Today’s Piece is Powered by Borsa — Where Spotify & Earnings Calls Collide:
1. Meta Platforms (META) – Earnings Review
a. Demand
Beat estimate by 2.8% & beat guide by 4.8%. Its 12.6% 3-year revenue compounded annual growth rate (CAGR) compares to 16.6% Q/Q & 19.6% 2 quarters ago.
Without unexpected foreign exchange (FX) help, revenue would have beaten estimates by 0.5% and guidance by 2.5%. FX neutral revenue growth was 22% Y/Y.
Reality labs beat by 31.6%. Note that the Quest 3 launch propped up revenue here vs. the prior year period. Quest 3 launched in Q4 2020.
Family of Apps (FOA) beat by 2.2%.
Meta set a 2 year record for DAUs as a % of monthly active users (MAUs). This is direct evidence of improving engagement.
Beat total daily active user (DAU) estimate by 1% & slightly beat total MAU estimates.
Price per ad rose by 2% Y/Y vs. expectations of -4.1% Y/Y growth.
Ad impression growth roughly met expectations.
Note that Y/Y comps are easy and 3-year comps are uniquely difficult.
b. Margins
Beat EBIT estimates by 4.7%.
Without any restructuring charges, EBIT margin would have been 43.6% instead of 40.8%.
FOA EBIT beat by 7%.
Reality Labs EBIT missed by 13.4%.
Beat free cash flow (FCF) estimates by 28%.
Beat $4.94 GAAP EPS estimates by $0.39.
Note that Y/Y margin comps are very easy and 3-year margin comps are very difficult.
c. Guidance & Valuation
Next quarter guidance was a robust 5.5% ahead of expectations. Very strong for a company at this scale with as many analyst eyeballs on it as it has.
2024 operating expense (OpEx) guidance was reiterated at $96.5 billion. Capital Expenditure (CapEx) guidance was raised from $32.5 billion to $33.5 billion due to ramping investment in AI infrastructure to support the technology’s boom. Infrastructure investments will continue to grow beyond 2024.
d. Balance Sheet
Meta declared its very first dividend of $0.50 per share. This may not seem meaningful, but it opens Meta up to significantly more investment demand. Many funds focus strictly on dividend-paying investments.
$65 billion in cash & equivalents (I wish regulators would let it buy something).
$18.4 billion in debt.
Basic share count fell 2.8% Y/Y.
$80.9 billion left in buyback capacity as it added a new $50 billion buyback program.
e. Call & Release Highlights
Playbook for Leading in AI:
Zuck’s talk was dedicated to laying out Meta’s 5 part plan to lead in AI in the decades to come. I’ll spell it out here. Part one is building world class GenAI infrastructure. Zuck spoke on the criticism he took for the heavy CapEx through 2021-2022, but now sees that decision as setting Meta up very nicely in the years to come. The company underspent when it debuted Reels on GPU clusters to support the product. So? It decided to get aggressive with a large, 350,000 Nvidia H100 Chip order this past month to ensure that it has the supply it needs. It’s going to “remain on the offensive here” and will “play to win.” Exploding FOA profitability and profits overall give it the permission to do this in my view. Zuck thinks “general AI” (multi-modal models that can do everything from reason to plan to code to remember” will be vital. This is how it will make sure it has the resources to deliver what it needs to. Investments here will remain aggressive in the years to come, as its guidance spells out.
“After underspending on Reels, I decided that we should build enough capacity to support both Reels and another Reels-sized AI service that we expected to emerge to avoid that situation again. The decision at the time was controversial, but I’m so glad we made it.” – King Zuck
Part two is its open source approach to software. Whether it’s through Llama 2 or subsequent models, Meta makes its “general infrastructure” available to all developers “while keeping specific product implementations and data proprietary.” This allows it to thread the needle of not giving up competitive edges while letting 3rd party developers build in its ecosystem. What does all of this building mean? It makes Meta a low cost operator for innovating in GenAI. Developers want access to consumer traffic and great models to build products they know will be utilized. That’s what Meta delivers with Llama and 4 billion people on its Family of Apps. By outsourcing some of this iteration and innovation work, Meta sheds costs associated with product development.
Foundational models are likely going to be commoditized in the future. Knowing this, Meta is flooding the market with open-source Llama models (like it’s doing with Quest) knowing that it can monetize elsewhere with subscriptions and app take rates. If there’s no differentiation between this great model and that great model where will consumers go? Where the best consumer apps are. And where will developers go? Where the consumers are. The open source approach nurtures this compelling positive feedback loop. Meta always takes its sweet time with monetization. Whether it was Facebook, Instagram, WhatsApp or now AI, it will only monetize once it knows it has already won. Other advantages of open sourcing include open-source simply being more popular among most developers. This helps it attract the best talent in the world.
