I’ll finish reviewing the Oracle earnings call and the Uber conference for Saturday’s article. For now, here’s what Uber CEO Dara Khosrowshahi said about the consumer environment:

"We haven't honestly seen much consumer change & that means that the trends continue to be quite positive."

UBER CEO Dara Khosrowshahi

Last week, PayPal launched its “PayPal Everywhere” program. The initiative extends PayPal’s already compelling online rewards program to offline settings. Now, customers can select a category to receive 5% cashback and can stack these discounts with more savings from in-app promotions. PayPal has hundreds of merchants offering these deals including DoorDash, Instacart, Domino’s etc. More tools include an “auto-reload” minimum balance option to replenish when needed, a new PayPal Debit Card integration into the Apple Wallet and more in-person tap-to-pay functionality.

One may wonder: “How can PayPal profitably offer such deep cashback perks.” In practice, customers must be a PayPal debit card user to access the perks. That’s key. PayPal enjoys a powerful halo effect in which debit card users contribute materially more interactions, product uptake, engagement and profitable transaction revenue. The lifetime value of debit card users makes these hefty rewards more than worth it to pay. That’s the thought process. It’s a bet on the revamped rewards being compelling enough to spur more debit card usage.

And I’m quite confident in this bet paying off for a few reasons. First, the financial benefits highlighted above were already in place before extending rewards to brick-and-mortar. It’s very easy to see the incremental utility driving incremental volume and preference. Secondly, adoption of the card rose 30% Y/Y last quarter to already offer signs of things working. Finally, and most importantly, new CEO Alex Chriss has no interest in running cash-burning promotions to gain volume like the old team did. Considering that strict commitment and the other data cited, I find it highly likely that this 5% reward will work well from a unit economics standpoint.

PayPal Everywhere is meant to help the firm stand out in a hyper-competitive field and to drive offline shopping frequency. It wants to be more of an everyday staple for customers, and this is a big piece of that. PayPal Everywhere is a consumer’s tangible reason to use PayPal instead of anyone else. Early on, the top categories selected for cashback are groceries and gas, which points to its primary objectives being met thus far. 5 years ago (even 2 years ago), nobody would have considered using PayPal at a gas pump, which means this campaign is changing behaviors and PayPal’s entrenchment in daily lives. 

Chriss split the Venmo path into short-term and long-term plans. Short-term plans are mostly review. Chriss aims to greatly enhance the appeal of keeping funds in a venmo account. Most of those funds immediately leave the ecosystem, and that is a massive wasted opportunity for this firm. Fixing that issue would mean more net interest income, more interchange revenue and more compelling unit economics for the large platform. If that’s the effect, the causes will be more rapid Pay with Venmo adoption (30% Y/Y growth last quarter) and more thoughtful integration/display of its cards in the app. I love my Venmo card and I know many more people would if only they knew it existed. That’s mainly PayPal’s fault, as driving awareness has (for whatever reason) been deprioritized until recently.

Longer term, Chriss wants Venmo to do a better job leveraging its innate strength. Venmo is the only platform in the world that merges its consumer scale, financial use cases and socialness all into one app. That’s impactful. It’s one thing to see a merchant advertise some random product. It’s a whole different thing to see your friend buy a product with a merchant, and then see that advertisement displayed right next to the notification. This is where Venmo can shine and leaves so much potential room to create two-sided value. Venmo can let local merchants easily tap into payment feeds to pay for highly targeted sponsored listings. These placements will be informed by unparalleled customer financial data profiles, thanks to all of the payment and transaction data PayPal has on these users.  That reality will mean hyper targeted marketing spend, which is only ever paid when PayPal actually drives conversions. Merchants will be highly motivated to tap into the Venmo ad product suite as it builds, and consumers will be more than happy to accept unique perks stemming from the app’s powerful scale. 

For now, foundational work is being conducted to ensure Venmo has the quality tools needed to be a player in this form of advertising. PayPal has been hard at work on building out an ads platform and team, and the financial benefits are ahead of us. This can be a very high margin growth lever for the company, without needing any further customer growth to create it. These advertising impressions are just sitting there with fees waiting to be harvested and PayPal is finally building the systems to take advantage.

