What is business moat
Moat can be defined as the time and energy needed to copy a business.
A business is often more than just its user-facing end product, but also the brand, unique distribution channels, existing enterprise contracts, legal contracts, patents and operational processes behind it.
The time and energy needed to rebuild the whole Amazon logistics ecosystem and global distribution by building a strong brand is way bigger than rebuilding an app like Todoist.
The moat of companies is dynamic and changes over time with new and improved tools allowing us to get things done more efficiently.
For example, large language models made it way easier to produce code; therefore, the time and energy needed to replicate software is drastically lower than it used to be before LLMs.
- If it would be easier to create strong brands on the internet, your brand moat would be weaker.
- If patent laws change, your IP moat would be weaker.
- If it would be easier to create logistics centers, your physical infrastructure moat would be weaker.
- ...
Also the overall moat of companies is often a combination of various sub-moats (Legal Contracts, Team, IP, Brand...).
Different types of moat
These are just high-level mental models. Don't think of each type of moat as actually being something different from other types of moat - it's all connected.
If you have an interconnected ecosystem, this mostly also leads to a corporate conglomerate moat. If you have a successful marketplace, this often leads to a brand moat, and vice versa...
Interconnected Ecosystem Moat
You could think that Meta, for example, has an interconnected ecosystem moat like Apple does because they own a number of social platforms like Instagram, WhatsApp, Threads, and Facebook.
But unlike Apple, Meta does not connect these different puzzle pieces in a way that makes the end-user experience better.
The user experience of WhatsApp or Instagram didn't become better after Meta's acquisition.
But the user experience of Shazam got better after Apple's acquisition because of its native integration into the operating system.
The interconnected ecosystem of Apple is insane:
First of all, operating systems like iOS or macOS themselves are interconnected ecosystems:
The work Apple did in building a great file manager, application launcher, system settings, system APIs, native development environments, native apps, etc., made the overall user experience better.
Additionally, Apple has an ecosystem of hardware products. If you own an iPhone, Apple Watch, Mac, and Apple TV, the overall experience is better than if you were to use fragmented hardware products from different vendors:
The iPhone leads to Watch, on Watch you benefit from a native Apple Wallet integration, shared notifications, Apple Health, Apple Homekit (there are hundreds of useful interconnections like this in Apple's ecosystem).
The important part here is that most of the subparts of Apple's ecosystem make the end user product experience better, you are not experiencing one single product but a whole ecosystem.
Tesla
Tesla’s moat builds up on a similar effect, as a Tesla customer you don't just experience solely the car but the whole ecosystem of supercharger availability, the Tesla wall connector for your home, the mobile app, and self-driving.
Amazon
Amazon's investment in logistics centers, their own driver network, and Amazon Lockers across cities has improved the overall shopping experience. Additionally, using Amazon as my primary online shopping platform simplifies order management within a single app.
Microsoft
Microsoft bundles VSCode, GitHub, and GitHub Copilot more and more into one ecosystem.
Windows bundles Copilot with Office.
Teams, Outlook, and Loop. The MS Loop vs. Notion distribution chart will look the same as the Teams vs. Slack chart in a few years, even with Notion and Slack being way better products than the MS counterparts.
Adobe
Adobe bundles the whole visual creation process into one ecosystem: vector illustrations, pixel editing, photo editing, video editing, UX/UI editing, video creation, photo creation - creating both interconnected ecosystem and habit moat.
Interconnected ecosystem moat is one of the strongest moats a business can have - because to offer an experience as good as the one of the current ecosystem you would need to rebuild the whole ecosystem.
Corporate conglomerate moat
I already mentioned that there are also a lot of businesses which bundle products in one organization that are not truly connected.
This can still make sense from an operational efficiency perspective or from a product distribution perspective.
If you're Meta and already running multiple social networks, acquiring another one could be strategically smart because you can use similar tech stacks, operational procedures and distribution strategies for all of them.
Same applies for companies like Unilever, Procter & Gamble, Nestle, LVHM, Walt Disney, Volkswagen which bundle multiple similar sub-products like in one organization because the operational and distribution processes of each of the single products overlap a lot.
- VW using the same core platform, interior modules across different brands
- P & G, Unilever, Nestle using similar manufacturing and logistics processes for similar products.
- Amazon using the same logistics platform no matter what you order makes it the long-term best e-commerce experience across every vertical.
