What We’re Reading (Week Ending 15 September 2024) - 15 Sep 2024
Reading helps us learn about the world and it is a really important aspect of investing. The late Charlie Munger even went so far as to say that “I don’t think you can get to be a really good investor over a broad range without doing a massive amount of reading.” We (the co-founders of Compounder Fund) read widely across a range of topics, including investing, business, technology, and the world in general. We want to regularly share the best articles we’ve come across recently. Here they are (for the week ending 15 September 2024):
1. The ROI on Generative AI – Tanay Jaipuria
The poster child for this has been Klarna which leveraged AI to elevate their customer support. Their AI assistant has taken over the work of 700 employees, reducing resolution times from 11 minutes to just 2 minutes while maintaining high customer satisfaction levels…
...Microsoft casually dropped that they too are expecting to save hundreds of millions of dollars a year on call centers after adopting Generative AI.
“Dynamics with Gen AI built in is sort of really … the category that gets completely transformed with Gen AI, contact centers being a great example. We, ourselves, are on course to save hundreds of millions of dollars in our own Customer Support and Contact Center Operations. I think we can drive that value to our customers”…
…We’re also hearing examples of measurable, tangible benefits from enterprises, as Amazon shared about their software development assistant Q which has saved them over 4,500 developer years in a recent code migration task…
…”With Q’s code transformation capabilities, Amazon has migrated over 30,000 Java JDK applications in a few months, saving the company $260 million and 4,500 developer years compared to what it would have otherwise cost. That’s the game changer. And think about how this Q transformation capability might evolve to address other elusive but highly desired migrations.”…
…eBay launched a new AI-assisted selling flow, and are already seeing improvements in customer satisfaction as well as faster time to list and get value for Sellers…
…YUM Brands is enhancing customer experiences at Taco Bell by rolling out voice AI driven drive-through systems. This technology is not only improving customer satisfaction but also boosting team member productivity, and the results are so promising they are accelerating their roll-out timelines…
…Manulife is an example of a company already seeing large ROI in using AI to assist salespeople…
…”We’re using GenAI and machine learning models to make it really easy for agents to understand customer opportunities but also to generate these personalized communications at the click of a button to help them engage with more customers more often.
In our first 2 weeks live, about 68% of our agents had already used the new GenAI capabilities. And in July, we will be broadening that user base to about 2,000.
Based on our analysis in Singapore, we anticipate a 17% uplift and repurchase rates for our customer base, when this is fully rolled out to all of our agents.”…
…Rocket Mortgage is utilizing AI to automating the transcription of client calls and completing mortgage applications…
…”Now the Rocket Logic Assistant seamlessly generates over 300,000 detailed transcripts every week from outbound calls. It supports over 100 data points on mortgage applications saving our bankers from inputting tens of millions of data fields each week.”…
…Walmart has harnessed generative AI to enhance its product catalog, improving the quality of over 850 million pieces of data…
…”We’ve used multiple large language models to accurately create or improve over 850 million pieces of data in the catalog. Without the use of generative AI, this work would have required nearly 100x the current head count to complete in the same amount of time.”…
…Mastercard is leveraging the new advances in Generative AI to enhance fraud detection, achieving a 20% increase in accuracy.
2. The Agent Era – Patrick O’Shaughnessy and Bret Taylor
Patrick
It’s such an interesting story because I think it becomes ultra relevant in today’s world. And you hear a lot about this, maybe the mythical 10x engineer, the 100x engineer, 1,000x engineer, the leverage available to one person with a growing tool kit.
And maybe that’s a great excuse to bridge the conversation into agents. I think everyone listening will have heard that term and maybe have thought about it a little bit, have gotten excited about the prospect of some sort of autonomous agent doing work on their behalf or their company’s behalf. But it would be great for you to ground us in your definition of what one of these things is, if this becomes a really critical part of the world of technology in the next year or two. I think it would be great for everyone just to have a level-set, simple definition from your perspective on what an agent is and does.
Bret
I’ll start with maybe the academic flavor of this, but then I’ll move into what I think is maybe the more — what I believe is the more relevant definition, but agent is like the word app. There’s not one definition, and I think it will be a noun that is quite meaningful in the age of AI. The word agent in the context of AI comes from the word agency and essentially is a system that can reason and take action autonomously is the way I think about it. And a system that is agentic is one where software and AI can reason and make decisions and take action without human intervention, which is really exciting but something that is relatively new though the idea is certainly not new.
