#211: Lending, Autonomous Vehicles, Debt, Bitcoin, & Sequencing

Using alternative data and machine learning, MercadoLibre seems to have gained a competitive advantage in identifying both high-risk borrowers and companies that otherwise would not qualify for credit. With data from more than 81,000 sellers in Argentina, MercadoLibre’s internal credit ratings contrasted significantly from those of traditional credit bureau ratings, based on research from the International Bank of Settlements (IBS). The IBS graphed MercadoLibre’s loans based both on credit bureau ratings and on MercadoLibre’s A-E rating scale.

As illustrated by the blue line, the average loss rate on loans that credit bureaus would have categorized as “High-risk” was 2.8% which, according to the authors, is “similar to the premium [small or medium enterprise] SME segment at traditional banks”. In other words, with alternative data MercadoLibre identified companies at low risk of default that traditional credit bureaus would have rated negative. According to the authors, such “High-risk” borrowers make up 30% of MercadoLibre’s credit portfolio. Perhaps more interesting, MercadoLibre identified high-risk borrowers whom the credit bureau classified incorrectly as “Low-risk” or “Medium-risk”, as shown by the outliers in category “E” within the bureau’s “Low-risk” and “Medium-risk” ratings.

This week, The Information reported that Waymo has increased its headcount from 800 to 1,500 during the past year. Now that Larry Page and Sergei Brin have ceded their leaderships roles, ARK has been wondering if Alphabet will look at Waymo with a more critical eye and perhaps consider a spinoff. News of increased headcount seems to suggest that Waymo is doubling down on its efforts to scale an autonomous taxi service.

Waymo has struggled to launch a commercial autonomous service in Phoenix, perhaps because of a data problem. Compared to its database of 20 million real world miles driven since inception, Tesla has the option to draw from more than 20 million miles of driving data in a single day thanks to its customer fleet of more than 700,000 Autopilot-enabled cars. Waymo’s prototype fleet might have more autonomous functionality than any other vehicles on public roads today, but without enough data to solve for full autonomy, it could be lapped by the competition.

For decades, startups have been funded almost exclusively by venture capital. As the startup ecosystem matures, however, debt could become a viable alternative for mid to late stage funding, argues Alex Danco.

While the debt market has ignored startups because of their unpredictable businesses and the lack of collateral, Software-as-a-Service (SaaS) companies are different. Individual SaaS customers may come and go but, as annual cohorts, their churn and cash generation profiles are fairly predictable. At the same time, the go-to-market strategy of SaaS companies typically relies heavily on sales and marketing which delivers recurring revenues. The upfront cash necessary to acquire predictable streams of future revenue is suitable to debt instead of equity financing. A SaaS startup should be able to issue a “2022 cohort customer collateralized bond” instead of raising equity to fuel growth for the next one or two years.

Whether debt will fund startup SaaS companies remains to be seen. If it does, late stage venture, already competing with Softbank and private equity, will face even greater pressures.

As fear about the coronavirus’s impact on the economy escalates, investors have been quick to assess bitcoin as a “digital gold, safe haven” asset.  On January 30th, The World Health Organization (WHO) declared the coronavirus outbreak a global emergency after the number of cases had spiked by an order of magnitude and the death toll had increased to 213.

Since the WHO declaration, bitcoin’s (BTC’s) price has dropped 8.5% while price of gold has appreciated 1%. Many are wondering why the price of “digital gold” did not mirror that of physical gold. As Coinmetrics notes, “While market participants are quick to point out instances where BTC’s correlation with gold is high, such as during the summer in 2019, much less attention is paid to instances where BTC should react to safe haven capital flows but doesn’t.”

Assessing asset class returns on a slightly longer time horizon, however, paints a different picture. Since China reported the first case of coronavirus on December 30^th, bitcoin has outperformed every major asset class, including gold. Bitcoin and gold have appreciated 23% and 4%, respectively, while the S&P 500 has dropped 9.5%. Over the longer-term time period, XBT’s appreciation seems to illustrate a lack of correlation to the returns of most traditional asset classes.

A subsidiary of China’s Beijing Genomics Institute (BGI), MGI recently unveiled an ‘extreme throughput’ sequencing platform ostensibly enabling a $100 human genome. In our view, Illumina opened itself up to this competitive incursion because its HiSeq X system, after enabling the $1,000 genome in 2014, has not dropped prices significantly in more than five years. ARK believes that Illumina has not cut prices because it has enjoyed an effective monopoly, until now perhaps.

While impressive technologically, MGI’s newest instrument, the DNBSeq Tx, seems too complicated and is unlikely to commercialize widely. Unlike Illumina’s self-contained sequencers, the Tx platform involves a large robotic workspace, as shown below. Like other sequencers, it achieves its lowest cost per genome only when running at full capacity. To hit the $100 genome, the Tx requires $70,000 – or 700 samples – to run the platform once, limiting adoption to high throughput customers such as large population sequencing sites.

Despite its commercialization challenge, the Tx system does contain novel advancements, importantly CoolMPS chemistry which will integrate backwards with the rest of MGI’s instrument portfolio. Once fully integrated, CoolMPS should enable higher accuracy and improved operating costs.

MGI’s second-highest throughput instrument, the T7, purports to sequence a genome for as little as $500 and could intensify competition in the United States later this year, unless Illumina’s patent infringement suit stops it. According to MGI, even at 25% capacity utilization, the T7 can sequence genomes for $500. Goaded by prospective competition, ARK believes that Illumina will upgrade the NovaSeq with super-resolution optics and its new two-color chemistry, aiming for a $100 genome and ultimately benefiting from the massive latent demand unlocked by lower sequencing costs.