“Usually, only when we reach that kind of scale do we start focusing on what monetization will look like.” – King Zuck
Part 3 is taking the “long approach to development” as it has already begun work on Llama models up to version 7. Part 4 is utilizing its world-class data set to train algorithms and part 5 is supporting a culture of rapid experimentation. Products coming from this philosophy in the near future will include Meta AI (its GenAI assistant which was just released into beta testing), Business AI products (for customer support), its Creator AI series to help anyone (like Snoop Dog and Tom Brady as early examples) create a personal AI and more.
This is the playbook for how Meta plans to succeed… and it mimics the playbook for how it has succeeded in the past.
Ads:
It continues to more effectively target ads for clients thanks to better consumer data leveraging. This will be a continuous process. Its Advantage+ suite (helps perfect campaign creation) is infusing more of Meta’s AI work into its processes to uplift campaign automation. It also just debuted Text Variations and Image Expansion, which are two GenAI products to further assist marketers in campaign optimization. Initial adoption of these tools was called “strong.” It’s also upgrading the conversions API (which allows for data sharing across Meta’s apps) to make data onboarding more intuitive. It’s in early beta testing on the WhatsApp AI tools for automated customer support interactions teased at the Meta Connect event. Finally, it told us that Meta Shops Ads (designed to pull consumers to your stores) crossed a $2 billion revenue run rate. This type of on-site conversion advertising, along with other business messaging products, make Meta even less reliant on 3rd party data access. That’s a big reason as to how it bounced back so quickly post Apple’s data sharing policy changes. That, and its massive supply of relevant 1st party data.
Metaverse:
Quest 3 is off to a “strong start” as it topped the app store rankings list for the holiday season.
Ray Ban will make more smart glasses due to “high demand.”
Reels, WhatsApp, Business Messaging & Threads:
WhatsApp continues to gain meaningful traction in the United States. This was once thought to be a pipe dream. WhatsApp’s newer “Channels” (one way messaging to large groups of people) product is growing rapidly. It has 500 million MAUs since launching last quarter. Business messaging from WhatsApp is becoming a material part of the business. It drove 82% Y/Y growth in FOA’s other revenue segment to reach $334 million.
Reels is now a revenue growth tailwind for Meta a quarter earlier than expected. It enjoys 3.5 billion re-shares per day. It will now work to “unify the recommendation systems across all types of video” to enhance content matching. Reels is enjoying 25% Y/Y growth in daily watch time across all apps as content ranking improves. Threads is “growing steadily” and now has 130 million MAUs vs. less than 100 million Q/Q.
Key Financial Metrics:
OpEx fell 8% Y/Y mainly due to lower restructuring charges and headcount reductions. Headcount fell 22% Y/Y, but rose 2% Q/Q as Meta resumes hiring.
Advertising revenue growth was 19% in UCAN, 23% in Asia Pacific (APAC), 33% in Europe and 32% in Rest of World (RoW). All regions accelerated vs. last quarter aside from Europe. Online commerce, entertainment, media and gaming where the sector standouts for ad demand. Ad pricing growth of 2% Y/Y was much better than -4% Y/Y growth expected. FX tailwinds helped, but “strong advertiser demand” did too. Notably, Chinese sellers drove strong demand growth in advertising and made up 10% of its total advertising revenue in 2023. China tends to be fine doing business with companies allowing their sellers to make more money. They’re less fine with companies trying to sell products directly to Chinese consumers. Meta doesn’t do that (Facebook is banned in China).
From a reporting standpoint, it won’t give us Facebook specific user growth anymore. Only overall DAU growth for its apps.
f. Take
This quarter was flawless. Elite team; elite product suite; elite execution; elite company. Elite. Enough said.
2. Amazon (AMZN) – Earnings Review
a. Demand
Beat revenue estimate by 2.2% & beat guide by 4.0%.
Met AWS revenue estimates.
Beat ad revenue estimates by 3.2%.
North American (UCAN) and International revenue were also both ahead of expectations.
b. Margins
Beat EBIT estimates by 26% and beat its EBIT guidance by 46%.
Beat AWS EBIT estimate by 7.8%.
Beat operating cash flow (OCF) estimates by 0.7%.
Beat $0.80 GAAP earnings per share (EPS) estimates by $0.20.
c. Balance Sheet
$86 billion in cash & equivalents.
$58 billion in debt.
Share count rose 1% Y/Y.
CFO Brian Olsavsky was asked about buybacks on the call. He flirted with the idea of entertaining them, but added that Amazon has many more investments to focus on in the years ahead.
Earnings season is in full swing and Borsa makes listening to an earnings call as easy as listening to a podcast. Dubbed “the Spotify for earnings calls,” Borsa is an incredibly simple tool that provides transcripts, presentation slides, and earnings releases.