I think the effectiveness of reasoning with AI systems has become so meaningfully better over the past couple of years that I think the concept is — like many parts of AI, the ideas are not new, but the effectiveness is, and so we’re living in an era of agents now.
In practice, I think the word agent, just like the word app or site in the age of the web, will become important to all of us. So one agent that I think is important is what my company Sierra does, which is your company’s conversational AI. And so just imagine you’re a retailer. I think you’ll put as much care and attention into your AI agent as you do your website or your mobile app. Or if you’re a bank, and you’ll put as much care and attention to your AI agent, which can help a customer look up the balance of their checking account or perhaps be an interface to your investment banking arm or wealth management arm. Or if you’re a streaming service, your agent might help people sign up for a plan or upgrade or downgrade their subscription, as an example.
In that case, an agent is something like website or mobile app that’s branded and it’s yours. And there are parts of it that are about agency and sort of the AI definition of the word. But more importantly, it’s your thing. It’s your digital asset. It becomes the digital manifestation of your brand.
And that’s what my company Sierra does. And we think that’s one really important part of an agent. Just like in 1995, the way you existed online was to have a website, we think in 2025, the way you will engage with your customers will be your AI agent, and we think it’s a really important new category.
But then taking, okay, what are the other types of agents out there? One will be, I’d like to think of them as persona-based agents. They’re internally facing. They do a job. You’ve talked about software engineering. I think there’ll be software engineering agents that will work to produce software. I was looking at a start-up called Harvey, I think, that’s making a legal LLM, which is super interesting. And I think across many job functions, there will be AI agents that produce the output of a — whether it’s a paralegal or a software engineer or an operations analyst, things like that. So that’s one.
So there’s your company’s agent, there’s a persona-based agent that does a job, and then the third one — category is probably personal agents. So this is the agent that will work on your behalf, whether it’s helping you plan a vacation or organize your calendar or perhaps triage your inbox and things like that. I think technically, they’re all similar, but my guess is they’re different enough in what job they accomplish for you that there’s — probably different companies will build those different categories of agent.
If you’re building a software to be a personal assistant agent, the breadth of systems you have to integrate with is infinite because different people use different calendars and different this and different that, and there’s lots of interesting investment into that. If you’re building a coding agent, it’s a much more narrow use case but very deep, and you’re probably evaluating it based on benchmarks of the effectiveness of the software produced and the robustness of the software it produces…
…Patrick
What do you think are the next most important unlocks for the power of these agents? You mentioned their access tools, access to the Internet. I’ve heard people talk about the ability to have some sort of stored memory about you, the customer or the specific customer or just memory in general that doesn’t just live inside of a context window that’s always re-fed in or something.
Are those the three things that we need to unlock the next tier of productivity out of agents? Are there other things that you and Sierra are focused on? I’d love to get down to the nitty-gritty capabilities and roadblocks that you’re thinking about and working on that might make these things as ubiquitous as you think they will be.
Bret
Yes. I’ll start with the vantage point of Sierra. We help companies build customer-facing AI agents. Today, if you’re setting up a new Sonos speaker, you can chat with an AI agent they’ve built on our platform to help you set it up. If you’re a SiriusXM subscriber, you can chat with Harmony, which is their AI agent they’ve built on our platform. And if you’re a WeightWatchers member, if you click on the 24/7 live coaching tab in their app, that’s an AI agent they’ve built on our platform.
One of the things that I think is a nuanced problem that is not strictly technical in nature is just the act of actually designing conversational customer experiences is a relatively new discipline. I remember in the early days of the Internet, most websites looked like DVD intro screens, like they’re very graphical, there’s four big buttons. It’s really interesting to go down the Wayback Machine and look at them.
And I would say it took a number of years to evolve into sort of the design idioms that we recognize with websites today. And now if you go to a retailer, they’ll have a hamburger menu on the top left, and the way you filter through items and these — they’re sort of emergent from people’s lived experiences, both designing and using websites.
And now you can talk to almost any web developer. And they’ll not only choose similar technologies to make a website, but even the design process and Photoshop or Figma to design a website, they’re sort of established practices, some of which are obvious and some of which are actually subtle, like why did this become the way these things are done, and it’s the cumulative experience we have building with them.
The difference between a website in a mobile app and an AI agent is both the breadth and non-determinism of AI agents. So if you have a menu on a website, you can control what links are there, and it’s essentially multiple choice, here’s the options available to you. If you have an AI agent with a free-form text box, people can type whatever they want into that. And so your concept of what your customer experience is defined by you, but it’s also defined by your customers, by what they write in there.