- Walt Disney using the same production studio across different projects
- Bosch/Siemens have a lot of technical overlap in engineering across automotive, medical, and infrastructure.
- LVMH using the same influencer network for different brands
Corporate conglomerates often lack strong interconnections between each of the products in the end-user experience.
Utilizing multiple products from the Unilever conglomerate simultaneously in my household does not enhance the overall end-user experience of a single one. Purchasing each one from various companies would result in the same user experience.
Therefore conglomerate moat is not nearly as strong as interconnected ecosystem moat because to copy one of the sub-products, you don't need to replicate the whole interconnected ecosystem to give the user the same experience. For Tesla to outperform the Volkswagen conglomerate, they didn't need to rebuild Porsche, Audi, Seat, Skoda, etc. They just needed to be better than the best carmaker of the VW conglomerate.
Proprietary dataflywheel moat
A lot of products, especially in the age of AI, get better the more data you can gather. Google's search engine improves every second by Google measuring what the world currently searches, and what the world searched over the last years.
The ability alone to gather a lot of data does not necessarily give you a data flywheel.
You need a to use the data continuously to improve the product.
- Every Tesla improving the self driving algorithm
- RLHF in language models
- Google Maps route planning intelligence
- Google search getting better with each search
Also the timelessness of the data you gather matters. If your product is great because it utilizes a lot of historical data + current data, the moat is stronger.
If the historical data matters less and the latest gathered data is the key ingredient.
Marketplace moat
Similar to the data flywheel, a marketplace improves with every user joining it. The moat of marketplaces is the users and creators who actively use it.
There are different types of marketplaces:
- Services: Doctolib, Hostelworld, AirBnB
- E-commerce: Amazon Marketplace, eBay
- Stock exchanges: NASDAQ / NYSE
- Social networks: WhatsApp, Bumble, LinkedIn
- Social Media: Instagram, Pinterest, Twitter, Wikipedia
If you want to create new marketplaces, you either need to enable a completely new kind of supply or compete with existing marketplaces by offering higher quality in lower quantity.
- Hinge having fewer users than Tinder or Bumble - but you can argue the quality is higher
- Cosmos having less content than Pinterest but higher quality.
- NFT marketplaces enabling a new kind of supply of goods.
Physical Logistics Moat
- Tesla’s Supercharger network
- Amazon’s logistics centers
- Walmart, Lidl, Aldi, Rewe
- Airlines
- Containership Logistics
- Next-gen preventive health clinics like Neko Health, Aware Health...
R&D / IP Moat
- Qualcomm owns the most important patents for wireless broadband communication and has the most skilled engineers and scientists in the world to both engineer and patent new wireless communication technologies before anyone else.
- Apple is doing the same for most human and computer interaction hardware like eye-tracking, touch interfaces.
- Nvidia IP
- Siemens Medical Devices
- BionTech mRNA development
- OpenAI having the best AI researchers in the world and being the first ones to create technological breakthroughs (event though they are currently lacking the distribution platform to make use of these early breakthroughs and build other moats on top of the timing)
Engineering talent moat often leads to IP moat and vice versa.
Legal Moat / Partnership Moat
If you are building a product in a regulated space or one that requires strong partnerships for it to function, securing these partnerships (ideally with a winner-takes-all arrangement like sports broadcast licenses that often go to a single service per country, or telecommunications partnerships like with the 5G network) can strengthen the moat of your business.
- Sony Music and Universal Music have deals with almost every big musician around the world.
- Deel figured out how to legally employ remote workers in nearly every country
- Airbnb's legal maneuvering with cities worldwide
- Same with Bolt and Uber
- Urban sports club partnering with lots of gyms and local sports facilities
- DAZN's partnership with the UEFA Champions League
Headstart moat through right timing
It is evident that timing plays a crucial role in building an empire. To be more specific, timing matters significantly in the following ways:
Interconnected ecosystem moats are almost always built over time. By being the first company to create or distribute a product, you can then leverage your head start to continually expand your ecosystem.
This principle applies to various types of moats: interconnected ecosystem moats, IP moats, team moats, physical logistics moats, interface-as-habit moats, and marketplace moats. A lot of them grow stronger over time.
