#345: AI Software Arrives On The App Store, & More

A photo-editing app launched in 2018, Lensa AI, recently exploded to the top of the App Store charts with its new generative-AI feature called Magic Avatar. For a few dollars, the app can train a personalized AI model based on user selfies and generate hundreds of stylized portraits. The product clearly has hit a responsive chord. Here are some of the results for ARK’s Chief Futurist, Brett Winton.

At an annualized rate, Lensa AI’s iOS App Store revenue now tops $1.2 billion, up ~100x since the launch of the Magic Avatar only 2.5 weeks ago.[1] Until the release of Magic Avatar, the app had averaged less than $10 million in annualized revenue, peaking at ~$16 million, as shown below.

According to our research, sales have spiked not only because of an explosion in the user base but also because of higher monetization rates. Daily active users are up 37x and monetization per user ~3-fold as users pay for the Magic Avatar AI training and image generation, as shown below.

While Magic Avatar could be a flash in the pan, we believe personalized AI models are likely to drive significant business opportunities in the future. Interesting to see will be Lensa AI’s ability to capitalize on the mindshare it has gained. The idea that users are willing to pay for personalized output should inspire the creation of many consumer- and enterprise-facing AI solutions.

[1] All data are from Sensor Tower as of Friday, December 9, 2022. [https://sensortower.com]

Molecular residual disease (MRD) tests use next-generation sequencing (NGS) to assess whether a cancer patient is responding to therapy. According to our research, MRD testing is poised for significant growth over the next market cycle.[1] Historically, clinicians have used medical imaging to measure residual disease or spot a relapse. In contrast, MRD tests use a tube of blood. The significant decline in NGS costs during the past decade has enabled diagnostic labs to offer profitable blood-based MRD tests for roughly the same price as CT-scanning. Although imaging may remain the gold-standard, we believe more oncologists are likely to adopt blood-based MRD testing based on the clinical evidence of its utility.

Recently, companies like Illumina (ILMN) and Ultima (Private) have predicted that they will reduce genome sequencing costs to $200 and $100 per whole human genome, respectively, next year, which would improve the economics of MRD tests. In our view, innovative diagnostic labs will broaden MRD tests well beyond the few dozen mutations in the bloodstream they investigate today to thousands more, potentially enabling the detection of cancer recurrence earlier than otherwise would be the case, as shown below.[2] Earlier detection of more mutations should help biopharmaceutical companies understand how various therapies, whether FDA-approved or experimental, will impact tumors.

Source: ARK Investment Management LLC, 2022, based on data presented in Zviran, A. et al 2020. Forecasts are inherently limited and cannot be relied upon. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

[1] Here we define a market cycle to be roughly five to seven years.
[2] Tumor fraction refers to the amount of circulating DNA that is tumor-derived (ctDNA) rather than healthy cell-free DNA (cfDNA).

Reports suggest that Apple’s autonomous car project may have switched gears yet again, lowering its odds of success. Apparently, the company plans to focus on a vehicle with autonomous features relevant only for highways and to delay the launch by one year to 2026.

ARK’s research suggests that autonomous taxi platforms will create a multi-trillion-dollar market opportunity, stimulated by price points much lower than those for personal transportation today. Importantly, our estimates assume that the vehicles will be equipped with fully—not partially—autonomous technology. Further, we expect autonomous taxi platforms to submit to natural geographic monopolies. If our assumptions are correct, Apple could be lagging far behind competitors by 2026. Waymo and Cruise already have launched commercial autonomous services in Phoenix and San Francisco, and ARK’s expectation is that Tesla will launch its fully autonomous service in 2024.

In the aftermath of the FTX fiasco, custody and proof-of-reserves are two of the most important topics in the crypto industry. FTX CEO Sam Bankman-Fried allegedly funneled FTX customer funds to sister hedge fund Alameda Research without customer consent. Such negligence, if not fraud, has underscored the importance not only of custodian counterparty risk when users hold cryptocurrencies on third party platforms but also of cryptographic guarantees that funds are backed 1-to-1.

In this context, the two most secure methods of holding cryptocurrencies are self-custody and collaborative custody. Self-custody is simpler, as the cryptocurrency owner holds his or her own private keys in a hardware-based, cold wallet. Collaborative custody removes centralized control by requiring multiple participants to sign—“multi-signature” or multisig—and approve a transaction.

Multisig distributes the coin’s private keys across three different signatures—two to the owner and one to a service provider. Transferring funds requires at least two of the three signatures, with the custodian service provider serving as a backup if owners lose either of the two keys. Among the leaders in collaborative custody today are Unchained Capital, which services bitcoin holders, and Casa, which services bitcoin and/or ether holders.

In addition to vaulting services, companies offering collaborative custody provide access to other financial services, like lending, in a multi-institutional model. In that model, the client holds one signature, the service provider another, and a designated third party the third. Through the multi-institutional protocol, clients can borrow in USD by collateralizing their BTC and maintaining 100% of it in cold storage. By providing a third party with the last key, clients give counterparties—like the lender and the custodian service provider—recourse to settle loans in the event of default.

At ARK’s brainstorm last Friday, Dhruv Bansal, Co-Founder and Chief Security Officer at Unchained Capital, noted, “[Collaborative custody] relieves the holder of what is a false dichotomy: the security of their assets and the usefulness of financial services such as lending.”

In our view, collaborative custody, self-custody, cryptographic proof-of-reserves, and decentralized financial services are emerging as best-practice alternatives to holding funds on centralized platforms.