It reminds me — going back to my web analogies here, it reminds me of going from Yahoo Directory to Google Search. Rather than having a taxonomy of everything available, it’s just free form, and there’s a much longer tail of queries in Google than there was in Yahoo! because of the expressiveness of a search box versus a directory.
And I think that that’s one of the really interesting and, I think, exciting opportunities with conversational AI for customer experiences is it’s a really authentic way to actually hear from your customers what they want from you. And I think we’ve — and so it sort of stands to reason, your website was the rails on which your customers communicate with you. And this is a free form that I think it’s much more expressive. And we’ve had multiple customers learn things about their customers that they didn’t expect by providing this really free-form experience.
And then similarly, I think the other really interesting thing when I mentioned non-determinism is the word agent comes from agency, and it’s really how much creativity do you want to give your AI in interacting with your customers. I think if you start from a position of control, you can say, I want to put guardrails around everything, but then your conversational customer experience is somewhat robotic. You’ve essentially defined the multiple-choice options of your customers’ experience. If you give your agent too much agency, in the extreme case, it will hallucinate, but in the more practical case, it just might not protect your brand in the way that you want it to.
And I would say that design question is both a technology question, which obviously we’re quite invested in solving, and I’m really excited about some of the work we’ve done there, but there’s a deeper question here, too, that’s actually a philosophical branding and design question as well. And what we’re trying to do at Sierra is not necessarily predefining answers to those questions. I think every company and every brand will have a different perspective on what’s correct for their brand experience but provide a platform that’s powerful and expressive enough. Whatever your answers are personally to that question, you can build your agent on Sierra.
Patrick
It’s so interesting to think about the customer experience going to a website where I buy shoes or something. I think one of your first customers was flip-flops, and there was a funny story around that, but I’m going to buy a pair of sandals, let’s say, on a website. And rather than click around, I just describe what I want and I can imagine like another pane on the right just starts showing me stuff. And then maybe I check out through this same thing as well, and that’s a simple version of tooling or ability to take action.
I’m curious what the hardest parts for you have been to build. It’s quite technically daunting to even think about how to build something like this, let alone one that’s adjustable and tunable to my specific brand. So talk a little bit about how hard of a technical challenge this is for Sierra, like the degree of difficulty you’ve encountered relative to, say, your expectation.
Bret
Yes. It’s a really wonderful question. I think that generative AI broadly is a technology with which it’s very easy to make a demo and very hard to make an industrial-grade system. And I think that’s the area of technical challenge that we’re really trying to dive into. And I think it’s one thing to say that this system does the correct thing 90% of the time. And it’s really an inkblot test whether 90% is a really good number or a horrible number.
And it also depends on the process. And so if it’s a consumer application that was helping you with your homework, maybe 90% is decent. If it’s something operating revenue impacting part of your business or there’s a compliance concern, it’s absolutely unacceptable to be wrong 10% of the time.
And so a lot of the challenges that we’re facing are, we like to say that software systems are moving from rule-based to goals- and guardrails-based. And it’s a very different mental model for building software systems. Rule-based systems, if you think about just the software development life cycle that’s evolved over the past 20 years, it’s really about how you make more and more robust rule-based systems, how do you ensure that the same input produces the same output, that it’s reliable, that it’s stable, and there’s a lot of true innovation in the way we make software to make them more secure and robust.
Now if you have parts of your system that are built on large language models, those parts are really different than most of the software that we’ve built on in the past. Number one is they’re relatively slow compared — to generate a page view on a website takes nanoseconds at this point, might be slightly exaggerating, down to milliseconds, even with the fastest models, it’s quite slow in the way tokens are emitted.
Number two is it can be relatively expensive. And again, it really varies based on the number of parameters in the model. But again, the marginal cost of that page view is almost zero at this point. You don’t think about it. Your cost as a software platform is almost exclusively in your head count. With AI, you can see the margin pressure that a lot of companies face, particularly of their training models or even doing inference with high-parameter-count models.
Number three is they’re nondeterministic fundamentally, and you can tune certain models to more reliably have the same output for the same input. But by and large, it’s hard to reproduce behaviors on these systems. What gives them creativity also leads to non-determinism.
