leads the world on key performance indicators such as long context understanding… and this brings us to another key idea. 

GenAI model winners will be those with the most money, infrastructure, talent and data. Google provides all 3 of those strengths as well as any other firm on the planet. Scale matters… GenAI favors established incumbents… this is that incumbent. Google is internally innovating across all layers of GenAI – hardware, models & apps – and thinks its in-house approach will allow its product teams to move faster than others. As foundational models and GPUs race to commoditization, that extra inch of innovation will make all the difference in the world.

GenAI Hardware News:

• The 6th generation of Google’s Tensor Processing Unit (TPU) for neural networks and machine learning is 67% more efficient with 5x higher peak compute capacity than the 5th version.
• Google’s A3 was a high performance compute chip. Google and Nvidia teamed up to create the A3 Mega, which is equipped with Hopper 100 (H100) GPUs. A3 Mega doubles GPU networking capacity compared to A3.
• Google isn’t worried about overbuilding data center capacity or overbuying chips. It has many ways to allocate excess compute (these assets have long useful lives) if it thinks it has too much. Leadership sees the real risk in not building enough.

AI Customers so far include – The U.S. Air Force, Uber, Deutsche Bank, Best Buy, Mercado Libre & Gordon Food Service (and almost all private GenAI unicorns).

Search:

Search was arguably the most impressive piece of this quarter. While this company was thought to be in trouble a year ago as Microsoft stormed back onto the search scene, it seems to have admirably fended off this competitive threat. A 2% beat on a revenue metric with this much analyst coverage is (in my mind) quite strong.

Search Generative Experience is now called AI Overviews. This gives GenAI-powered responses to fielded queries at the top of Google search results. These experiments across North America and some other geographies are showing clear signs of higher engagement and higher user satisfaction (especially for younger users). More value always leads to monetization potential for Google, and it doesn’t think this time will be any different.

Importantly, Google isn’t supplanting the publishers and merchants that have made its search database special… it’s using AI overviews to feed them more traffic. New ad placements right above and below these passages are proving quite valuable and relevant for searchers. Google will add shoppable links to these AI overview ad placements in the coming months.

In Google’s obsessive pursuit to “re-engineer the cost base” foundational model usage is a key focus area. It now has 4 different Gemini model sizes with various amounts of parameters. It’s actively working on better matching these 4 models with the appropriate types of queries. This will help cut cost per GenAI query, which had been a key concern for Google until very recently. It will also improve latency and the overall user experience (UX). Commentary last quarter about the margin profile for GenAI search ads being similar to other ad types was encouraging.

New AI-powered search tools:

• Search by taking a picture with your Android device. This pairs very nicely with AI overviews.
• Circling an item on the web to search debuted this week.
• Virtual try-on-and-shop ads are now beta testing. This is leading to large engagement and click through rate (CTR) uplifts early on.
• Project Astra is a new multimodal AI model Google is working on for day-to-day life. This is a buzz phrase to remember in the coming quarters.

Cloud:

• Google was not asked about its rumored bid to purchase a cloud security platform called Wiz.
• Google thinks its cloud platform’s cost for model training and inference is an edge. Most public cloud platforms will say the exact same thing.

Advertising Tools:

Google introduced 30 new GenAI tools for better campaign creation, targeting and measurement while upgrading several existing tools. There are 4 big tools to discuss. Performance Max (PMAX) is its campaign creation service with Gemini infused right into it. The automated campaign material generation tool PMAX offers leads to a 6% conversion boost. Demand Generation (Demand Gen) is exactly what it sounds like: a tool to extend audiences and sharpen targeting to uplift the number of relevant eyeballs a campaign can affordably reach. Broad Match is another tool that it uses to sharpen ad matching once demand generation has uncovered the most desirable customers to reach. Product Studio is its group of free AI tools to create images and designs for advertising.

• Demand Gen + PMAX = 14% higher conversions for advertisers. When also using broad match, that positive conversion impact rises to 25%.
• Advertisers using its AI-powered impression profit optimization tool are enjoying a 15% profit uplift vs. advertisers only using revenue as their key performance indicator (KPI).
• Tiffany used Demand Gen to deliver a 2.5% uplift in brand consideration, 5.6x lower cost per click and a 100% boost to CTR.

YouTube:

YouTube revenue was the lone disappointment on the top line this quarter. The reasons weren’t very concerning. Comps simply got much tougher, it lost the benefit of leap day, FX headwinds intensified and content timing issues were the icing on the cake. Leadership remains adamant that the underlying business tends and the ad environment are both fine.