If you utilize your head start effectively and do not rest on your laurels, you gain more time than other companies. For instance, convincing local politicians to approve a second supercharger network is challenging if Tesla has already established one. Similarly, attracting talented engineers to join your company over OpenAI is difficult if OpenAI has a significant head start in building foundational models, acquiring customers, partnerships, and funding.
This dynamic underscores the value of startups. If your product bet or hypothesis proves correct, the risk of innovation can result in a significant competitive moat.
Your headstart as a moat can fade away when the technologies you built on become outdated.
We saw this with Tesla replacing the foundational principles German car manufacturers relied on. They built around internal combustion engines and formed the best manufacturing and engineering teams in the world. However, Tesla created better technology to solve the same problem, thereby rendering the German car manufacturers' foundational principles obsolete.
Creating your own distribution rules - by owning a platform
If you are the creator of a platform, you can create your own rules.
Even though the internet was conceived as a federated standard not owned by anyone, Google is creating the core products that most people use to use core parts of the internet.
Chrome Browser, Google Search Engine, Google Sub search engines (maps, images, shopping, flights, finance, books...) etc.
Google owns the search engine market. They can create their own rules and, for example, distribute their own products at the top of their search results or even integrate them natively as a widget.
Same with Amazon listing Amazon basic products at the top of search results etc.
Brand Moat
Elite universities are prime examples of brand moat.
Every great person enrolling at your university each year strengthens the brand, as students often consider the success of previous graduates when choosing a university.
Capital moat
Having access to large amounts of money can be a big advantage, especially in fields where big investments can take a business to the next level.
We saw this in the early days of the cloud industry. Companies like Microsoft, Google, and Amazon used their access to money by investing huge amounts into cloud infrastructure. No other company could do this on the same scale, which made their ecosystems even stronger.
The same thing is happening with Tesla. The company can invest more and more in building gigafactories around the world, increasing their production in ways that companies with less access to money simply can’t.
Switching cost moat
If the users have high rates of switching away from your product, you have a problem. Switching to a different social media platform when you already have a great audience on one of them does not make much sense. Switching from Adobe Photoshop to a competitor does not make much sense when you and your team have already learned Photoshop over the years and are not very familiar with it.
Language models in their current form have a small moat because of low switching costs:
Language models today don't have nearly the same moat as operating systems because they don't rely on proprietary standards. They function as single endpoints with low switching costs to switch to another model (taking about 5 minutes) and use a universally recognized standard: natural language.
This demonstrates that the time and energy needed to build a technology does not directly correlate with the strength of its moat. Other AI labs can train and distribute similar models; thus, the moat of a company like OpenAI is not in people building on top of them, but in having a head start in assembling the most talented AI research team in the world, raising significant funds, and building infrastructure for model training—which is harder to replace. Therefore, always having the best model in the world at the best price point will be the moat, rather than the fact that developers are building on top of proprietary standards.
If OpenAI uses their advanced models to improve subsequent model developments, this moat could become extremely strong. Otherwise, it remains a challenging moat to defend, especially compared to the robustness of an interconnected ecosystem moat like Apple's.
Habit as moat
Creating a habit moat is one way to increase switching costs.
Being an "industry-standard" to solve a specific task often means nothing more than being the solution that most people have trained their brains on.
Whether it’s the programming language you learned, the operating system interface you’re familiar with, Excel macros ingrained in your workflow, the medical equipment you trained on, the CAD program from university...
In a world of human knowledge workers, this moat is incredibly strong. However, as AGI starts replacing more human knowledge work, this type of moat may become less critical because AGI can quickly learn any novel solution in seconds and does not need months and years to learn it.
Also one might argue that companies could simply copy a competitor’s core interactions to minimize changes in users’ muscle memory. However, for highly complex products like SolidWorks, replicating every single interaction isn’t feasible. If you were to build a new SolidWorks, you’d likely start from first principles and current technology, resulting in different interaction paradigms.
Collaboration standard moat
Even as interface habits become less important due to the increasing role of generative AI, information standards will remain crucial for reliable communication and data transfer between multiple AIs. AIs will still produce prototypes in standardized tools and share them with other AIs.
Providers will continue to facilitate code creation based on these prototypes. Additionally, there will be more data for AIs to create Figma-to-code workflows than for any other tool, leading to better performance. We already see this with AI generating superior code in JavaScript and Python compared to less widely used programming languages and frameworks.