And so this combination of it, we’ve gone from cheap, deterministic, reliable systems to relatively slow, relatively expensive but very creative systems. And I think it violates a lot of the conventions that software engineers think about — have grown to think about when producing software, and it becomes almost a statistical problem rather than just a methodological problem.
And so that’s really what we’ve tried to solve. We shared on our website, but we have a process we call the agent development life cycle, which is the name comes from, say, in the software development life cycle, here’s what you should do with these agentic platforms. It’s also — we’ve developed a lot of unique technology to make these systems more robust with having one AI model supervise another AI model to layer different models on top of each other to produce statistically more robust results.
And then as importantly, we’ve developed ways that folks who aren’t experts in AI can express the behavior that they want in their agent. You shouldn’t have to be an AI expert to make an agent just like you shouldn’t have to have a PhD in computer science to make a website. I don’t think we’re there yet, but that’s really what we’re trying to solve.
And broadly speaking, I would say, on the spectrum of fundamental research institutions like OpenAI, we’re not that we’re applied. We’re really thinking about how do we engineer on top of these foundation and frontier models to produce robust or reliable agents for our customers.
Patrick
I love the title of this one Kevin Kelly book, What Technology Wants, and I’m curious what agents want. If I’m a customer, I’m a prospective customer, and I want to go work with Sierra to make the best possible version of a conversational agent for my customers to use, what can the companies provide that make the agent do the best job?
Bret
Yes, it’s a great question. I would say that there’s two types of knowledge that I think really produce a really robust agent. One is the factual knowledge of your company. This just grounds the agent so that it won’t just make something up.
There’s a pretty widely-used technique called retrieval augmented generation in AI right now that effectively means rather than relying on the knowledge encoded in the model to answer questions, you present the model with knowledge, maybe stored in a knowledge base or a database and say, “Hey, summarize the content from here. Don’t rely on the information you’ve been trained on.”
That has been an effective technique for two reasons. One is that it means that you don’t necessarily need to train or fine-tune a model to use it with proprietary data, which is a much cheaper deployment methodology. And it also can be effective at preventing hallucinations as well because you’re effectively — rather than relying on the AI to determine what it knows or doesn’t know, you present the AI with the knowledge that it’s allowed to know, a simple way of putting it.
And that’s factual knowledge. And I would say that’s necessary, but woefully incomplete because that would enable your AI agent to answer questions, but it wouldn’t necessarily enable it to orchestrate a complex process or take action on your customers’ behalf.
The other type of knowledge is procedural knowledge. We have a Sonos speaker. It stops working. What would the best Sonos engineer ask you and do to figure out whether it’s a problem with your hardware, a problem with your Sonos app or a problem with your Wi-Fi? Like what is the process by which you do that.
If you’re a subscription streaming service, what is the process of processing an upgrade or downgrade to your membership? Are there different offers available based on your membership level? Do you have a promotion running? What’s been the most effective technique to keep people, a subscriber for a long period of time?
This is all the stuff that if you are a person and expert in it, and so coming in with that knowledge of not only here’s the factual knowledge for our company, but here’s the processes that represent our greatest customer experience. What does the best salesperson do? What does the best customer service person do? What is — the most effective marketeer at your company, how do they describe your products? And that’s often there we work with our customers to improve when they deploy AI.
And then the third thing is just access to the underlying systems themselves. I think the AI agents shouldn’t just be about answering questions or having a conversation, they should actually be able to take action on your behalf, whether that’s a retailer processing a return or a subscription service, changing your level of membership or connecting to the telemetry system of a consumer electronics company. So we can say, “Hey, we know your device phoned home. You’re connected. We now figured out this other problem.”
Or even with something like SiriusXM sending a signal down from a satellite to refresh your radio if your radio stopped working. So three ingredients, factual knowledge, procedural knowledge and systems integrations, I think, are the three key ingredients. And then with the right methodology, your agent can do anything that a person could do on a computer, which is just an incredible opportunity for customer experiences.
3. Here’s What Happens When Credit Markets Go Dark – Joe Weisenthal, Tracy Alloway, Jared Ellias, and Elisabeth de Fontenay
Joe (11:53):
You spell out this evolution of the debt markets and the historical things you’re taught in law school about the dangers of single lenders. We’ve talked to people in the industry and they have their explanations for why this particular market has boomed. But from your research, what would you say are the drivers of this? Or when you talk to people, what problems does the private credit market solve for them?
