New Amazon Position

A brief overview of this investment case.

This is meant to be a broad overview of the Amazon investment case. This is not a deep dive and does not attempt to be nearly as thorough as one.

a) Introduction

Obvious secular tailwinds often make for great investments. These tailwinds, however, are never linear. They blow in fits and starts in a manner much less predictable than the compounded annual growth rate (CAGR) estimates would indicate. While these performance fluctuations can be anxiety-provoking, they’re also what foster opportunity.  When tailwinds temporarily fade is often when that opportunity becomes the most promising to secure compelling risk/reward and alpha.

Amazon has two structural tailwinds moving in its favor: Cloud computing and e-commerce. Interestingly, both are currently facing short-term, macro-related headwinds. The cloud computing industry is enduring added budget scrutiny, workload optimization, sales cycle elongation and so revenue growth slowing. E-commerce is dealing with a post-pandemic demand realignment, making Y/Y comps wildly difficult. To make things even more challenging, broad-based discretionary consumer spending continues to weaken amid a softening economy.

As the largest player in both spaces, Amazon is not at all immune. These obstacles will not go away overnight, but for this to work as an investment, they need to eventually fade. Fortunately, it’s likely that both will and that is the bet I am making by starting a position this week.

b) E-Commerce Titan

Amazon owns U.S. e-commerce. Depending on where you look, its market share sits between 30% and 40% while its growth rate consistently remains in excess of competitors’. As the clear leader in the area, it also prospers from a fragmented industry with smaller marketplaces everywhere from which to take share. Those potential share gains come for the players who can create the best dynamics for buyers and sellers. That’s what Amazon’s scale provides. Its leading SKU selection paired with consistent, lightning-fast fulfillment is a key recipe for attracting buyers. 3rd party sellers will always follow that demand to feed selection and traffic further.

Counterintuitively, the razor thin margin profile of this segment is one of the things we find most appealing. The gaudier the margins, the more competition a firm will invariably attract. That added competition requires a stronger competitive moat to maintain traction. Well? Amazon has made a super low margin business profitable thanks to incredible scale. That scale has created one of the most compelling and concrete competitive moats on the planet.  Emulating it requires tens of billions of dollars that nobody is willing to spend for the small EBIT prize it currently represents.

With this formula of two-sided loyalty firmly established, the company has earned the ability to create more margin. Just a bit of expansion with its massive revenue base would be quite needle moving. The question is where and how this can occur. A few areas look promising.

First, like all e-commerce players, Amazon got too excited in extrapolating pandemic trends as permanent. It built out huge excess capacity and hired tens of thousands of employees to meet its fulfillment growth expectations. It assumed the e-commerce pull forward would normalize at a lower rate of growth without giving back any gains. That didn’t happen as you can see below:

Brick and mortar shopping instead came back with more momentum than virtually anyone expected. That led to a lot of Amazon’s added capacity rapidly morphing into deadweight loss. And just like Meta’s year of efficiency, Amazon has taken the last few quarters to cut that capacity, pull back on expansion, right size its headcount and rationalize its cost base. This work is still ongoing with the firm’s North American marketplace EBIT margin at 1.2% last quarter vs. over 4% pre-pandemic. It sees a clear path to getting back to 4% which is also the main source of our bullishness. That 280 basis points of expansion alone would represent over $2 billion in incremental quarterly profits flowing through the income statement. The boost requires no investments and little risk; it merely requires operating more rationally.

That 4% margin, however, may have more upside. As part of Amazon’s ongoing fulfillment overhaul, it will take a more local, regional courier approach. The firm is reorganizing its national network into 8 regional networks all equipped with abundant SKU availability. This is set to cut miles to fulfill and costs while further shrinking delivery times. Win, win. As the team told us last call, it built out a last mile delivery infrastructure the size of UPS’s in just 24 months. Now, it’s time to extract as much value out of those assets as possible while e-commerce growth resumes and accelerates over time. In another move to extract more value from its courier footprint, it’s also allowing merchants to utilize the service in a white-labeled manner via its newer “Buy with Prime” product. That means more usage of its current fulfillment capacity which should help margins a bit too.

Beyond these initiatives, more 3^rd party selling on the marketplace is yet another margin tailwind to look forward to. This one is key. These 3^rd party sellers generate higher margin business for Amazon than its 1^st party business (more asset light service with incremental fees and no wholesaler take rates) and represented 59% of total unit sales last quarter vs. 55% Y/Y. That durable trend should continue and will inherently make the marketplace more profitable as 3^rd party sales slowly grow as a % of total. Most recently, this 3^rd party segment outpaced overall marketplace growth of 3% Y/Y by a full 1700 basis points.