• 17 straight months of leading streaming market share. Across all content forms, it trails only Disney.
• Sports streaming viewing hours rose 30% Y/Y.
• YouTube Shorts monetization rate continues to improve Q/Q.
• Shopping ads will be a key focus area going forward. Watch time for these ads rose 25% Y/Y as it introduced product tagging to help with product ID and its affiliate marketplace to help match creators with opportunities.\

Final Notes:

• Google is earmarking another $5 billion for Waymo.
• Google won’t deprecate cookies as it “now favors consumer choice.” It probably has more to do with how many times its privacy sandbox product replacement has been delayed.

f. Take

This was a good quarter. The search revenue result was excellent, YouTube revenue was meh, and the rest of its segments were strong. YouTube weakness seems to be transitory and outperformance across other segments seems to be more structural. I find it impressive that this company is investing as aggressively as it is in GenAI while briskly expanding margins. I find it impressive that search has fended off Microsoft, OpenAI and others so effortlessly. I find Google impressive. Good quarter from an elite team and an elite company.

2. Tesla (TSLA) – Earnings Review

a. Demand

• Beat revenue estimates by 3.2%.
• Beat auto revenue estimate by 0.9%.
• 2 out of 3 Tesla sales during the quarter were to brand new customers.

b. Profits & Margins

• Missed EBIT estimates by 13%. OpEx rose 39% Y/Y.
• Missed $0.58 EPS estimates by $0.06.
• Missed $0.49 GAAP EPS estimate by $0.07.
• Beat GAAP GPM estimates. This was powered by impressive profit strength from its non-auto businesses.
• Missed 16.2% Auto GPM ex-regulatory credits by 160 basis point (bps; 1 basis point = 0.01%). Auto GPM ex-credits: (auto rev ex-credits - auto COGS) / (auto rev ex-credits).

Tesla makes no non-GAAP adjustment to its EBIT. This quarter, it incurred a $642 million GAAP charge related to restructuring. Without this impact, it would have generated $1.85 billion in EBIT and beaten depressed estimates comfortably. Still, the non-GAAP EPS miss does tell me the sole source of the profit misses was not one-off charges. There were other factors at play, which are discussed later on.

c. Balance Sheet

• $30.7 billion in cash, equivalents & investments.
• Inventory levels rose 43% Y/Y.
• $7.7 billion in total debt ($2.26 billion is current).
• Diluted share count was roughly flat Y/Y.

d. Guidance & Valuation

Tesla doesn’t offer much forward guidance. It sees slower auto growth in 2024 vs. 2023, strong energy storage performance and no need for any capital raises.

Tesla trades for about 95× 2024 EPS. Earnings are expected to fall by 18% Y/Y this year and rise by 38% Y/Y next year. Estimate trends had begun to bottom before this report, but this will likely lead to more downward pressure. Here’s how its earnings multiple compares to historical norms:

e. Call & Letter Highlights

Macro:

A heavy discounting environment from peers and high interest rates continue to heavily weigh on this business. Automotive and energy are both cyclical. Tesla’s structural share gains do buffer some of that cyclicality, but not all of it. Musk sees the pricing competition as very temporary and something the company can seamlessly weather.

Auto Business:

The mass market $25,000 model is on track for 1H of 2025 deliveries. Generally speaking, its new lineup of vehicles is still on track for 1H of 2025 deliveries too. As a reminder, these pull from existing manufacturing capacity and are more so iterations of existing models. The next-gen vehicle lineup should be a 2026 event.

Tesla continues to lean into more financing programs and affordability initiatives (like some price cuts to models and FSD) to help consumers through these tougher times. This will remain the case in Q3. Bulls will tell you what Tesla leadership tells you. They think these decisions make it even harder for negative GPM EV alternatives to try to compete with Tesla. They also think Tesla can “harvest more margin” down the road via software upsells like full self driving (whenever that comes) to eventually plug this upfront profit gap. Finally, bulls will rightfully call out how positive near-future rate cuts would be for this business. 

What do bears say? That competition is catching up, Elon’s polarizing dialogue and focus on XAI are hurting the brand and investor sentiment. They’ll say FSD is a pipe dream and will simply keep getting delayed as Waymo and others race ahead. As I have been since I started investing, I’m entirely on the fence. I see credibility and weakness to both sides of this argument and have Tesla placed firmly in my too hard pile to invest. If I had to own shares or short it, I’d own shares… Elon is impossible to bet against. But luckily, as Buffett tells us, we play in a no strike game.

