ARK Disrupt Issue 137: Autonomous Tech, EVs, AI, and CAR-T

Please enjoy ARK Disrupt Issue 137. This blog series is based on ARK Brainstorming, a weekly discussion between our CEO, Director of Research, thematic analysts, ARK’s theme developers, thought leaders, and investors. It is designed to present you with the most recent innovation takeaways and to keep you engaged in an ongoing discussion on investing in disruptive innovation. To read the previous issue, click here.

1. Tesla’s Autopilot Blows Competition Out of the Water

Follow @tashaARK on Twitter

In a recent independent study, Tesla’sTSLA latest generation Autopilot performed much better than its peers. The Insurance Institute for Highway Safety (IIHS) conducted a series of tests comparing a 2016 Tesla Model S (with first generation Autopilot technology), a Tesla Model 3 (with second generation Autopilot), a VolvoGEELYF S90, a BMWBMW.DE 5 series, and a Mercedes-BenzDAI E-Class. Important to note, Tesla’s first generation of Autopilot used Mobileye’sNVDA camera system.

Lane-keeping was the most basic and important among the tests. As shown below, the Tesla Model 3 with second generation Autopilot hardware outperformed its peers by a landslide. The second best performing vehicle was the Mercedes while the BMW failed miserably, straying from its lane or disengaging more than 90% of the time.

Bolstering these results during the next few years will be Tesla’s massive autonomous data advantage. Tesla is the only automaker that can send data to and collect data from its customer fleet with over-the-air software updates. Currently, state laws and/or independent dealer networks prevent Tesla’s competition from doing the same. Until legal or structural impediments that are impeding the progress of traditional auto manufacturers change, the Tesla Model 3 will be the only vehicle in its category able to improve its performance over-the-air over time.

ARK Disrupt Issue 137 Graph 1

2. Do Global Auto Executives Really Believe that Fuel Cells Will Be the Key Trend?

Follow @skorusARK on Twitter

Each year KPMG surveys auto executives, asking them to rank the importance of various industry trends. The chart below shows the percent ranking trends as “extremely important.”

ARK Disrupt Issue 137 Graph 2

Source: KPMG

To ARK’s surprise, fuel cell electric mobility ranked the highest, while battery electric mobility, mobility-as-a-service, and autonomous vehicles ranked third, seventh, and eighth, respectively. ARK’s research suggests that the demand for hydrogen fuel cell vehicles is unlikely to scale fast enough to drive down costs and become price competitive with electric vehicles (EVs). Furthermore, hydrogen fuel cell vehicles will require a more extensive buildout of re-fueling infrastructure than will electric vehicles which can charge at homes over-night. Moreover, because each refueling station would cost millions of dollars, an order of magnitude more than that of a supercharger, such a buildout would be cost- prohibitive.

3. OpenAI Trained a Dexterous Robotic Hand

Follow @jwangARK on Twitter

OpenAI—a nonprofit artificial intelligence (AI) research institute—has made remarkable progress in creating software that can control a robotic hand with human-like dexterity. While highly flexible, five fingered robotic hands have been available since 2005, programming them to perform reliably has been next to impossible.

To solve the control problem without programming, OpenAI turned to deep reinforcement learning, a trial and error approach to robotics control. Simulating 100 years of experience in just 50 hours, computers trained a 3D model of the robotic hand in a virtual environment. By varying parameters such as friction and gravity, the resulting program worked well not only in simulation, but also in the physical world. When given a physical cube, the robotic hand was able to manipulate it correctly 12 times before making a mistake.

The combination of deep learning and digital simulation is turbocharging progress in robotics control, perhaps heralding a golden age for robotics.

4. CMS Sets Pricing Guidelines for CAR-T Therapy

Follow @msamyARK on Twitter

Chimeric Antigen Receptor T-cell (CAR-T) therapy is revolutionizing the treatment of metastatic cancer, with complete remissions in more than 50% of the patients who had failed all other lines of therapy and had no other options. Last year, the FDA approved the first two CAR-T therapies – Novartis’s Kymriah for pediatric acute lymphoblastic leukemia (ped-ALL) and Gilead’s Yescarta for aggressive non-Hodgkin’s Lymphoma (NHL).

This week, Medicare announced its final 2019 reimbursement decision for CAR-T. Prior to this decision, it had classified CAR-T under the Drug Related Group (DRG) code for autologous bone marrow transplants which qualify for $161,000 in reimbursements, well below Kymriah’s $475,000 list price and Yescarta’s $373,000. After this decision, CAR-T will qualify for an additional $186,500 in reimbursements, or $347,500 in total. While not quite up to Novartis’sNVS and Gilead’sGILD list prices, Medicare’s reimbursement is higher than most analysts expected.

As detailed in the announcement, Centers for Medicare & Medicaid Services (CMS) indicated that it will consider a value-based pricing system as it analyzes the claims data from CAR-T therapies. Early proposals indicate that CMS will not have to reimburse unless a patient lives beyond a certain period of time.

ARK's statements are not an endorsement of any company or a recommendation to buy, sell or hold any security. For a list of all purchases and sales made by ARK for client accounts during the past year that could be considered by the SEC as recommendations, click here. It should not be assumed that recommendations made in the future will be profitable or will equal the performance of the securities in this list. For full disclosures, click here.