Chriss told us that the backdrop is shaping up about as expected this quarter. There was “nothing of note to call out” and no changes to current guidance. PayPal bears were out in full force for about 4 minutes during the event as the stock meaninglessly fell a bit and then recovered on these comments. Last quarter, Alex Chriss told us the quarter was trending ahead of expectations, so perhaps some wanted more commentary like that and algorithms did the rest. From my perspective, whether or not PayPal meets expectations or beats by a percent or two next quarter is not material to the overall investment case. And furthermore, I’d rather have Chriss save any positive surprise for the actual quarter to gear up for easier under-promise, over-deliver. As I frequently mention, that’s fixated on by a somewhat shallow, short-sighted Wall Street.

The shift to profitable Braintree growth is going well. Customer conversations on pricing to value remain in full force and will last for another few quarters. All of these conversations have been “very healthy” per Chriss. Braintree is determined to move on from its irrational pricing approach used to gain market share from 2020-2023. Now, it thinks best-in-class authorization rates afford it the luxury to command better pricing, along with software up-sells like Hyperwallet and its risk and foreign exchange services. Braintree also has another sales secret weapon in Fastlane guest checkout. It equips its go-to-market team with another wildly appealing tool in best-in-class guest conversion rates. More value means lower contract friction and larger deals.

As a reminder, Fastlane is a modern guest checkout. It can identify repeat shoppers at any onboarded Fastlane merchant (not only PayPal merchants… all partners too). Identification means customers skip manual, annoying card entry. That entry process routinely cuts shopper conversion rates down below 50%, while Fastlane is around 80% due to incremental convenience. Hard to overstate how game changing converting 30 more shoppers out of 100 is for any merchant. That’s life in the fast lane… a compelling new PayPal product and an elite Eagles song. This is where PayPal’s two-sided network shines, as it has the consumer scale to make this meaningful for PayPal and its partners. And? It has the merchant scale to make this work across much of the internet, with promotional and order tracking tools to augment consumer utility.

As Fastlane builds, PayPal can recognize more customers, the guest checkout edge can extend to a larger portion of the market and a network effect can grow. This is also why it’s so important that PayPal is signing new Fastlane distribution deals with Adyen and Fiserv. Competitors are acknowledging the superiority of this product and are thus motivated to offer it to its own clients. As this happens, integration will become less optional and more table stakes. All of this means more Braintree transaction revenue for merchants that aren’t even direct customers and a larger portion of customers who can enjoy this flow.

As an important aside, non-PayPal customers using Fastlane with other partners also means another touchpoint for PayPal to drive account growth. When it can create incremental delight via faster checkout and incorporate special merchant discounts (thanks to consumer scale), it’s likely that this touchpoint will be an impactful one. Early on, 40% of non-PayPal users are electing to have their data vaulted with the company for future checkout.

Shopify joins Fiserv and Adyen on the list of recently deepened PayPal partnerships. There’s nothing in the news about a Fastlane integration, but there are two exciting pieces of the development.

First, Braintree was named as a processor for some Shopify Payments in the USA. This news is far more encouraging than it would have been if announced last year. Why? Because again, Braintree isn’t undercutting the competition on price anymore. It’s competing on authorization rates, uptime, scalability and software. This hints at Braintree’s capability to win without accepting poor margins.

Secondly, PayPal wallet transaction data will be combined with Shopify payments for a broader view of business health and trends. Shared merchants (there is a lot of customer overlap) will now gain better interoperability and data from two of the leading commerce platforms on the planet. That should augment the value propositions and usability of both product suites vs. the rest of the competition.

There’s still a ton left to do on cost cutting via more automation and GenAI usage. Early innings here.

Guidance & Valuation:

Oracle trades for 23x forward EPS. EPS is expected to grow by 13% this year and by 15% next year.