Cybertruck is still ramping and still a cost per vehicle headwind (Model 3 is still ramping to a lesser degree too in some factories). Without that impact, cost per vehicle fell Q/Q as it continues to enjoy better auto scale. Maturation of its 4680 cell is also a key source of this improvement, as production rose 50% Q/Q and costs tumbled. General input cost disinflation likely helped a bit too, although that wasn’t mentioned on the call.

“We saw a sequential rebound in vehicle deliveries in Q2 as overall consumer sentiment improved and we launched attractive financing options to offset the impact of sustained high interest rates.”

Slide Deck

• Global EV share of the auto market returned to positive Q/Q growth. 
• Cybertruck production rose 3x Q/Q (small base). It’s on track to debut its dry cathode 4680 cell for the Cybertruck, which will be a “major cost reduction milestone.”
• Regulatory credits were highly elevated due to fulfilling emissions requirements.
• Discounting and lower deliveries were revenue headwinds while Cybertruck and regulatory credits were tailwinds.
• AI investments led to sharper OpEx growth while continued 4680 cell maturation helped Tesla find some operating efficiency gains.
• The Shanghai factory materially grew deliveries Y/Y.

Tesla Market Share Chart:

FSD, Robotaxis & Optimus:

Tesla set a new October 10th date for its robotaxi event to give the team more time to add more products to the pending launch. Musk sees version 12.5 as a “step change” of improvement, which is similar to his commentary from the last few rollouts. 12.5 merges highway and city software for a slicker experience and continues to widen the gap between driver interventions. Elon sees unsupervised FSD as happening by the end of this year or next year. To be fair, he is routinely wrong and overly optimistic when it comes to launch timelines. He will explicitly tell you that. Flying reusable rockets and solving for autonomous driving is kinda hard.

“Based on the current trend, we should get miles between interventions high enough for unsupervised [FSD] possibly by the end of this year. I would be shocked if we cannot do it next year.”

CEO Elon Musk

In terms of licensing FSD to legacy manufacturers, that’s still the plan. These automakers, however, move very slowly. It will take them years to implement Tesla FSD in their own cars and for revenue to start flowing. That’s both annoying and encouraging for gauging the durability of Tesla’s EV edge. GM and Ford just don’t move quickly. Tesla does.

Musk doubled down on his belief that Optimus will be the firm’s most valuable product to date. Version 1 is planned for 2025 production for internal Tesla usage. It expects to have “several thousand” Optimus robots “doing useful things” in 2025 before launching this product for other external customers in 2026. Tesla is convinced that it’s far ahead here.

• Musk told investors that GM delaying its autonomous vehicle concept wasn’t due to regulators but that he thinks “GM can’t make it work.” Considering how quickly Waymo and Zoox are progressing, I agree with Musk.
• Tesla will soon ask for approval on supervised FSD in Europe and China. They’re confident in getting the needed licensing this year.
• Optimus is now performing some tasks in one of its factories.

Dojo, GenAI Hardware & XAI:

Tesla is hard at work on its supercomputer and GenAI infrastructure. It has its sights set on catching up to Nvidia, and finds that as an operational imperative. I found that highly interesting. Other mega-caps have all but conceded they can’t catch Nvidia’s GPU lead. Most are making their own chips for other, less complex use cases. Tesla says not so fast. Nvidia’s lead is fostering absurd pricing power. Tesla doesn’t want to pay up. And? It sees vertical integration here (and across all layers of the AI stack) as an efficiency force multiplier for margin and pace of innovation. 

Musk “supports” more direct collaboration between Tesla, Grok and XAI. He got a bit defensive about XAI, which was likely a response to all of the conflict of interest criticism he has recently received. To be fair, I do get why Tesla shareholders would not want GPU orders going to XAI. Musk told us that Tesla had no use for these GPUs yet, as it was still building out capacity to support them. He also told us his goal was to hire these people at Tesla. They supposedly just had no interest in working on anything but general AI.

In terms of how the autonomous fleets will work, Tesla cars will “only be used in the Tesla fleets and not for 3rd party autonomy. I still think it’s somewhat inevitable that these Tesla fleets will plug into Uber and Lyft demand to optimize occupancy rates and car margins.

It’s almost done with its Giga Texas expansion, which includes its “largest training cluster to date.” This includes 50,000 H100 chips. As a reminder, Tesla considers itself the “most efficient AI inference company in the world.” If they’re right (or even close to right), these investments make sense. It reiterated its $10+ billion 2024 CapEx guide.

Energy Storage:

• Energy deployments for Tesla were quite strong, with 158% Y/Y growth. Megapack and its Powerwall product both performed well. 
• The Powerwall 3 product (newest iteration in that line of hardware) is enjoying a very successful rollout.