Elisabeth (12:19):
The interesting thing about this is that there’s multiple stories going on at the same time. So one is that, this is just actually substituting for a lot of the activity that banks did because the banks, ever since the financial crisis, have been really constrained for a lot of reasons. One, they’ve primarily been constrained because of regulation, and sort of regulation designed to discourage them from making risky loans and from, you know, to have diversification in their portfolio, and so on. And just their evolving model of doing business, that they prefer to be sort of the middleman and get some fees rather than lend directly. [There are] all kinds of reasons why banks have retreated from particularly the lower middle market, but also all the way to the largest companies. A second story is just that there’s been too much bank regulation. So, I’m not going to take a position on whether that’s true or not, but that bank regulation is stifling the banks and they can’t really lend and so on.
A third story is one that we find really interesting and appealing, which is that, it may just be that it never really made all that much sense to fund loans using bank deposits. That essentially, you have a very short-term liability, which is customer deposits, and very long-term assets. So some of these loans, of course, are multi-year loans. And that’s just a fundamental mismatch that banks have always struggled with and that bank regulation has always struggled with. And this is a really nice, neat solution to that. And the reason it’s showing up now is that, thanks to sort of loosening of some of the securities laws and other things, it’s finally the case that you can get these investment funds that are big enough to actually take over the role of banks. And for them, the sort of positive side of private credit is that you now have a better match between the funding source, which is you have these big institutional investors putting capital into private credit funds that is locked in for a number of years, and you’re matching that really well against the loans that are also multi-year. So in some sense, it’s actually a better fit than banks for financing this type of loan…
...Joe (20:43):
It sounds pretty good to me. Okay, so there’s less legal fees, less creditor on creditor violence, liability asset matching, the better user experience. So what’s the catch? I don’t see any problems.
Elisabeth (20:57):
One potential problem is, of course, these are, in some cases, absolutely massive loans. And so you do lose diversification benefits. These are very risky investments. I would say, the private credit structure has a partial solution to that problem, which is that, the investors themselves in a private credit fund oftentimes are so massive themselves that they really don’t lose diversification, which is to say, their portfolios are so large that they can make this enormous investment in one private credit fund because that’s a tiny piece of their portfolio. So that’s one downside of private credit. The other of course, is the absence of trading. So before. you had pretty good signals of what your position was worth. There were lots of syndicated loans that had pretty active trading and there were indices tracking all of this. The [Loan Syndications and Trading Association] LSTA provides lots of data on the loan market, and, of course, the bond market is public in terms of the pricing there. So exit is always going to be a concern in this market, and I don’t think this market really has been truly tested yet. So we’ll have to find out. But that illiquidity can be an issue depending on what kind of investor you are and what your expectation is for getting out of these things…
…Tracy (30:14):
Just to play devil’s advocate for a second, I think this is something you actually deal with in the paper, but one of the things you hear from people in the private credit industry is tha, ‘Oh, well, if you’re getting funding from a private entity, maybe a single lender or maybe a club of lenders but it’s a smaller group than you would have in the public market, maybe there’s greater potential for working out your issues if you get into trouble. So you can renegotiate your debt with a smaller group of creditors and maybe they know your business better than like a big fund that is buying pieces of all these different types of bonds and things like that.’ What’s your response to that argument? This idea that, well, private credit actually allows you to have more room for workouts or maybe even stave off bankruptcy for longer?
Jared (31:09):
So, I guess my answer is that, that all sounds great, but it’ll depend. And it’s hard to really understand which way any of these forces cut. The one thing that’s clear cut, that’s important is, we’re losing the claims trading markets. Like, that’s just going to look a lot different. Like, the active market and the claims of Chapter 11 debtors, when that debtor is a private credit funded firm. But, as to the question of, ‘Well, you know, aren’t these private credit lenders smarter, more versatile, more nimble, able to commit capital? And won’t that be good for companies?’ You know, at the end, it depends. So something you worry about is, well, maybe private credit lenders will have incentives, not to adjust their marks on their books and instead, just to do ‘amend and extend’s, and just keep loans going when the company really needed to liquidate or should have filed for bankruptcy sooner.
Think about how different the GM bankruptcy would’ve been had they filed for bankruptcy in like 2005 versus 2009 when their business had already eroded so much. So we think of that erosion as something that limits reorganization options. And it’s not necessarily obvious how private credit interacts with that. Because private credit lenders have their own incentives and maybe their incentives are to say, ‘Look, we make loans to sponsor backed companies and if the sponsor wants to continue, we’re going to keep doing that because we really want to participate in their next deals.’ Or they could say like, ‘Let’s pull the plug on these things earlier.’
