1. A New Ridehailing Company May Turn Uber And Lyft Into Sitting Ducks
ARK hypothesized1 that once fully autonomous cars commercialize, Uber and Lyft will lose the 20-30% platform fees they take today. Indeed, they will be lucky to get 1-5% in lead generation fees as they are forced to partner with autonomous technology providers.
Not only in the long term but perhaps in the short term as well, autonomous technology companies seem well positioned to capture the lion’s share of the economics from Uber, Lyft, and other ridehailing companies. On its recent earnings call, Tesla reiterated that it plans to launch a ride hailing service with drivers as early as this year, before solving for full autonomy. If so, it will learn how to operate a ridehailing network and will collect much more data to train its Autopilot system in preparation for the launch of its fully autonomous taxi network. As it competes with Uber and Lyft, Tesla also will generate recurring cash flow from a new revenue line item on its income statement, mitigating the need to go back to the capital markets as it scales manufacturing capacity globally.
A Model 3 is likely to provide ridehailing drivers with superior economics compared to the widely used Toyota Camry. According to ARK’s analysis, thanks to lower maintenance and fuel costs, the Model 3’s cost per mile driven will be roughly one third lower than that of the Camry. Not included in this analysis are insurance costs: Tesla probably will self-insure, offering lower rates for drivers activating Autopilot on a regular basis. Based on all of these lower costs, not only should drivers on the Tesla Network take home more pay than with Uber, but Tesla also may enjoy a higher take rate than Uber’s 23% average today. This potential win-win could become a triple win perhaps as ridehailing customers choose what we believe is the more appealing car. In fact, Tesla might be able to charge UberBlack rates on the Tesla Network as it adds a recurring revenue stream to each Model 3 sold.
The value proposition of Tesla’s ridehailing network is likely to hit Uber and Lyft hard. Competition in the ridehailing market is already fierce, and Tesla is about to turn up the heat!
[1] Please refer to pages 40 – 45 of ARK’s Big Ideas 2020
2. 2020 Could Be the Year That Robots Pick and Place at the Same Rate as Humans
Amazon held its first robot pick and place challenge in 2015. As shown below, that year the winning team picked and placed the equivalent of 30 items per hour, paling in comparison to the human average of 400 items. Then, in the three years ended 2018, robot picking and placing improved more than eight-fold to 250 items per hour.
This year, German electrical supplier Obeta reportedly introduced robots that outperform humans, thanks to AI robotics company Covariant. Though not completely comparable to items picked and placed per robot per hour, Obeta claims that a Covariant-powered robot can fill more than 200 orders per hour, surpassing the 170 average of humans. Perhaps most impressive is the AI system’s learning rate. After training for five months, the percent of Obeta’s products that the robots picked successfully improved from just 15% to roughly 95%.
3. A Clever Flash Loan Has Exploited “DeFi” Protocol bZx
bZx, a decentralized finance (DeFi) protocol for tokenized margin trading and lending, was exploited by a flash loan. A flash loan allows an individual to borrow capital without collateral, as long as he returns it in a single transaction. In other words, in seconds users can tap significant capital and carry out an attack at no cost. While not the enablers of the attack, flash loans have raised questions about the DeFi’s sustainability.
Blockchain security company PeckShield detailed the attack in 5 steps:
- Attacker takes advantage of the dYdX flash loan feature to borrow 10,000 ETH.
- Attacker deposits 5500 ETH into Compound as collateral to borrow 112 WBTC.
- Attacker deposits 1300 ETH into bZx to take advantage of the margin trade feature and to short ETH in favor of WBTC. bZx calls KyberSwap to swap the borrowed ETH for WBTC in return. Consulting reserves, Kyberswap finds the best rate for swap on Uniswap, artificially driving WBTC price up on Uniswap.
- With inflated WBTC price, attacker sells 112 WBTC borrowed from Compound for WETH in Uniswap, netting ETH in return.
- With the ETH, attacker repays the flash loan to dYdX, pocketing the remaining ETH.
A full breakdown of the attack can be found here.
As Block journalist Celia Wan notes, “The recent attacks on DeFi lending protocol bZx, enabled by flash loans, serve as a wake-up call to other DeFi projects that may have been underestimating their adversaries.”
4. ‘UNCALLED’ Enhances the Power of Nanopore Sequencing Instruments
Oxford Nanopore Technologies (“ONT”) recently announced several hardware and software upgrades that ARK believes will make its sequencers more attractive to clinical diagnostic providers. Previously, diagnostics companies shunned ONT instruments because of their inferior accuracy, frequent update requirements, and lower sample throughput. We summarized several of its improvements in a recent newsletter. Now, two academic groups have published papers highlighting the benefits of ONT’s system upgrades and the advantages of sequencing with nanopore technology.
In clinical settings, typically diagnosticians prepare DNA samples by making many copies of genetic regions of interest prior to sequencing, the most expensive part of the DNA sequencing process. The chemicals used in this step can cause errors or destroy portions of the genetic material. One academic group built UNCALLED, an open-source software tool that eliminates the need to pre-select regions of interest using chemicals. UNCALLED takes advantage of ONT’s ‘read-until’ API, allowing researchers to enter digitally the parts of the genome in which they are interested. As shown in this chart, a nanopore will sequence DNA completely only if it matches the region specified by the researcher or diagnostician. UNCALLED should allow for lower sample prep costs while preserving the fidelity of DNA samples.
While this improvement increases the likelihood of clinical adoption, ONT has had a history of over-promising and under-delivering on an “Illumina-killer” for roughly a decade. If successful in any way, however, ONT will introduce more competition into the sequencing space, perhaps giving Illumina (ILMN) an incentive to accelerate its own cost declines and maintain its competitive advantage.