Investors, analysts, and journalists have been using Borsa for over 6 years and it was featured live on CNBC with Kelly Evans. Join me in using Borsa this earnings season - you’ll be glad you did.
d. Guidance & Valuation
Missed Q1 revenue estimate by 0.9%.
Beat Q1 EBIT estimate by 11%. Like Meta and Google have already done, Amazon extended the useful life of some assets from 5 to 6 years. This lowered depreciation expense and helped EBIT. Without this help (which shouldn’t be excluded but I will for context) it beat EBIT estimates by 1%.
Amazon trades for 39x 2024 EBIT, 25x FCF and 45x 2024 EPS. EBIT is set to compound at a 3 year clip of 34%. FCF is set to compound at a 3 year clip of 30%. EPS is set to compound at a 3 year clip of roughly 34% as well.
e. Call & Release Highlights
Fulfillment Overhaul and Margin Expansion Work:
Amazon’s aforementioned regional fulfillment overhaul is going even better than expected. It's creating outperforming benefits in miles to fulfill and touches per fulfillment. This has been paramount in Amazon lowering per unit cost to serve in 2023 for the first time in 5 years. Specifically, it fell by a brisk $0.45 per package in the USA. This matters a lot when a revenue base is as massive as Amazon’s is. As an added positive byproduct, this is shrinking industry-leading delivery times further and making Amazon a larger player in daily essentials. It’s leaning into this reality as it added 65 new same day delivery sites Y/Y. If you can get me a toothbrush or an onion in an hour, I don’t need to go to CVS or Publix. While progress has been impressive, they’re not done… not even close. When pairing that with macro factors like labor supply recovering (lower wage inflation) and transportation disinflation, we’re left with Amazon’s exploding North American (UCAN) EBIT margin. Just 3 quarters ago, Jassy had to convince analysts that UCAN EBIT margin could get back to 4%. It’s now nearing 8% with more upside to come per the team.
“We can keep doing better on the cost to serve… the improvement in margins is not capped.” – CEO Andy Jassy
This overhaul has been the key source of marketplace and fulfillment margin expansion, but it has many more levers to pull. 3rd party sellers setting a new record of 61% of marketplace sales is margin accretive as Y/Y growth accelerated from 18% to 19% this quarter. Sponsored listing growth is margin accretive; better utilizing excess capacity via Supply Chain by Amazon (SCA) and Buy with Prime (BwP) is margin accretive; improving inbound inventory algorithms to better place goods is margin accretive; All of these additional tailwinds continue to play out. It doubled the size of its already massive logistics network in a very short period of time. Now, it’s reviewing this rapid expansion and, line by line, optimizing cost.
Buy with Prime added 24/7 live chat support, box free returns, and live order tracking (all managed by Amazon) for participating merchants. Buy with Prime is its white label fulfillment service allowing merchants to offer its world-class network from their own sites.
Debuted SCA in China for its large seller base there.
For years and years, Amazon famously punted profitability to future periods. Macro turned on a dime in 2022 to force companies to show near term profitability. Amazon pivoted admirably. This is the result.
AWS Demand Trends & Cloud Demand Appetite:
AWS growth accelerated a bit from last quarter. Encouragingly, leadership told us on the call that the acceleration would “continue into 2024.” Like last quarter, Jassy again told us that “cloud optimization trends continued to attenuate.” Less optimization means more new migration and workload demand and so accelerating growth… hence the expectations for 2024. It accelerated the pace of large deal signings, shrank the sales cycle and is renewing existing client contracts with healthy net revenue expansion. All of this is wonderful news for Amazon and for countless cloud players like Snowflake, CrowdStrike, Zscaler etc.
AWS added $1.1 billion in revenue sequentially, which is the largest gain for any of the three cloud providers. It’s not giving up market share to Azure or anyone else. Notable wins and customer expansions during the quarter included Amgen (expansion), Salesforce (expansion) Mitsubishi, Nvidia, LG, Hyundai, Merck and Byd.
Mitsubishi used Amazon QuickSight to drive tech stack maintenance costs down by a wild 70%. QuickSight is Amazon’s business intelligence tool to leverage 1st party data to guide decision making.
Amazon SageMaker is another important AWS tool that was highlighted this quarter. It’s Amazon’s fully managed service for building, training and deploying AI models. With it, Amazon reduced foundational model deployment costs by 50% and model latency by 20%.
New AWS European Sovereign Cloud to give public sector clients the ability to more seamlessly meet uniquely strict regulation.
AWS now has 105 availability zones and will soon add 12 more via recently announced investments in places like Mississippi.
To help AWS customers control cost, Amazon debuted new “Capacity Blocks” for ML. This allows customers to use exactly as many GPUs as they’ll need over shorter periods of time to bolster flexibility and to minimize waste.
