#428: Robotaxis Continue To Scale, & More

We’re excited to see Baidu and Waymo scaling their robotaxi operations.

Last week, Waymo confirmed that its sixth-generation robotaxi, developed by Geely’s Zeekr, will cut the number of onboard cameras from 29 to 13 and lidar sensors from five to four, compared to the hardware on Waymo’s Jaguar I-PACE.¹ Although Waymo has not disclosed details, the cost of the new vehicle is likely to be much lower than the rumored ~$100,000+ for the current model² —a critical step toward scaling operations.

Waymo recently announced that it has ramped the number of autonomous rides per week to more than 100,000.³ Recently, ARK’s Tasha Keeney tested Waymo’s service in LA and was impressed by the smooth ride.⁴

Separately, in its recent earnings call, Baidu highlighted that during the second quarter Apollo Go, its autonomous ride-hail platform, averaged ~75,000 rides per week.⁵ Notably, while Baidu now operates 100% autonomously in Wuhan with a small fleet of ~400 vehicles,⁶ its operations in other cities include both human-driven and autonomous vehicles.⁷ Despite Baidu’s progress, management noted that rolling out nationally and gaining a significant share of the ride-hail market could take years.

While Waymo and Baidu are first-to-market in robotaxi operations, as shown in the first chart below, our research suggests that Tesla is training its autonomous system with ~70X more data, as shown in the second chart below. As will be the case with most AI (artificial intelligence) projects, Tesla’s data scale advantage is likely to lead to commercial dominance in this “winner-takes-most" opportunity.

We look forward to monitoring this space, with keen interest in and special attention to Tesla’s robotaxi event in October.

Source: ARK Investment Management LLC, 2024, based on data from a range of external sources, as of August 23, 2024, which can be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

Source: ARK Investment Management LLC, 2024, based on data from a range of external sources, as of August 23, 2024, which can be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

Advances in high-throughput proteomics,⁸ which measures thousands of proteins from a single drop of blood, have revealed that proteins circulating in our blood can help detect, treat, and even prevent complex diseases like cancer, diabetes, and Alzheimer's. Using large datasets from studies like the UK Biobank, researchers have identified patterns that link proteins to health outcomes, importantly, how changes in protein levels are precursors to certain diseases.

Recently, three studies have demonstrated that high-throughput techniques analyzing proteins are enabling earlier diagnoses of disease and personalized treatments.

Researchers from the University of Edinburgh created ProteinScores that have outperformed traditional genetics risk markers, demonstrating that blood proteins can predict the onset of age-related diseases like diabetes, heart disease, and Alzheimer's—often before symptoms appear.⁹

Researchers from the University of Oxford studied the Proteomic Aging Clock and demonstrated that plasma proteins can predict both mortality and the risk of common age-related diseases. The study’s key finding—the "ProtAgeGap"—measures the difference between biological and chronological age: a higher ProtAgeGap is linked to increased risk of diseases like heart disease, diabetes, and Alzheimer’s, while a lower gap indicates slower aging and reduced risk. The distinction allows for earlier disease detection, timely intervention, and personalized healthcare.¹⁰

Researchers from the Icelandic Heart Association identified key protein markers linked to Alzheimer’s/dementia. Specifically, proteins impacted by the APOE-ε4 gene—a known genetic risk factor—are relevant for diagnosis and treatment, even for individuals without the APOE-ε4 gene.¹¹

These studies could reshape healthcare by enabling the early detection of diseases. Routine blood tests might identify health risks years in advance, enable early intervention, and delay or prevent disease progression.

Personalized medicine also should improve dramatically. Understanding individual protein profiles is likely to empower doctors to tailor treatments more effectively, evolving healthcare practices that are more proactive in response to routine diagnostics that predict disease and inform treatments long before symptoms surface.

While discussions about automation often focus on lost jobs and other negative financial impacts on employees, our research suggests that the productivity boost associated with automation could lead to higher wages. When automating, companies might allocate more revenue to capital investments than to labor, reducing “labor share”— the share of revenue going to labor. The labor share in large companies–presumably with the wherewithal to automate—tends to be lower than that in smaller companies, as shown in the chart on the left, but the productivity gains associated with automation seem to result in higher average pay per employee, as shown in the chart on the right.

Source (left): ARK Investment Management LLC, 2024, based on data from IZA Institute of Economics 2019¹² as of August 23, 2024. Source (right): ARK Investment Management LLC, 2024, based on data from United State Census Bureau 2023¹³ as of August 23, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.