#222: Bricks and Mortar Retail May Have Peaked in 2015, Batteries, & More

Shattering records, retail sales fell 16.4% from March to April, including an unbelievable 80% drop in clothing and accessories store sales. Previously, the worst month on record was during the Global Financial Crisis in 2008 when retail sales dropped 3.9% from October to November 2008.

Meanwhile, while e-commerce surged nearly 50% in April, according to Adobe, we believe the peak in bricks and mortar sales actually took place nearly five years ago. Now COVID-19 appears to have accelerated e-commerce adoption, perhaps compressing what would have been three years of growth into months. ARK estimates that e-commerce will increase from 15% of US retail sales in 2019 to 40% in 2025, putting at risk trillions of dollars of retail real estate and $1.6 trillion of retail enterprise value.

On Thursday, Reuters published a piece highlighting its expectations for Tesla’s Battery Day, originally scheduled for April and then May. The most notable rumor is that Tesla could introduce a Model 3 in China at the same sticker price as comparable gas-powered cars this year, or early next year, a year or two earlier than we have expected.

To achieve price parity, Tesla supposedly will use cobalt-free lithium iron phosphate batteries for which China’s CATL (Contemporary Amperex Technology) will charge less than $60/kWh. At that cost, ARK estimates that a Model 3 with a 75kWh battery pack could sell for roughly $23,000.  In ARK’s latest forecast, EVs hit price parity with comparable gas-powered vehicles in 2022, so Tesla’s battery breakthroughs, if true, will accelerate the timeline by one to two years.

Given its long-standing goal to export vehicles globally, China may have found the right partner in Tesla. Unfortunately, we will have to wait a bit longer to confirm or deny these rumors. Musk tweeted this week that Tesla has postponed Battery Day until June.

On Thursday Nvidia announced its new flagship A100 GPU, the first major update to its AI training processors since 2017. Manufactured on TSMC’s 7nm process, the A100 is the fastest, largest, and most power-hungry chip today. Beyond its increase in raw performance, the A100 has new features that make the gargantuan chip far more efficient for artificial intelligence (AI) workloads in the data center.

The A100 GPU supports a new virtualization technology that Nvidia calls Multi-Instance GPU (MIG). MIG allows the A100’s seven cores to run workloads for seven different users. Today, public cloud providers use multi-tenancy to increase the capacity utilization of their CPU servers. MIG turns the A100 into a multi-tenant GPU, making it especially attractive to cloud customers.

The A100 also includes new optimizations to eliminate redundant AI calculations. By ignoring zero or near zero values and applying compression, the A100 can perform inference up to twice as fast as Nvidia’s last generation chip. First championed by AI chip startup Cerebras, this feature is called “sparsity”. By adding it to the A100, we believe Nvidia is demonstrating once again that it can innovate and execute as quickly as nimble startups.

Having completed its acquisition of Mellanox within the last two months, Nvidia also debuted a new product that merges GPU and network functionality. The EGX A100 is a smart network card with an A100 GPU onboard, enabling network, security, and AI acceleration with minimal input from CPUs.

In the past three years, we believe AI chip startups have taken little share from Nvidia’s GPU business. With the A100, Nvidia may be even harder to catch.

According to a report by the Financial Times, UK-based challenger bank Monzo is raising a new round at a valuation of $1.5 billion, 33% less than the $2.3 billion in its last round nearly a year ago in June. Earlier this year, Berlin-based challenger bank N26 raised new funds at a flat valuation while UK-based Revolut did a round that tripled its valuation to $5.5 billion but was 45% below its target.

While the Financial Times believes the COVID-19 crisis has made venture capitalists wary of highly unprofitable companies, we believe the real reasons behind the drop in Monzo’s valuation is likely to be a slowdown in its customer acquisition and other key performance indicators. According to our calculations, Monzo’s user growth has dropped from more than 280,000 per month in the first half of 2019 to roughly 100,000 in the last few months before the coronavirus hit. In fact, commenting on Monzo’s down round, an industry insider said that “the time for grandiose PowerPoints was over and had been replaced with hard numbers in spreadsheets”.

Important to note, challenger bank ‘user numbers’ refer to those who have signed up, not to those actively using the application. Under the generous assumption that 50% of registered users still use Monzo on a monthly basis,¹ investors in its upcoming down round will be valuing Monzo at roughly $800 per active user, down more than 50% from $2000 in June 2019.

To put these numbers into perspective, if public market investors valued monthly active Cash App users² as highly as Monzo’s, Cash App would be worth roughly $23 billion, or 70% of Square’s market capitalization even though it accounts for only 16% of revenues. If its monthly active users were to have been valued at $2,000, Monzo’s peak valuation per user in 2019, Cash App would have been worth twice as much as Square’s current valuation. Instead, today Cash App users may be valued at only $225.³

¹ Venture Capitalists familiar with challenger bank KPIs have told ARK in the past that on average a third of challenger banks’ registered users can be considered active users. Chime, a US-based challenger bank, reported that half of its registered users use Chime for direct deposits.

² We believe that Cash App has surpassed the 30 million monthly active user mark on the back of strong user growth over the first four and a half months of 2020.

³ Assuming that public market investors don’t value Cash App at a premium to Square’s merchant business.

Notarizations are necessary in executing many business, legal, and financial documents, with online notarizations increasing in popularity. According to the National Notary Association, 24 states permit online notarizations. In response to social distancing amid the COVID-19 crisis, 18 new states have enacted emergency orders allowing notarizations online and probably will legalize them in the aftermath of the crisis. Innovation certainly seems to gain traction during tumultuous times.

Compared to traditional notarizations which take place in person, we believe digital notarizations can make important transactions not only more efficient and secure, but also less expensive. They should benefit online companies like Zillow, Redfin, Carvana, and DocuSign.

CAR-T cell therapy has demonstrated efficacy that some patients and their families have described as miraculous. The FDA has approved two therapies, but they not only are expensive, requiring a hospital stay, but also involve complex logistics and long manufacturing timelines.

Allogeneic, or donor-derived, CAR-T cell therapy could solve most of these problems. This week, Allogene Therapeutics (ALLO) provided evidence that allogeneic CAR-T cell therapy could be efficacious in treating patients with non-Hodgkin’s lymphoma. With preliminary data, it reported an overall response rate of 78% in nine patients, three complete responses, four partial responses, and three relapses within six months. We believe this data shows that allogeneic CAR-T cells could be manufactured and administered safely for anti-tumor activities. Important to note, Allogene Therapeutics licenses its allogenic CAR-T technology from and has a strategic collaboration with a company not as well known, Cellectis (CLLS).