As my colleague on Twitter (@SupBagholder) pointed out, this proliferation also directly supports its high margin advertising business. These 3^rd party sellers exist within a sea of merchants on Amazon’s site. They are all driven to stand out via sponsored product placements and preferred treatment in ways that Amazon 1^st party business does not incentivize. This is one of the many levers contributing to Amazon’s 23% Y/Y advertising revenue growth to buck trends of broader industry slowing. The combination of unparalleled traffic and embracing 3^rd party sellers is a powerful one to say the least.

c) Marketplace Peanut Butter to Prime Subscription Jelly

Amazon transformed and extended its fulfillment business and marketplace into a Prime subscription with nearly 200 million members by infusing more customer utility to fortify its edge.

Just like with Uber and Uber One for example, Amazon can afford to offer unique perks to subscribers. This results from Prime Subscription’s meaningful retention benefit and how much scale it features. Retention means small boosts to revenue. Scale allows the frequent retention-based gains to aggregate in a more meaningful fashion. That recipe coincides with slower fixed cost growth to power more appealing unit economics and generate positive returns. Cross-selling/bundling is always cheaper than winning a new customer. This all justifies Amazon’s heavy investment in Prime Video content, Gaming etc. which further motivates member loyalty. The more services it provides vs. the competition, the more bundling potential it has, and the more revenue it will generate thanks to less churn. Amazon is in pole position to offer the most valuable subscription in its space.

Over time, this subscription should consistently get more valuable as it adds new perks like discounted prescriptions and telehealth, grocery delivery and so much more. That’s how it raised the Prime subscription price from $119 to $139 annually this year with virtually no user push-back.

Like 3^rd party seller growth supports the ad business, more engaged customers lead to more traffic to further juice ad demand. Amazon Prime’s leading scale means it has more traffic and eyeballs shopping on its platform than anyone else… by a mile. Those eyeballs are ripe for ad-based monetization, and Amazon is taking advantage. The company continues to enhance ad load across its ecosystem. Between sponsored seller products and grocery delivery placements, there’s a long way to go to realize the earned benefits of being the biggest. Ad revenue is very high margin revenue as it requires minimal fixed cost to deploy. The impression real-estate is there… it’s just a matter of monetizing it.

d) Cloud Titan

AWS is the other vital growth driver. As a public cloud vendor, it offers Fortune 500 clients the compute and data capacity/storage needed to operate in today’s software and app-powered world. Customers pay for secured usage of this to take the headache out of maintaining complex infrastructure.

AWS is the largest in the space with Microsoft’s Azure quickly closing in. Both are dealing with macro-related challenges. While Amazon is clearly trying to do more with less this year, so are all of its AWS clients. All C-suites are laser-focused on becoming as lean as possible. This has meant added layers of deal approvals, elongated sales cycles and workload optimization -- leading to slowing usage growth. AWS is openly supporting customers in this optimization journey to align with their cloud evolution trajectories and focus on the long-term. Still for now, this led to its cloud growth slowing to 11% Y/Y for the month of April vs. about 16% for Q1.

That slowing was more noticeable than for Microsoft’s Azure… but there are two caveats. First, ChatCPT and OpenAI proliferation singularly feeds Azure cloud usage. The historical pace of that application's proliferation has slightly buffered the macro-obstacles for Azure. AWS doesn’t have this buffer. While one could argue that Google doesn’t either and GCP grew by over 30% Y/Y last quarter, the scale in which GCP operates is still peanuts vs. AWS. That makes differing growth rates somewhat inevitable. Secondly, and much more impactfully, Azure skews more to large enterprise clients vs. AWS’s small and medium business (SMB) leaning niche. AWS dominates with startups and smaller firms with less internal resources that are more fragile across macro cycles. Intuitively this makes sense as virtually all Fortune 500 enterprises are already using Microsoft’s productivity apps which paves the way for seamlessly cohesive Azure cross-selling.

This diverging AWS and Azure performance has led some to believe Amazon will not be able to compete as effectively in cloud as AI use cases become a larger piece of the industry’s growth. To that we say not so fast. While Microsoft has garnered all of the headline-based hype, Amazon is more than capable of being a prominent piece of this computing wave. In terms of AI capabilities, Amazon has been investing heavily into its training chip set (called Trainium) for years. It sees these investments as giving it the ability to train large language models (LLMs) more cheaply and with more scale than any other competitor on the market. I’m sure Azure, Nvidia and others would disagree, but the point is that AWS is very much so in this conversation and very much so trying to hold its own.

