#286: Bitcoin Is Legal Tender in El Salvador, & More
1. Bitcoin Is Legal Tender in El Salvador
El Salvador has become the first country to adopt bitcoin as legal tender. Approximately three months after announcing plans to do so, El Salvador’s government has rolled out its bitcoin wallet, Chivo. To encourage its citizens to download Chivo, El Salvador is airdropping $30 worth of bitcoin to adult citizens who have downloaded the app and registered their wallets.
While some Twitter users have showcased the app as intended, including bitcoin payments at Starbucks and McDonald’s, many others have reported technical glitches. El Salvador’s President confirmed on Twitter that the bugs plaguing Chivo necessitated a reboot to improve the user experience and server capacity. Many but not all of the bugs have been fixed.
El Salvador’s rollout is an important milestone for the adoption of bitcoin in developing countries. As noted by CNBC, the impact on the remittances industry could be significant, as the launch of Chivo in El Salvador could cost Western Union $400 million a year.
Other countries are following El Salvador’s lead, hoping that bitcoin will help them leapfrog the traditional financial services industry. This week, for example, Ukraine passed a bill legalizing cryptocurrencies and Panama unveiled a bill to make bitcoin legal tender. As ARK has noted in the past, bitcoin is catalyzing currency demonetization in emerging markets and could scale as its infrastructure reaches critical mass.
2. Coinbase Receives a Wells Notice Focused on Lend
This week, Coinbase announced that the SEC issued it a Wells notice focused on its impending Lend product. Coinbase Lend will allow users to lend crypto assets, starting with a 4% annualized yield on USDC, a USD-backed stablecoin. Other companies in the crypto space like Gemini and Circle also allow users to earn interest on loans, while decentralized finance platforms like Compound and Aave use smart contracts to enable high-yield loans.
The Wells notice suggests that the SEC is planning to sue Coinbase because it classifies Lend as a security. On Twitter and a recent blog post, Coinbase claims that it has engaged with the SEC for months but has received neither a warning nor explanation of the decision. Coinbase believes that Lend should not be regulated as a security but, awaiting more clarity, has delayed the launch of Lend.
Given developments like this, ARK is concerned that regulatory uncertainty will force crypto and other fintech projects to flee the United States. Mindful of the technologies transforming the financial services industry, we hope that regulators will continue to collaborate with those of us who are focused on the way that technology is changing how the world will work.
3. Mammoth Biosciences Becomes the Latest Unicorn: Now What?
In 2017, Drs. Jennifer Doudna, Janice Chen, Trevor Martin, and Lucas Harrington co-founded Mammoth Biosciences, a biotechnology company focused on developing CRISPR based diagnostic tools. On the heels of its series D round, Mammoth became a unicorn as its valuation surpassed $1 billion.
In August 2020, Mammoth licensed a new family of Cas proteins called Casɸ from the University of California, Berkley. This family of Cas proteins is smaller than legacy enzymes, potentially conferring upon it some advantages.
Recently, a plethora of new enzyme discoveries is stirring even more optimism. Dr. Feng Zhang’s former students, Dr. Jonathan Gootenberg and Dr. Omar Abudayyeh, discovered Cas 7-11. Dr. Feng Zhang’s lab discovered new systems IscB, IsrB and TnpB, a class of transposon-encoded RNA-guided nucleases, likely to be programmable as gene-editing enzymes.
ARK believes that gene editing companies will innovate beyond the original intellectual property (IP) associated with CRISPR enzymes, potentially impinging upon the IP of the Cas-9 enzymes. Already, companies are using Cas12a to circumvent Cas9 IP. That said, we believe the likely outcome is cross-licensing of tools and enzymes as gene editing becomes essential to therapeutics and diagnostics, curing diseases.
4. Will Synthetic Data Erode Real-World Data Advantages in Artificial Intelligence?
An artificial intelligence (AI) model is only as good as its training data. When the data-value asymptote is high, more training data typically results in better performance in complex models. Companies with large volumes of high-quality training data often enjoy a data advantage that results in a competitive moat.
Increasingly, researchers are generating synthetic data sets with a variety of techniques. Synthetic data fills gaps in real-world data sets, reducing data acquisition costs. One estimate suggests that the cost savings could be greater than 98%.
Now we wonder if synthetic data will eliminate the need for real-world data and erode competitive moats? In our view, for highly complex tasks like self-driving, real-world data provides important validation that synthetic data cannot replace. While large volumes of real-world data can surface weird, unknown edge cases like a truck carrying a load of stop signs or a human throwing ice cream at cars from an overpass, simulations can train models with rare edge cases like trailers jack-knifing on snowy mountain roads. The best training solutions are likely to combine large volumes of real-world data and synthetic data in simulations.