GenAI:
I wanted to frame this section by reviewing the 3 layers of GenAI products and Amazon’s updates within each of them.
Layer one encompasses algorithm training and inference. It’s where Amazon’s internal chipsets and a partnership with Nvidia come into play. Graviton chip is its CPU product designed for basic computing. It debuted the 4th generation of this chip during the quarter which boasts 30% better performance than the predecessor. It also released the latest version of its Trainium chipset (intuitively named based on its role to train GenAI models). This quadruples machine learning training speed vs the old model. Airbnb and Hugging Face are among the users of these chips early on. It also expanded collaboration efforts with Nvidia to “deliver the most advanced infrastructure, services and software to power GenAI innovation.” It will be the first public cloud player to get its hands on Nvidia’s latest Grace Hopper Superchips. The two are also working together to design the “world’s fastest” supercomputer to be hosted by AWS. It will be a healthy mix of 1st and 3rd party chips powering Amazon’s GenAI ambitions.
Model creation as a service is the second layer where Amazon’s foundational Bedrock model is vital. Bedrock allows developers to build custom use cases on top of it while offering a plethora of 3rd party models such as Meta’s Llama 2. Per Jassy, GenAI is in a rapid experimentation phase at the moment. Considering how new this field is, developers are actively toying with different models to see which work the best for what they need. Bedrock’s large partner integration roster allows these developers to freely move between model templates in a safe, secure, fully-managed cloud environment. This quarter, AWS and Salesforce expanded their collaboration to add support for Bedrock in Salesforce’s ecosystem. Salesforce is adding more AWS products as part of the news.
All in all, Bedrock now has “thousands of customers” and is “deeply resonating” with customers. Per Jassy, clients want their GenAI work to be done in the same cloud environment they’re learned and have grown to trust over years and years. They want the cloud provider with the “best security track record.” That’s what Bedrock hosted by AWS provides. This is a large piece of Jassy seeing tens of billions in GenAI revenue in the coming years. That’s exciting to think about. Bedrock added new products like multi-step task agents and model fine-tuning during the quarter as it added Cohere’s and Anthropic’s latest models among others to bolster choice.
The third layer is the consumer app layer. This week, it debuted “Rufus” as its marketplace shopping companion. This allows shoppers to ask questions like “which football has the best grip” with detailed explanations taking you to the best option. It’s trained on Amazon’s 40% share of U.S. ecommerce and commanding global commerce share, which is something smaller competition just can’t possibly match. Codewhisperer, which is used by Adidas and BMW among many others, was among its first apps here which is a general coding companion for developers. More recently, it debuted Q. Q is another coding companion, but is trained specifically for AWS to “write, test and translate code” with access to AWS’s wide range of data repositories. Many more consumer GenAI apps are coming.
Ads:
Prime Video will debut advertising throughout 2024. This is giving Jassy “increasing conviction” that Prime Video can turn into a profit driver on its own. Thursday Night Football (TNF) will be a material piece of this. Amazon enjoyed a whopping 24% Y/Y increase in TNF viewers and is licking its chops and the potential to add more sports (like with its Diamond Sports investment during the quarter) to bolster Prime Video’s appeal and ad impressions. It’s fine tuning measurement, identification and targeting capabilities to capitalize on this momentum. It’s also using its work in GenAI to automate campaign creation (like Google and Meta). Ad revenue rose 26% Y/Y on an FX neutral basis as growth continues to accelerate on a larger and larger base. It’s the fastest rate of growth in over a year. There is much more ad load increasing to be done here.
Macro:
Consumer motivation to deal hunt remained very elevated this quarter with inflationary concerns still in place. It delivered a 70% Y/Y increase in cumulative savings during Black Friday - Cyber Monday to cater to these needs.
Other Bets:
Kuiper successfully launched two satellite prototypes (for broadband internet) and validated all systems. It will launch more throughout 2024 as part of ongoing beta testing.
Tweaks and consolidation within the grocery business is leading to a “nice profit trajectory.”
Added new 3rd party grocery delivery partners.
As previously covered, it partnered with One Medical to deepen virtual care services as part of its push into healthcare.
f. Take
This was a fantastic quarter. AWS is accelerating; ads are proliferating; fulfillment margin levers are all working; margins are exploding. There is nothing to pick at in this report. The thesis has come fully to fruition, and I have no plans to sell any shares following the after-hours pop. I expect sell-side profit estimates to keep soaring as cloud demand recovers and it proves to the world exactly how profitable fulfillment can be – especially with the international profit drag now gone. Wonderful quarter. Wonderful team. Wonderful company. Simple, wonderful.
Thank you for reading. If you found this valuable, please feel free to share it far and wide. That is how we grow. Have a great night!
Congratulations on Meta.
Well written, incredibly insightful, succinct yet detailed, and amazingly presented. Thanks Brad!