Products that leverage habit as a moat often also rely on being a collaboration standard. For example, when two people want to exchange money, both using PayPal simplifies the process. The same applies to using Figma, Google Sheets/Excel, and other collaborative tools. The world collectively builds habits around these standards, making it challenging for new products to replace established ones.
Examples:
- US Dollar: The collaboration standard of the Dollar is a moat of the United States (Ray Dalio has great thoughts on this on how the ruling nation controls the world's reserve currency).
- BTC / ETH: as crypto standards
- Money Transfers: PayPal
- Stock Exchanges: NASDAQ, NYSE
- IDE and extensions for IDE's: VSCode
- Browser Extensions: Chrome
- Code Sharing and Collaboration: GitHub
- Calendar: Google Calendar
- Spreadsheets: Excel
- Multiplayer Games: CS:GO, LoL, Dota, FIFA...
- Communication: iMessage, WhatsApp, Email...
- Social Profiles: Instagram, Twitter, LinkedIn
- Open Source Developer Frameworks: Flutter, NextJS, SvelteKit (code is open source, but community does not really have any control)
Operating systems: The ultimate "featureizers" of personal computing
Many technical-founders solely focus on building a product but not on building an empire around the product.
That's fine if you:
- Don’t mind being “featureized” or acquired by other companies.
- Are fine with working on niche problems with very small markets which are not big enough to acquire or copy by bigger players (also called indie-hacking).
What Does It Mean for a product to get “featureized”?
Being "featureized" means that your whole product becomes just one of many features of another meta-product.
For example:
- Calendly's meeting scheduling was just a missing feature in Google Calendar.
- Chatting with a PDF is just a feature of ChatGPT not a separate product.
- My product Fixkey, along with Grammarly and Raycast AI, turned out to be native features of macOS.
While I was never really bottlenecked in providing features for Fixkey on macOS due to deep system APIs, building Fixkey for iOS would have been impossible the way Apple did it.
Apple's iOS does not offer many ways to integrate features deeply into their operating system, so you will be limited in your capability as a developer to provide the best experience.
For apps like Salesforce, SAP, or Figma, this is not the case because the data they handle is not, or only slightly, connected with the rest of my personal operating system data.
Given the importance of operating systems for personal computing but the increasing restrictions on customization for app developers, the Unix idea of a community-owned operating system might become more appealing in the future. Especially when going towards cloud computers which are not running on proprietary hardware anymore (personal cloud computing mostly applies to desktops though, which is a decreasing market).
Nevertheless, with LLMs and future AI systems making it easy to copy software, this is not only beneficial for proprietary vendors but also open source community-driven projects which could theoretically recreate proprietary operating systems.
Personal standalone Apps - a common victim of operating system integration
I cannot think of any business-empire which was built by creating a “personal standalone app”.
Let's define them:
For personal standalone apps, the core product is the app itself, not a physical service or proprietary information accessed through the app.
So you are not using the app to:
- Access physical services: Amazon, Airbnb, Uber, Uber Eats, DoorDash, Bolt, Booking...
- Access licensed media: Amazon Prime, New York Times, Netflix, Brilliant, Duolingo, Kindle, Audible, Spotify, DAZN...
- Access proprietary intelligence: ChatGPT, Midjourney, Elevenlabs, Luma AI...
- Access social networks or marketplaces: TikTok, eBay, Airbnb, Instagram, WhatsApp...
Excluding all of these what's left are Note-taking, Calendar, Contacts, Personal Memory Management (Rewind), Meeting Recording, personal CRM's, Browsers, Grammar Correction / Text Replacement, Bookmark Manager, File manager, and Mail apps.
These apps are solely handling personal data like your todos, notes, calendar entries, contacts, mails, bookmarks - and if you wouldn't mind syncing between devices, most of them could run fully locally.
If you are a startup building one of these kinds of apps where the main value you provide is the user experience itself, not something behind the app.
If your app is useful enough for a general audience, you can be sure it will eventually be rebuilt or integrated natively by the operating system. This is because:
- It’s relatively easy for companies like Apple and Google to recreate these standalone apps, as they don’t require intensive operations (booking partnering with hotel chains or Netflix creating movies and acquiring licenses).
- Native integration almost always results in a more seamless and superior user experience.
- Built-in solutions are naturally marketed and distributed to every user, unlike custom apps that require active installation.
Examples:
- Apple integrated features from startups like Multi-On, Nox, Adept through app intents, and the ferret-model.