The new AWS CEO has an interesting background that I don’t think is broadly known. He’s been at the job for just a few months, but with AWS for 18 years. Fascinatingly, Garman was actually Andy Jassy’s summer intern when he was a college student. During his internship, he worked directly with Jassy on a top secret start-up within the company. That startup is now AWS and its $100 billion+ revenue run rate. He was instrumental in building out early services like S3 and EC2, and led several different AWS departments before taking over this year. 

From here, the opportunity for AWS remains massive in Garman’s mind. That’s how the company is guiding to accelerating AWS growth (reiterated during the interview) despite the massive base of business. Optimism stems from many of the product areas we will focus on through the rest of this section. Perhaps the most powerful tailwind, however, is lower than 20% of compute workloads run on clouds. That remaining 80% is on-premise, with most of that majority able to pocket real cost, efficiency and pace of innovation gains by letting a hyperscaler like AWS run their infrastructure instead. A cost edge is always popular, but efficiency and innovation edges are especially popular in the age of GenAI and rapid change. That rapid change is hard to embrace if you’re running on static, antiquated, finite on-premise infrastructure. Bottlenecks under this approach are inevitable. This is one of many reasons why cloud optimization demand (cutting usage) from clients is easing and new workload demand is ramping.

Garman thinks AWS does a better job listening to its customers and building what they actually want. While this seems easy to do, he doesn’t think any of the competition rivals Amazon in this regard. 

That meshes extremely well with world-class security and a philosophy that has embraced freedom of database, hardware and other vendor choice. This freedom of choices is the peanut butter to AWS’s massive partner integration roster jelly. Developers can build however they want, with the certainty that their work is secure, fully managed and scalable. In summary, Garman sees unique competitive value in its unmatched systems and infrastructure pairing beautifully with a willingness to build the solutions customers want.

Throughout this portion of the chat, he threw two subtle jabs at Azure. First, he spoke on Amazon being security first from the beginning. Not “bolting on” products in response to observed shortcoming. Secondly, Garman seemed to criticize the vendor lock and lack of vendor choice that Azure creates. This lack of choice is increasingly unpopular, per Garman. His opinion is certainly subjective and biased, but Azure’s new data product (Fabric) is newly open-sourced, which does point to him being correct.

Here, Garman took us through the various layers of cloud computing and how AWS stands out in each. AWS has been investing in foundational hardware to support cloud infrastructure since inception. Between chips, networks, data centers, power infrastructure and more, it thinks it has invested appropriately in the “base cloud layer.” That foundation is what enables it to innovate rapidly for its customers at “lower cost and better performance than anyone else.” Sticking first with chips, its work here started a full 10 years ago. It has not been scrambling to catch up like many social media pundits seem to think. The latest version of its general compute central processing units (CPUs) called Graviton4 outperforms all X86 processors by an average of 20% and with nearly 50% price performance gains thanks to better efficiency. Amazon’s ability to “control the whole process” or use its own chips in its own data centers routinely leads to deeper customer optimizations and outcomes.

A hypervisor is a tool unlocking the ability to run many virtual machines on a single server. Amazon has been offering these for 15 years, so “no operator can access compute instances on AWS” for customers needing that option. This vertical integration, it thinks, offers a superior security posture to others on the market.

Amazon also continues to invest heavily in high-performance compute (HPC) hardware to spur the GenAI boom. It has been making AI processors for 5 years, starting with Inferentia. The latest Inferentia chip (for AI inference) was used in Alexa to cut inference costs by 70%. Garman was understandably asked how he thinks about competing with Nvidia/AMD/Intel on chips while also partnering with them. The answer goes back to Amazon’s perpetual dedication to customer choice. It does not want to force you to use anything. It wants to make great products, and accept that those great products will not always be the perfect fit. It then wants to make sure customers have whatever that perfect fit is within the AWS environment. Sometimes it’s Graviton… sometimes it’s not. Today, Amazon is adamant that it’s the “best place to run Intel, AMD and Nvidia processors” (as well as its own). They seem to be executing here, as Nvidia gears up to use AWS infrastructure to run their servers.