“While we expect production to continue to grow sequentially, deployments will continue to fluctuate. Deployment timing depends on many factors, including project milestones and logistics timing as we deliver product globally from a single factory.”

Slide Deck

Politics:

European tariffs on Chinese car imports are leading Tesla to rethink its supply chain a bit. Per CFO Vaibhav Taneja, Tesla is “working furiously to offset rising costs, but they may impact profits near term. Tesla may lean away from Shanghai vehicle and battery exports to lower costs. It actually sees these new taxes as a layer of inefficiency that will benefit it long term. Why? Because inefficiency hurts weaker competitors with lower margins more than Tesla. Today’s relative pain is tomorrow’s material gain. Trump’s Mexican tariff threats are also leading to Tesla putting Giga Mexico plans on hold.

f. Take

This was a rather uneventful quarter for Tesla. We didn’t learn a lot. As long as rates remain this elevated and discounting from competition this steep, I don’t see how results could sharply inflect. Bulls are betting on that macro cycle turning (which it soon should) to get this auto business humming once more. If that can happen while unsupervised FSD is cleared, Optimus revenue begins to ramp, and its next-gen vehicle program finds traction, there could be a lot to like here. You just may have to wait a few (or several) quarters for certainty surrounding those projects to begin to surface. 

3. Spotify (SPOT) – Earnings Review

a. Demand

• Met revenue estimates & slightly beat its revenue guidance.
  • FX neutral (FXN) revenue growth was 21% Y/Y; FXN premium revenue growth was 22% Y/Y; FXN ad-supported growth was 12% Y/Y.
• Missed monthly active user (MAU) guidance by 0.8%.
• Slightly beat 245 million premium subscriber guidance by 1 million. Beat 6 million net new premium subscriber guidance by 17%.
• Premium revenue per user rose 10% Y/Y FXN vs. 7% Y/Y FXN growth posted last quarter. Price hikes were the main help, and accelerated overall premium revenue growth considerably. Still, a mix shift to higher-priced countries helped too.

b. Profits & Margins

• Beat 28.2% GAAP GPM estimate by 100 bps & beat guide by 110 bps.
• Beat GAAP EBIT estimate by 12.3% & beat guide by 6.4%.
• Beat $1.05 GAAP EPS estimate by $0.28. $1.33 in GAAP EPS for quarter compares to a loss of about $0.70 (excluding restructuring charges from the Y/Y comp).

c. Balance Sheet

• €5.4B in cash & equivalents.
• No traditional debt.
• Diluted share count rose 6% Y/Y; basic share count rose 2.8% Y/Y.
• Also guided to 639 million total MAUs and 251 million total premium subscribers.

d. Next Quarter Guidance & Valuation

• Revenue guidance met estimates. Guidance includes slightly elevated churn levels due to more price hikes.
• EBIT guidance beat estimates by 38%.
• 30.2% GAAP GPM guide beat estimates by 150 bps. Really good.

Spotify trades for 70× 2024 earnings. Earnings are sharply inflected positively Y/Y and are expected to grow by 44% Y/Y next year as well. These estimates will surely rise following today’s report.

e. Call & Letter Highlights

Monthly Active User (MAU) Variability:

Spotify from Q4 2022 through Q4 2023 Spotify enjoyed strong MAU tailwinds including product debuts, marketing efficiency gains and competition exiting some markets. That all helped user growth and all of those benefits were lapped last quarter. Temporary tailwinds always lead to tougher comps down the road.

During that last quarter, leadership also spoke about getting ready to lean back into more aggressive marketing spend. That led to its guidance of accelerating MAU adds throughout the year – starting this quarter. That did not happen. Spotify incurred “lower than expected” marketing spend, which, when paired with the large profit beats, tells me that’s where this weakness came from. That’s not concerning as long as it’s not running out of productive places to find new growth. That’s not happening as it sees subscriber growth “hasn’t shown any signs of slowing.”

Still Ek also spoke a bit about pursuit of new users in developing markets as its opportunity in North America and Europe becomes more mature. The return on investment in these developing markets still needs some work. To combat this, it’s focusing on crafting a more relevant, local message and also improving its free product to strengthen the top-of-funnel. Ek sees a “reversal in MAU trends as more of a when rather than an if.”

MAU to premium subscriber conversion was also quite strong, as you can see above in the outperforming result vs. expectations. This was especially true in developed markets where potential subscribers are the most valuable.