So something that I’ve heard from lawyers working in this space is that when private credit lenders replace like your mid-market banks, like your Citizens and that kind of bank, when you have like a private credit lender with a $30 million loan that might have been done by a syndicate of two regional banks, the private credit lenders are much more aggressive and much more willing to pull the plug on the company and to own the asset then that bank might have been, but the world could look very different for larger companies where private credit lenders might be easier for companies to do workouts with. So it’s really hard to tell. But I’m certainly a bit skeptical of the idea that all of this is unidirectional and the private credit is just better in every way for everything. It’s different and there’ll be different pros and cons and we’ll learn more about them, and the law will adapt and hopefully deal with some of the ways in which the incentives of private credit lenders distort bankruptcy outcomes.
Tracy (33:28):
Since you mentioned GM, could you maybe talk about another specific example of a liquidation playing out a bit late, as you describe it? I’m still salty over the collapse of Red Lobster, which you mentioned in your paper. So could you talk a little bit about that one and what it tells us about private credit?
Jared (33:47):
Sure. So, something that has been the case over the past few years is you’ve had private equity owned restaurants and retailers that just ended up doing quick liquidations after stalling for a very long time. Red Lobster is really interesting. Red Lobster had been struggling for a little while and then Fortress Investment Group, which was its private credit lender, came in and took over the company and basically just owned the asset very quickly. And something that is so interesting about that is that, traditionally, other lenders would’ve been a lot more cautious about doing that, because other lenders are very cognizant of what we call ‘lender liability’ and this line of law that suggests that you shouldn’t, if you’re a lender, play too much of a role in business decisions of companies that you lend to.
And like, there’s an example of like a private credit lender just behaving in this really aggressive way, which is interesting. Like, again, it’s hard to tell exactly what’s going to happen, but certainly that example doesn’t fit well with the story of, well, you know, the private credit lender is just like the banker and you know, it’s your corner bank in 1925, who’s going to work with you on your farm. The answer is, maybe some of the time that’s the story, but other of the time, you’re dealing with a very sophisticated party who may have different incentives and be worried about different things than traditional bank lenders or investors in the broadly syndicated market.
4. Flash Crashes Are Getting Faster – Ben Carlson
In the spring of 1962, the stock market was already in the midst of a double-digit correction. Then on May 28, there was a flash crash, sending stocks down nearly 7% in a single day. It was the biggest one day sell-off since the Great Depression…
…It’s becoming clearer by the day that last Monday’s stock market swoon was also a flash crash. As of August 5, the S&P 500 was down more than 6% for the month. It’s now positive in August…
…Flash crashes happened in the 1920s, they happened in the 1960s and they happen today.
The biggest difference between now and then is the interconnected nature of the global markets. You have computer and algorithmic trading. Information flows at the speed of light. Every piece of economic data is parsed in real-time with a fine-tooth comb.
Overreactions can happen much faster now.
Just look at the biggest gap downs over the past 40+ years:
This chart shows the biggest difference between the opening price of the stock market and the prior day’s close. All of them have occurred this decade outside of the 1987 crash…
…We are likely to see more of these flash crashes in the future due to a combination of increased leverage in the system, globalized markets and computer trading.
The hard part for investors is that it’s now easier to lose control during these types of market events. You don’t have to call your broker on the phone to place a trade. You can change your entire portfolio on your phone with the push of a button.
Just because markets are getting faster does not mean your decisions must be made faster.
5. Gaining Currency – Rachel Cheung
In its effort to cement its role as an innovation powerhouse, China’s most ambitious technological debut was also its most controversial: The digital yuan was rolled out as the legal tender of choice for the Olympic games. Instead of cash or Visa (the corporate sponsor that had dominated the sports event for three decades), visitors were encouraged to exchange foreign currencies for digital yuan at automated teller machines and to pay digitally through the e-CNY app on their phones or through a card that can be used offline…
…Yet, despite all the attention, the launch of the digital yuan largely fell flat. The COVID-19 pandemic meant Olympic visitors were confined to “bubbles” with little opportunity to travel, shop and dine out, and very few foreigners chose to use the digital yuan over their credit cards. Beijing saw just $315,000 in digital yuan processed every day over the course of the games — a small fraction of the usual revenues at the Olympics. At the 2008 Olympics in Beijing, for instance, the city generated roughly $264 million per day…
…But while China acknowledged its Olympic failure, it has also quietly doubled down on the digital yuan, including a big push to drive adoption. Last year, several cities began paying civil servants and collecting taxes in digital yuan. Jiangsu province saw the most recorded transactions in the country after it gave away 30 million yuan ($4.18 million) in digital “red envelopes.” And this past May, the digital yuan expanded for the first time outside of mainland China when it became available for use in Hong Kong. Though there is no timeline for a nationwide launch yet, China has rolled out pilot schemes in 26 cities and 17 provinces since 2019.