“We think more processor choice is better for customers.”

Matt Garman

All of the aforementioned work Amazon has done to internalize development of more pieces of cloud infrastructure will bear fruit for years to come. Chips and hypervisors are good examples, but networking equipment is too. If you follow GenAI hardware, you’ll notice that Ethernet connectivity is becoming wildly popular again. This technology is generally better for connecting massive compute clusters and “often outperforms” Nvidia’s InfiniBand for price, uptime and performance. InfiniBand dominates AWS for smaller cluster performance. Amazon knew the Ethernet connectivity wave was coming, and has been building for the moment for over a decade. All of these investments, to Garman, mean lower reliance on 3rd party vendors. In turn, that means relatively lower CapEx intensity as well as vertical integration driving better performance.

Garman sees most of the AI chatbots that have gained popularity as “scratching the surface” on GenAI utility. Considering its main rival (Azure) is ahead in chatbot monetization, this opinion does make sense. Still, I think it’s entirely valid. Most of the apps gaining popularity thus far have been for driving more efficient processes for traditional work. Garman sees this going so much farther. He sees drug discovery companies 100,000Xing pace of protein discovery; he sees Japan’s new Bullet Train using SageMaker to create custom models for predicting component failures. He sees GenAI creating better decision makers and outcomes and thinks the Microsoft Copilots of the world are only the beginning.

SoFi is testing a product called “cash coach.” This organizes liquidity across all accounts to nudge customers to optimize fund placements. For example, if you have a savings balance but are paying credit card interest, it will tell you to pay off the bill to avoid more interest expense. If a customer has money in a checking account, it will suggest moving funds to a higher yielding savings account or to SoFi Invest. This should drive more cross-selling and growth, considering many of the nudges will include products that a SoFi user isn’t yet interacting with.

Next, SoFi will debut “SoFi Plus” later this year. This will offer its 4.5% savings yield to non-direct deposit customers and will offer more rewards, loan discounts, SoFi’s team of financial planners, more early IPO access (as that market reopens), more diverse assets to invest in etc. Noto made it a point to say the product comes with “no asterisk.” That’s him throwing shade at other players that offer somewhat similar subscriptions with several more access caveats. These two developments have one thing in common: Both rely on a fully product suite to actually connect recommendations to actionable tasks. It’s one thing to say “You should do this.” It’s another thing to say “You should do this and if you want us to we can do it for you.” It’s also way easier to provide more subscription utility when you have a larger, compelling product suite.

Financial services is now firmly contribution profit positive, despite investing in sub-scale new products. SoFi Invest & Credit Cards are the two, currently cash-burning examples. SoFi’s products are always unprofitable at the beginning. It takes its time to build out the scale needed to cover natural fixed costs. It uses positive variable (contribution) profit as permission to lean more heavily into growth while using net profit as the next green light to get even more aggressive.

This approach is always preferred to me. It’s wise to go slowly in driving product market fit, rather than guessing and hoping that fit has been secured. I think that’s especially true for the credit card business. Losses here can pile up in a hurry and it would be irresponsible to grow these originations amid such an uncertain backdrop before it’s fully confident in underwriting algorithms. That confidence is only now approaching acceptable levels, which means growth here should begin to ramp (as the team has told us to expect).

For some evidence of both of these products being future profit drivers, consider the following thoughts: First, SoFi Invest (which is slightly more mature than credit cards) is approaching a sub two-year payback period. This milestone is despite SoFi Invest being 50% under-monetized via SoFi not yet having the full product suite in place. Credit cards are following in its footsteps, and both products compellingly resemble the promising profit trajectory its SoFi Money product has enjoyed. Give them time.

As an important aside, SoFi is also not using Invest or Credit Cards as member lead generators yet. It’s solely focused on cross-selling. That’s partially due to needing more scale, but also because it has such better data on its own existing customers for better credit card underwriting.