- Microsoft’s Internet Explorer effectively killed the Netscape Browser.
- Apple’s Photos app diminished the need for personal cloud storage providers on iOS.
- Google Chrome / Android Password offering a more integrated experience compared to Bitwarden and 1Password.
- Google Drive/Photos serve similar functions by integrating personal storage solutions.
- Apple Pay provides a more elegant system than Cash App and Venmo through superior OS integration.
- Apple's integration of contact sharing through RFID displaced many startups offering similar solutions.
- Native integration of Google Search / Google Assistant into Android. Native integration of Siri into iOS / macOS / watchOS...
- Whoop features will continue to be integrated into Apple Health and Apple Watch.
I think of the Operating System as the ultimate bundler of useful startups resulting in the closest you can be to an “everything company,” a “roman empire of technology", especially because I cannot think of a future where personal computers will not be one of the core products shaping the human experience.
Future scenarios where the operating systems companies might lose their dominance are:
- Open-source community approaches to building operating systems offer faster and higher quality innovation than proprietary solutions, as is already the case for dev tools
- Human-computer interfacing changing in drastic ways which needs from scratch rebuilds of core parts of the OS - together with current OS vendors sleeping on them and not integrating them early enough which makes room for startups building new OS empires (LLM OS and moving away from apps towards generative UI, voice interfacing).
Special case: standalone prosumer apps
We looked at a bunch of companies having strong moats. Now, let's examine companies with weak moats (to be clear it's a spectrum, not binary).
I also want to clarify that many of these are products I love using, so this has nothing to do with the product itself being bad, but is more about evaluating if the business has a chance in becoming an empire.
There are several powerful standalone apps for power users providing more powerful alternatives to more popular generalist apps, such as:
- Apple Notes vs. Obsidian, Heptabase, Craft, Todoist
- macOS Spotlight vs. Raycast
- Native browser bookmarks of Chrome / Safari vs. Readwise Reader
- Google Calendar vs. Amie, Cron Calendar
These apps are unlikely to be fully replicated by operating system vendors because they would be too complex for the general population to use. However, since they target a specialized, professional-consumer audience, the market size is often very limiting.
These startups make a bet that we will see the 10x superhuman developer equivalents in a number of verticals, so that single people creating the productive output of 10 people: 1) need power-user software, and 2) would pay 10 times more for it.
Empires to build:
Big fragmented markets fragmented markets - which don't have a clear standard (yet)
It's clear that the most powerful companies in the world can often be seen as ubiquitous standards for human needs:
- Amazon: The standard for buying physical items (clothes, food, tech, home miscellaneous, medicine).
- Tesla and Uber: Becoming standards for human transportation.
- Apple and Google: Sharing the standard for personal computing.
- Google: Standard for web search.
- Nvidia: Standard for accelerated computing.
- OpenAI: Standard for building and distributing intelligence.
- SpaceX: Standard for space transportation
So the obvious questions are: what empires are left to build?
- What big markets are still fragmented and don’t have a clear standard yet?
- What markets are not even created yet? Jensen Huang refers to these as zero-billion-dollar markets, similar to how Nvidia created the GPU market, and Tesla created the EV market.
- What current empires offer outdated technological solutions and could be replaced with new approaches?
Fragmented:
- Housing
- Construction sites
- Corporate law
- Nutrition
- Healthcare
Potential zero billion dollar markets:
- Physics simulations for real world engineering AI (Nvidia Omniverse)
- BCI's
- Personalized genetic engineering
- ...
Obviously there are also markets where fragmentation is a feature, not a bug. Take restaurants and bars, for example: they should be fragmented because each bar or restaurant you visit should feel slightly different and unique every time. You don't want to build habits for entertainment experiences, as that would make them boring.
Deeptech
An obvious bet to make with AGI being right in front of us is that products which are currently hard to build will be easy to build tomorrow; products which are impossible today will be hard tomorrow.
- Real-world simulation
- Neuroimaging technologies
- Future of compute: quantum, photonic, neuromorphic
- AI for drug discovery (isomorphic labs)
- ...
Aside from hardware projects being harder to build, another dynamic is that there is generally less platform risk involved with hardware than with software. This is because, with hardware-first products, every startup operates within the same fundamental physical constraints and battles the laws of physics, rather than navigating artificial software systems governed by potential competitors.