The efforts have paid off. In a press briefing last week, the PBOC announced that total transactions reached $7 trillion yuan ($982 billion) in June — a four-fold jump since last June.
Digital yuan usage is still only a fraction of China’s $40-trillion payment market, of course. The total number of e-CNY wallets opened — 120 million as of last July — also trails behind that of Alipay, which had over a billion users by 2020 and recorded $118 trillion worth of transactions in one year alone.
But as Beijing continues to crackdown on its fintech giants, it is creating room for the digital yuan to rise. In fact, officials see the transition to digital currency as both necessary and inevitable. According to Yi Gang, former governor of the PBOC, the current moment of transition is not unlike that of the Ming Dynasty, when the government started taking tax payments in silver instead of labor and grains. China’s currency has evolved with time, he said during a speech at Fudan University in April, and “the digital yuan is no exception.”…
…Officials are also trying to expand the scope of e-CNY beyond consumer retail transactions. The Bank of China, for instance, has tested the use of “smart contracts” for afterschool programs in Chengdu of Sichuan province: Parents can pay a deposit in e-CNY to educational institutions, and the latter only receives the money after the lessons are taken.
These business-to-business and government programming applications could be a “game changer,” according to Warwick Powell, a senior fellow at Taihe Institute, a Beijing-based think tank, because they “ensure that the provision of certain funds can only be used for certain activities.”
Yet that same function triggers concern for others. For instance, although some local governments and banks have offered loans in e-CNY, companies are reluctant to take them, says Yang You, a finance professor at University of Hong Kong. “The nature of e-CNY is that a policymaker can generate a loan and see where it flows to,” says You. But companies, he notes, would much prefer non-traceable loans, despite repeated assurances from the People’s Bank of China that it will not hold information against them…
…Instead, the PBOC says the digital yuan follows a principle of “anonymity for small value and traceable for high value” as a way of striking a balance between privacy protection and combating criminal activities, such as tax evasion and money laundering. The e-CNY wallet, for instance, requires users to undergo a more complex verification process in order to unlock higher transaction limits…
… If anything, the search for an alternative to the U.S.-backed Swift, the global messaging network for the banking system, has gained momentum since the U.S.-led sanctions on Russia.
“China has used the sanctions as a reason to advance the cause of de-dollarization,” says Elizabeth Economy, a senior fellow at the Hoover Institution at Stanford University and recent advisor to the Department of Commerce. “It has made the case that the United States is weaponizing the dollar, hence other countries should begin to trade in their own currencies. It’s actually a deft diplomatic move on the part of China.”
According to the Bank of International Settlements (BIS), a survey of 86 central banks last year showed a sharp uptick in experiments with “wholesale CBDC” — transactions between banks and other financial institutions, rather than consumers and businesses. In October, for instance, the e-CNY set a new milestone: At the Shanghai Petroleum and Natural Gas Exchange, the state-owned PetroChina used digital yuan to purchase a million barrels of oil from an undisclosed seller.
“There’s still a conversation about the e-yuan [for domestic retail transactions], but there’s more discussion about a regional payment system,” says Victor Shih, an associate professor of political economy at the University of California. “An alternative to Swift potentially has more legs.”
The oil purchase seems to be a one-off so far, but a new project called mBridge hopes to make such transactions routine. It is a collaborative effort between the “innovation hub” of BIS and the central banks of five jurisdictions: China, Hong Kong, Thailand, United Arab Emirates, and most recently, Saudi Arabia.
Underpinned by distributed ledger technology (which records transactions in multiple places at the same time), mBridge aims to be a multi-CBDC platform that can support instant cross-border payments. The idea is to make international settlement faster and cheaper than Swift. But it also means things are not dependent on the U.S. dollar.
Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. We currently have a vested interest in Amazon, Mastercard, and Microsoft. Holdings are subject to change at any time.
