Please enjoy ARK Disrupt Issue 69. 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.
Disruptive Chart of the Week: The Death of TV
1. No Bank Account or Credit Card?… No Problem! You Can Shop with Cash on Amazon
AmazonAMZN is targeting millions of unbanked people by allowing them to shop on Amazon using cash. The new service, “Amazon Cash,” allows customers to add cash to their Amazon balances by showing a barcode at a participating retailer, then having the cash applied immediately to their online Amazon accounts.
At first blush, Amazon Cash appears similar to buying a gift card; however, it has significant differences and benefits. Amazon Cash not only ensures immediate availability of funds in an Amazon account but also eliminates the tedious process of loading a gift card into an Amazon account. Most importantly, the product is entirely mobile based, making it simple and more attractive to millennials. We believe this is an interesting experiment to reach new demographics outside of the traditional payments ecosystem, which could be a precursor to Amazon’s success in emerging markets like Mexico and India, where cash is used predominantly.
2. Adidas Turns to Carbon 3D
Adidas plans to use Carbon 3D’s machines to 3D print 100,000 pairs of midsoles for its Futurecraft 4D shoe by 2018. Traditional manufacturing uses metal molds to create midsoles, which can take a month to build and can be expensive as molds can require over 10,000 uses before the manufacturer breaks even on the investment. Carbon 3D, however, can 3D print midsoles in 90 minutes, and plans on reducing that to 20 minutes. Carbon 3D’s CLIP technology allows Adidas to create more flexible midsoles than previous 3D printing methods, as well as different zones of cushioning and support for feet in one print. Traditional manufacturing uses different densities of foam that must be glued together. While Adidas has not given specific prices for Futurecraft 4D, the shoes are expected to be at premium price point.
Futurecraft3D, Adidas’ previous generation of Futurecraft shoes, had 3D printed midsoles that were printed in partnership with MaterialiseMTLS. The shoes cost $330 and printing time was about 10 hours. It appears Carbon 3D could have edged Materialise out of its relationship with Adidas.
Eventually Adidas hopes to create 3D printed shoes customized to each users’ weight, gait, and sports type. ARK believes this could create customer stickiness, as shoppers may not want to go through the fuss of rescanning their feet at a competitor’s shoe store once Adidas has their measurements. This customization process will also arm Adidas with valuable customer data to design new products and new materials with Carbon 3D.
3. 23andMe is Back in the Game
The U.S. Food and Drug Administration (FDA) announced on Thursday that it would allow 23andMe to market Personal Genome Service Genetic Health Risk (GHR) tests. Such tests will allow the company to provide genetic health risk information about 10 conditions:
“Parkinson’s disease; late-onset Alzheimer’s disease; celiac disease; alpha-1 antitrypsin deficiency, a disorder that raises the risk of lung and liver disease; early-onset primary dystonia, a movement disorder; factor XI deficiency, a blood clotting disorder; Gaucher disease type 1, an organ and tissue disorder; glucose-6-phosphate dehydrogenase deficiency, a red blood cell condition; hereditary hemochromatosis, an iron overload disorder; and hereditary thrombophilia, a blood clot disorder.”
These tests are not diagnostic-tests, but provide information about an individual’s genetic predisposition to a disease or condition.
The tests are being sold directly to consumers, and will be “the first direct-to-consumer (DTC) tests authorized by the FDA that provide information on an individual’s genetic predisposition to certain medical diseases or conditions.” Notably, 23andMe was told by the FDA in 2013 to stop marketing these results as the FDA said they “could be inaccurate and risky to consumers.”
The technology underlying the tests is not based on next-generation sequencing, but isolates DNA from a saliva sample, testing for more than 500,000 variants. Despite this, the news is still a boon for sequencing company IlluminaILMN, since many of the direct-to-consumer tests in the market run on its equipment. Moreover, the diagnostics sector overall stands to benefit as the FDA direction provides a roadmap for regulation of direct-to-consumer tests. The FDA even noted, “GHR tests from other makers may be exempt after submitting their first premarket notification. A proposed exemption of this kind would allow other, similar tests to enter the market as quickly as possible and in the least burdensome way, after a one-time FDA review.” We believe the FDA’s announcement is a big step forward for consumer-focused DNA testing.
4. Google’s First In-House Chip Crushes AI Benchmarks
This week GoogleGOOG released the first performance numbers for its Tensor Processing Unit (TPU), a chip built by the internet giant specifically for deep learning applications. The results are impressive. By our estimates, Google’s TPU is 57x faster per watt than NVIDIA’sNVDA K40, a GPU built using the same 28nm process. NVIDIA, however, is not standing still: the latest generation of Pascal GPUs is built on the more advanced 16nm process. New instructions were added to speed up deep learning operations. Taken together, performance has improved about 10x.
But Google isn’t done either. The company is likely working to bring its TPU to the 16nm node. Assuming it can double the chip’s performance (not an aggressive target), a gen 2 TPU would still outperform NVIDIA’s comparable 16nm Tesla P4 accelerator by 8x.
Google’s mastery of chip design for AI has many implications. It could give the company a competitive edge in internet services, cloud computing, and even autonomous driving. Apple’s success rests in large part to integrating hardware (including chip design) with software and services. Google may end up doing the same for the cloud and the car.
5. Industrial Robot Cost Decline
By 2025 industrial robots could cost roughly half of what they cost today. In fact, ARK’s research demonstrates that industrial robots are declining in cost faster than BCG has forecast.
Source: ARK Investment Management LLC, BCG, International Federation of Robotics
ARK uses Wright’s law to model cost decline, which states that for a cumulative doubling in production, price falls by a certain percentage. Interestingly, BCG is attributed with coining the idea of the “experience curve,” which is a rebranding of Wright’s law. Yet, BCG has the cost of robots remaining relatively flat for the next eight years. A common mistake when extrapolating cost declines is the assumption that sales will remain stable as the cost comes down. In reality cost declines often increase future sales by expanding the market, and thus increase the speed at which another cumulative doubling in production occurs.
6. CRISPR Debunks Studies Based on Older Gene-Editing Tools
What if you were told that a number of successful drugs on the market were developed by luck, and that while based on faulty scientific analysis, they worked because of some other biological phenomenon? Scary, but true.
Nature this week released some eye-opening studies in which scientists used CRISPR-Cas9 to interrogate the function of a well-characterized cancer-mediating gene, MELK, and were unable to reproduce the data. As it turns out, while silencing that gene was helpful in cancer remission, its mechanism of action was quite different than researchers expected. Such discrepancies undermine the scientific foundation of other clinical trials associated with MELK.
While some cell types respond better to gene suppression, others respond better to gene-edits. While CRISPR is generally used for its editing function, CRISPRi will allow for gene-silencing — that is, it will simply stop a protein from forming without making edits.
The larger takeaway, however, is that data on previously characterized genes may be wrong. Luckily, CRISPR technology is cheap and fast enough to redo these screens and insure that future therapeutic trials are based on sound science instead of pure luck.
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.