Beyond computer, inference and prediction is another massive piece of AI infrastructure. Like Tranium, Amazon has been hard at work on a 2nd chipset called Inferentia. CEO Andy Jassy thinks the “combination of price and performance” advantage it boasts is “significantly differentiated.” The vertical integration of these semiconductor-based capabilities meshes quite perfectly into cloud and AI computing. Owning these assets gives Amazon a great shot at becoming a default vendor in the space while its balance sheet ensures it can keep investing into the opportunity as it sees the need. That last point is quite important. Model building requires years and billions of dollars to effectively execute. Not many companies can afford to make these investments while deferring profits as Amazon has.

Related to these projects, Amazon will soon debut a generative AI tool called CodeWhisperer to automate code creation and expedite app building -- similarly to Microsoft Copilot. All of this utility will be fully open through Amazon’s foundational model called Bedrock. Bedrock will free developers to access the powerful proficiencies and build on top of them for more industry-specific use cases. Notably, as open-sourced AI models become better thanks to player like Meta, AWS should slightly benefit as smaller firms are more likely to tap into this tech. These open-sourced models are actually affordable for them to pull and customize unlike building complex language learning models from scratch. That utilization through Bedrock will mean more revenue within its SMB bread and butter.

And while these Amazon tools are being built for a more open, collaborative wave of computing, they’re being built to upgrade Amazon’s own processes as well. Among the first consumer facing use cases from its generative AI investments will be a near-future upgrade to its marketplace search. Like ChatGPT and Bard, this will feature a more conversational experience and more sophisticated product matching. Notably, more U.S. commerce search begins on Amazon than on Google. Amazon’s lagging search features had meant it was losing a piece of its traffic and demand to alternatives. This is now being addressed.

As an important aside, these investments should vastly enhance Amazon’s line of Echo hardware and Alexa voice control. The conversational style that this update will boast should push this segment closer to Amazon’s vision of giving everyone a personal assistant. Google’s Bard recently showcased an ability for its AI model to call and book appointments on a person’s behalf. It’s very easy to see this soon being a reality at Amazon based on the products now being rolled out. Amazon just released a new line of Echo speakers and it will be interesting to see how quickly these generative AI models are infused into the hardware.

It’s the concoction of its chips, developer tools and models that Amazon hopes to become standardized in the world of AI as use cases explode. This is why Andy Jassy, with his background in AWS and cloud, was named as Bezos’s replacement.

None of this is to say that Azure and GCP aren’t capable competitors -- they are. you can easily find brilliant industry experts (like we did) arguing in favor of any of these three vendors as being best in class. This is simply arguing that the slowly growing narrative around Azure eating AWS’s lunch is overblown. It reminds us of the Facebook/TikTok argument from 2021-2022 and the ChatGPT/Bard argument from this year. Cloud computing is a trillion+ dollar industry growing at a CAGR well over 10% with 90%+ of all workloads still being on-premise. The low hanging fruit remains abundant and there will be no “winner take all” outcome. Azure and AWS can co-exist, and both can keep attractively growing even if Azure continues to take share.

Macro headwinds holding back the sector are showing initial signs of abating. In our conversations with 3 different software engineers at 3 different Fortune 500 firms, all expressed recent investment re-acceleration in cloud infrastructure and migrations. Smaller cloud players like JFrog last quarter also called out easing usage optimization trends and accelerating demand. We will see if these green shoots have legs, but cloud’s re-acceleration will be a matter of when, not if.

e) Quantitative

Demand:

Amazon’s history and share lead allow us to extrapolate industry CAGRs for its long-term growth. Conservative cloud growth forecasts call for 15% compounded expansion for the next several years while e-commerce projections assume roughly 10% compounded growth.  Continued share gains in e-commerce are more dependable than in cloud computing with Azure and GCP being formidable alternatives. For this reason, we’re comfortable assuming AWS grows safely at 15% Y/Y for the next 3 years while the e-commerce segment grows by 10% over the next 3 years (call it 11% with modest sector outperformance). The blended growth rate would leave us with revenue CAGR of roughly 14% through the end of 2026. That growth will likely be closer to 10% for 2023 and accelerate in 2024 and 2025 to 15%+. Comps will get easier while macro headwinds fade…firms can’t just endlessly optimize cloud workloads forever. Eventually they’ll need to invest in usage expansion regardless of what macro looks like.

Margins:

We are confident in North American EBIT margin getting back to the 4.1% pre-pandemic mark just like leadership tells us to expect. We’re somewhat hopeful that the margin will expand more than that (for the reasons covered) but wanted to remain conservative in projections. The international segment’s run rate EBIT margin is the least certain of the 3. While we view breakeven EBIT as an overly negative assumption, it provides a margin of safety appropriate when modeling several years down the road. For AWS, we again pessimistically assume a halt to operating leverage for a 28.5% margin at maturity. That leaves us with a blended EBIT margin assumption of 7%.