Profitability:

GPM expansion was powered by the same items on the premium and ad-supported sides. Better music and podcasting profitability, general “cost of revenue efficiencies” and lower-than-expected personnel costs (slower headcount growth) all contributed. Furthermore, the restructuring that Spotify conducted in Q2 2023 hurt GAAP GPM during that period. Of the 510 bps in Y/Y GPM expansion this quarter, 160 bps of that was related to the easier comps from this item. Audiobooks are still a material gross margin headwind. It’s incurring front-loaded costs to rapidly buildout and scale this business. Just like with its overall ad-supported segment, these investments should eventually slow as a % of overall business contribution to foster GPM leverage.

Spotify incurred €59 million in social charges during the quarter vs. €13M expected. Social charges are payroll taxes paid to employees. That expense is tied to its share price. Share price rose a lot, so charges rose a lot. Spotify doesn’t adjust guidance mid-quarter as social charge expectations evolve. But? Analyst expectations do freely float. So? That is why the EBIT beat vs. analyst consensus was so much larger than the beat vs. its own guidance. Operating expenses overall fell by 16% Y/Y due to lower personnel and marketing costs. Excluding the impact of social charges and restructuring, OpEx fell by 14% Y/Y.

“A few quarters ago, I said that while while many believe that Spotify has a great product, we needed to prove that we could also be a great business. I think we're really starting to show this now. “

Founder/CEO Daniel Ek

Products & Subscription Packaging:

For video podcasting, Spotify now has 250,000 shows in its catalog and 170 million users who have already interacted with this newer product. It also now has 250,000 audiobook titles as part of its premium subscription in Canada, Ireland and New Zealand. As of last quarter, this form of content was already incrementally raising engagement per user by a full hour. More engagement means more value which means more pricing power. In music, Spotify debuted live listening parties for an artist’s most passionate fans to get exclusive, intimate and early access to new music. Its AI Playlist tool, which lets a user create playlists with a conversational prompt, debuted in some markets this quarter too. Creative lab was announced as its in-house advertising creative agency to assist brand campaign building. Quick Audio was announced to automate ad buyer campaign content creation with GenAI.

As Spotify continues to bolster the depth and value of its suite, prices will surely rise to augment monetization. But there’s another key monetization tool that stems from expanding beyond a music library into podcasting, digital education, audiobooks and AI discovery tools. More products means more mixing and matching and subscription tiers to more granularly and relevantly match users with compelling offerings. It allows its customers to purchase in more of an á la carte manner to sidestep paying for content they don’t really want. That should improve premium funnel conversion. For example, it just launched a new Basic Plan in Australia, the UK and the U.S. to offer ad-free music without audiobooks.

“We implemented a price increase in several key markets, including in the U.S., which we're rolling out now with great success. In fact, we're seeing less churn in this round of increases than we did in our prior one, which was already very low by any measure… high engagement in developed markets gives us tremendous confidence in our ability to raise prices, allowing for strong revenue growth even as those markets continue to mature.

Founder/CEO Daniel Ek

“In the U.S. today, access to all of our content [elsewhere] would cost a user approximately $26, significantly more than a Spotify subscription. Spotify remains a pretty outstanding deal.”

Founder/CEO Daniel Ek

Ad-Supported Environment:

The Spotify Audience Network grew Q/Q publishers and advertisers. We didn’t learn much more beyond this. (SPAN is Spotify’s way of serving as a conduit between buyers and platform publishers across most podcasting apps. This enjoyed 10% Q/Q growth in participating publishers following expansion into 5 new countries last quarter.

• It called slower ad-supported growth a byproduct of volatile marketer spend. This seemed to be mainly in podcasting pricing, as music impression and pricing growth were both called tailwinds.
• Ad-supported revenue rose by double digits Y/Y in all markets.

User Growth Trends:

Y/Y MAU Growth

Q2-24

Q1-24

Q4-23

LatAm

22%

21%

22%

UCAN

18%

20%

22%

Europe

28%

29%

32%

Rest of World

33%

30%

24%

Y/Y Premium Sub Growth

Q2-24

Q1-24

Q4-23

LatAm

22%

21%

21%

UCAN

27%

28%

29%

Europe

38%

39%

39%

Rest of World

13%

12%

12%

f. Take

I thought this was an excellent quarter. The very small MAU and revenue misses are due to lighter than expected marketing spend. Profit metrics (especially gross margin) were all excellent and the guide points to that trend continuing. This company continues to slash costs, jack up pricing, maintain low churn and strongly compound revenue. That ability is somewhat rare and shows you exactly how sticky their offering truly is. Congratulations to shareholders.