B2B software
I think that advanced B2B software will continue to have strong moats at least for the next 4-7 years. It's unclear how this will play out with autocoders lowering the cost of software drastically and making it possible to copy any software out there just by observing it (including every update made to the software).
Unlike personal standalone apps, most B2B software at least would not profit from deep operating system integrations. SAP, HubSpot, Salesforce, or Celonis couldn't be used in a vastly more efficient way if they were integratively integrated into macOS, Windows, or the Chrome Browser.
Also, at least as of right now, they would be too complex to engineer (Figma, Solidworks, Adobe Premiere...) and run operationally (sales, legal, support) for operating system creators like Google and Apple. Even though there are good counterexamples with Apple's Logic as an Ableton competitor, or Apple Final Cut Pro as an Adobe Premiere competitor, or Windows Power Automate as a UiPath competitor.
Random thoughts
-
More startups getting featurized:
- Speechify getting featurized by Elevenlabs mobile app
- Pika and Runway getting featurized by OpenAI Sora
- Inflection AI Pi getting featurized by ChatGPT with GPT-4o
- Super.so getting featurized by Notion Sites
- Leap Motion featurized by Apple Vision Pro
- GPT-5 with real-time voice and real-time vision language model, seeing iPad or Mac screen, will featurize human math tutors
-
How many of the current empires looked like they had weak moat in the first ten years of their existence?
-
YC advice is that founders shouldn't care about competition because most startups are killed by suicide not murder. In the last year, this saying became questionable with Google IO, Apple WWDC, and OpenAI Dev Days being ceremonies of killing dozens of big and small startups.
-
It's fascinating to think of nation-states as companies with the most formidable moats imaginable:
- Vendor Lock-In: Many companies operate within or on top of nation-states, creating a form of vendor lock-in.
- Rule-Makers: Nation-states set the rules of the game—including laws, taxes, and political strategies.
- Physical Ecosystem Moat: They have complex infrastructure, developed over centuries, that would be incredibly challenging to replicate (cities, electricity grids, etc.).
- Strong Brand: Nation-states possess such a strong brand that people view them as natural entities rather than products. When a state becomes the ubiquitous definition of something, like "googling" is for web search, it essentially holds a monopoly.
- There are startups like Praxis trying to build new states
-
How much is the most happening, and is it largely uncontrollable and organic over time by just building things you would like to see in the world? Or how much is it strategically developed by folks like Satya Nadella, Sam Altman, Jeff Bezos playing business chess?
- Pretty sure there's no absolute answer here—it's clear that, as an early-stage founder, you first need to find product-market fit and be aware if you are unique with the value you provide to avoid being easily replicated or outdone. Thinking too much about building a moat or acquisition strategies early on would likely be a distraction.
- However, over time, the idea of just building tools and products you would like to see in the world, which at some point magically/passively form an empire, is wrong. The best founders I've met had a strategy to actively scale the product to an empire early on.
-
It's impressive how Google, Apple, and Microsoft have avoided the innovator's dilemma by creating startups within their own organizations:
- Google Ventures, Google X, and the culture of creating internal startups are prime examples. At Google, moving up the ladder often involves creating your own startup within the company. This explains both the rise and fall of many internal startups, but also drives significant innovation within the conglomerate.
- Sundar Pichai wouldn't be the CEO of Google today if he hadn't led the successful launch of the Google Chrome browser.
- These companies maintain their power by combining the best of both worlds—incubating and acquiring startups with small, agile teams while leveraging enterprise capabilities such as access to capital and global distribution.
- In contrast, Volkswagen (VW) has fallen victim to the innovator's dilemma, lacking a culture of spinning up startups within the enterprise. They missed out on acquiring a company like Tesla, which has been a major oversight.
- If the people and leaders in an organization do not think like startup founders, the innovator's dilemma is likely inevitable. Companies like Tesla, Meta, and Microsoft still maintain a startup-like leadership style. Based in San Francisco, these companies are deeply immersed in the innovative startup ecosystem, which contrasts sharply with a company like VW, led from Wolfsburg, Germany, by executives who certainly do not live and breathe technology.
-
I used to think personal fintechs have a moat, but Apple starting to kill Venmo and Cash App with their latest WWDC 24 announcements is impressive. Also, switching costs for personal customers in banking are way smaller than for businesses. To be fair, Apple and Google are the only ones who can pull this off.