From a free cash flow perspective, there is no reason to believe Amazon cannot get back to its 2019 level of 9% as it enters a more favorable part of its CapEx cycle. We were tempted to assume expansion from there but refrained from doing so. We again pessimistically (probably too pessimistically) assumed an 8% free cash flow margin for the end of the planning period.

Outlook:

When taking this all together, we are left with $60 billion in 2026 EBIT and $68 billion in 2026 free cash flow. As profits will be growing in excess of markets at that point, we are comfortable assuming a 25x EBIT and FCF multiple. From an EBIT point of view, that leaves us with a 2026 enterprise value of $1.5 trillion. In that scenario, we get a return CAGR of 9%. If instead using free cash flow, the return CAGR would be 14%. We feel this bakes in overly pessimistic assumptions and leaves material, compelling upside.

“We remain focused on building long-term sustainable growth in free cash flow, including our efforts towards a strong cash flow accretive working capital cycle.” -- CFO Brian Olsavsky April 2023

Balance Sheet:

• $54 billion in cash & equivalents with another $16.1 billion in marketable securities.

• Inventory levels are flat QoQ at roughly $34.4 billion.

• $67.2 billion in long term debt.

• It did not buy back any stock this past quarter vs. buying back $2.7 billion in the YoY period.

• Diluted share count has grown at a compounded rate of 0.7% over the last 3 years.

f) Risks

It’s a bit difficult to pick pressing risks to Amazon’s business model. Some of these are a bit of a reach.

Direct to Consumer (DTC) & Copycats:

The main risk in Amazon’s business model is its tendency to be possessive over merchant data and branding. This business model is why alternative approaches like Shopify’s have been so successful. Merchants ideally want full control over their brand, data and the customer relationship which is not something Amazon fully provides. Its “Amazon Basics” private label unit also often copies successful 3^rd party products to sell directly.  Sellers essentially bear the risk of finding out if the products work; then Amazon creates a knock-off which means less DTC revenue for merchants. Most sellers have to accept this concession due to the incredible value Amazon’s marketplace provides, but not all need to.

Allbirds and Nike both stopped selling through Amazon citing concerns stemming from these issues. Making this move requires significant brand power and scale which most sellers do not have. That’s why these examples remain rare… but something to keep an eye on. Today, in our profit-obsessed world, DTC and maximum revenue are inherently more attractive vs. an intermediary marketplace like Amazon. That could push more to leave the site and embrace more direct consumer selling.

Going Global:

Amazon is also taking its talents across the globe. In places like India it partners with Future Retail while self-running some of its own fulfillment capacity as well. It also very recently announced plans to invest $12.7 billion in cloud infrastructure there to fortify AWS’s presence.

That market, like for Apple, will become increasingly important over time. India is massive, is earlier on in its e-commerce evolution, is young and is wildly fragmented. Unsurprisingly, Europe and Latin America are also key focus areas. Amazon is opening operations in Belgium, Colombia and Chile in 2023 with plans to expand into South Africa as well. Germany today is probably its most important international market where Amazon.de is the share leader (more than double second place).

While international expansion will likely keep bearing fruit, that is less certain than its foothold into North American commerce. Success here is all but inevitable at this point. That’s not true abroad. MercadoLibre dominates in LatAM, Sea Limited and Coupang are entrenched in APAC and the complex Chinese market led to Amazon leaving that market in 2019.

Azure + Google Cloud:

Simply put, Azure is a juggernaut in cloud computing. It’s growing faster than AWS and quickly catching up to its revenue base. Microsoft is well funded, well run and talented. Google isn’t far behind. Both are strong competitors that will mean Amazon must continue aggressively investing in AWS to stay relevant. That could lead to a pause in EBIT margin expansion (assumed in our model). AWS EBIT powers the company’s overall profitability and must remain robust for this investment to work. Competition has a way of eroding margins… Especially mega cap tech competition.

Willing to Fail:

Amazon is famously known for a willingness to try and fail. It will bet on expansion projects featuring a compelling probability of success and a favorable risk/reward. While that mindset is loved when money is free and real rates are negative, it becomes less popular amid poor macro and hawkish monetary policy. Amazon being happy to loudly strike out is how it found AWS, how it found Alexa and how it blossomed into a trillion-dollar enterprise. Will it feel free to make some uncertain investments in today’s climate to preserve its long-term edge? Its current focus on margins says no, but its balance sheet allows it to walk and chew gum.

g) Plan

I started Amazon at 3.3% of total holdings. I have another 3.3% left of capital to allocate. I will review and assess over time.