Noto reiterated SoFi’s 7%-8% life of loan loss rate target for the millionth time. SoFi offered some highly valuable cohort analysis last quarter that Noto again cited during this interview. I’ve included my coverage of that from the last earnings report. If you read the review, this next paragraph will sound familiar, but is highly important to know for newer readers.

It took us back to 2017, as that was the last year its loss rates approached 8%. Its 2020 - Q1 2024 vintages at the same amortization rate boast significantly better loss rates vs. 2017. Not just this, but newer vintages are now showing a concrete pattern of improving credit health. The 2020-2024 vintages overall are 56% of the way through repayment and have an average loss rate of 3%. To breach its 8% target, that loss rate for the remaining 44% of principal would need to be 11%. SoFi has never seen loss rates remotely close to that level. Even in Q4 2022, as it was tightening credit parameters and dealing with riskier borrowers than it wanted, the charge-off rate of 5.02% compares to 6.07% in 2017. 

It’s hard to overemphasize how positive rate cuts would be for this bank. That may sound weird, considering cuts weigh on net interest margin, which drives bank financials. But SoFI isn’t a normal bank. The fixed cost advantages inherent in its business help embolden it to cut account yields as those rates fall to help offset the net interest income hit. Furthermore, it’s rapidly replacing higher cost debt with cheaper credit and lower cost deposits too. That will continue and should help it maintain a 5%+ NIM (as guided to).

There are also several more positives that would stem from rate cuts. Dovish policy would directly prop up mortgage and student loan refi demand. SoFi is “willing to originate as much of this credit” as there is acceptable borrower demand. Both are intuitively considered lower risk than unsecured personal loans and both would thrive in a falling rate environment. For example, during the August 5th Japan Carry Trade drama, SoFi saw variable rate home equity lines of credit (HELOCs) applications 2x. That coincided with just a 25 bps hit to mortgage rates, showing you how eager borrowers are to take advantage of future cuts.

Even on the personal loan side, despite cuts hurting variable rate fixing demand, SoFI has massive pent up personal borrower demand. It also has strong demand from capital market buyers, which would merely grow as those participants also get access to cheaper funding. That way, SoFi can grow personal originations without growing balance sheet risk.

On the tech platform side, enterprises gaining cheaper debt will accelerate demand for its products. Faster velocity of money will prop up usage of its payment processing arm too. And for financial services, that same velocity of money benefit will boost every single product in that bucket. Cuts are coming… and SoFI is excited.

SoFi lowered its savings yield by 10 basis points (bps) to test the waters and respond to deeper cuts from competition. It saw zero blowback and zero net deposit growth impact from this move. That offers some evidence of SoFi’s value being far from solely in a higher APY. Maybe its financial planners, cheaper loans, unique public investment access and budgeting tools matter just a little bit after all.

Still, SoFi expects to remain near the top of the pack in APY. It fully plans to lower its rate more slowly than others. How? It has no branches, owns its own tech stack and utilizes fewer 3rd party vendors than others. That separates its cost structure from other disruptors. But it also has its banking charter, which means access to lower cost deposit funding and a freedom from sponsor banks forcing it to set certain yields. It has the freedom to set its own and the business model to make sure its rate is higher. Simple enough.

Rate hikes have led to elongated sales cycles, delays to launching new contracts and slower than expected tech platform growth. We’ve seen this across countless B2B fintech vendors. Some customers are postponing start dates and prospective clients are awaiting monetary certainty to set new budgets. No large deals have been lost, but many prospects remain in wait and see mode. Still, SoFi is waiting on a sizable backlog of clients to go like next year and Noto is adamant that a few cuts in 2024 would position Galileo and Technisys for a strong 2025. It has the multi-core banking tech needed (with things like real-time asset/liability management to avoid another Silicon Valley Bank fiasco) to thrive in an easier environment.

SoFi uses the tech platform for buy now, pay later and will soon use it to work with commercial partners as a sponsor bank by the end of the year. Next year, SoFi money will be finished migrating to this platform. This update depicts how long these things take in the highly regulated banking sector and why I don’t think investors should start assuming large deals aren’t coming. More patience required than I think pretty much all of us initially thought.