View as web page here.
It's Sunday, August 18th, 2019. Please enjoy ARK's weekly newsletter curated by our thematic analysts and designed to keep you engaged with disruptive innovation.

AI Is Accelerating Our Understanding of Genetic Disease

Follow Simon on Twitter @sbarnettARK


According to ARK’s research, the clinical adoption of next-generation DNA sequencing (NGS) should increase 40-fold during the next five years as the number of whole human genomes sequenced annually increases from roughly 2.4 million last year to 100 million by 2023. While costs still are the primary obstacle to widespread adoption, knowledge - or the lack thereof - is another. Diagnostic companies often surface variants of unknown significance (VUS) in genomic data, or mutations not yet linked to disease.


As the research community links genetic mutations to diseases, NGS-powered diagnostics and therapeutics based on gene-editing should proliferate as insurers become more willing to reimburse them at scale. Thanks primarily to software innovation, the pace of discovery and classification of new variants appears to be accelerating. The open-source AI-engines listed below are sorting through sequencing data more effectively and discovering the mutations most likely to cause disease. In turn, researchers are focusing their budgets on the most promising candidates. Further improvements should accelerate the clinical adoption of NGS and surface new targets for gene-editing companies that are developing novel therapeutics.


SNL 08.18.19 4-1




What Is BERT and Why Does Nvidia Keep Talking About It?

Follow James on Twitter @jwangARK


During the last two earnings calls, Nvidia CEO Jensen Huang has highlighted that BERT, a neural network, will have a significant impact on GPU sales. What is BERT, and why are Nvidia, Google, Microsoft, and Facebook all focused on that particular neural network?


In 2017, Google released an influential paper called Attention Is All You Need showing how a neural network based on “transformers” handily outperformed recurrent neural networks (RNNs), the industry standard at the time. A year later, Google released BERT, an improved transformer model that took top place in both SQuAD and GLUE, two important benchmarks for natural language understanding. Since then, Microsoft, Facebook, and Nvidia have designed and improved upon the performance of their own variants of BERT.


BERT is exciting for a number of reasons. First, its performance is astounding, leaving behind entire classes of neural networks like CNNs and RNNs. For the first time in history, in both the SQuAD and GLUE benchmarks BERT networks outperformed the human baseline. Deployed in commercial products like Alexa, Siri, and Google Assistant, BERT actually could make digital assistants smart.


Second, BERT’s training regime is far more efficient than today’s supervised learning. Today’s AI systems learn primarily from carefully crafted data sets that must be labeled by humans. BERT learns in an unsupervised fashion by ingesting data from books and sites like Wikipedia. Learning from unlabeled data and transferring that knowledge to a variety of tasks, BERT networks are more scalable and versatile than previous language models.


Third, though BERT is not the largest transformer model, its cousin - GPT-2 from OpenAI - is notoriously large and difficult to train. Both BERT and GPT-2 are large language models with an unlimited appetite for data and computing power.


No wonder Nvidia keeps talking about BERT—it has an insatiable appetite for graphics processing units (GPUs), Nvidia’s base business.


[all performance claims in previous paragraphs are referencing the BERT paper linked above]



What Is the Optimal Way to Denominate Cryptoasset Prices?

Follow Yassine on Twitter @yassineARK


Recent price moves in the crypto markets have raised questions about the optimal way to price cryptoassets. As we discussed in ARK Disrupt Issue 173, during the month of May the top 10 coins by network value underperformed bitcoin (BTC) by 30%. Since then, while their network values have appreciated significantly in USD terms, their prices against bitcoin have continued to unwind.


At current USD price levels, every coin in the top 10 is at a new low in terms of BTC since the spike in late 2017. Notably, ether (ETH) is down 77%, litecoin (LTC) 90%, and ripple (XRP) 67% from their all-time highs against BTC.


This week, ETH has been trading around $200, still well above its recent low as measured in USD but at a new low relative to BTC compared to the three most recent times it has traded at this price, as shown in the graph below. The first time it hit this range in mid 2017, ETH/BTC hit .091, followed by .081 a few months later, and then .034 towards the end of August 2018. Today, ETH/BTC is trading at .0185.


ARK SNL 08.18.19


Three Arrows Capital Founder Su Zhu first postulated that the number of coins that will hit all-time highs against USD probably will exceed the number that hit all-time highs against BTC. Because bitcoin appears to have evolved into the reserve currency of the cryptoasset ecosystem, measuring the performance of cryptoassets probably will continue to shift away from a USD benchmark toward BTC.



Square Could Turn Its Merchants into ATMs

Follow Max on Twitter @mfriedrichARK


In July, Square received a patent for a technology that would enable Cash App users to withdraw cash at nearby Square merchants. Cash App users already can use their Cash Cards, through Cash App debit cards, to withdraw cash from a network of ATMs, for which Square charges $2 per use. If a user receives $50 worth of direct deposits every month, however, Cash App scraps the fee. Now with the patented technology, Cash App users will need only a phone, not a physical card, to make withdrawals.


With this patented technology, Square will provide users with a list of withdrawal locations, either Square merchants or actual ATMs, as shown below. In some cases, Square will earn transaction fees for withdrawals, another source of monetization for the Cash App. We believe if merchants also use Square bank accounts integrated into the Square Card, the friction associated with fund transfers will drop precipitously, suggesting that withdrawals could be instant and nearly free. Square also could incentivize merchants to offer cash withdrawals with free credit card processing dollars, a common and low-cost Square marketing incentive.


From a high-level perspective, this technology should give Square a powerful opportunity to bring its consumer and seller ecosystems together. While not as revolutionary as some of its other innovations, Square could increase the overall engagement of Cash App users and, at the same time, boost the business of its merchants.


SNL 08.18.19 3






Yes, Snap is Still a Camera Company

Follow Nick on Twitter @GrousARK


SNL 08.18.19 1

(From the Creators of Snapchat – Spectacles 3)


This week Snap announced the release of Spectacles 3 – the company’s 3rd generation smart-glasses. Fashion aside, Spectacles 3 offers great improvements over the previous two versions. That said, we are not quite sold on them.


Snap’s glasses will come equipped with 2-HD cameras that will allow users to capture “depth and perception the way your eyes do,” as shown in the link below. Users also will be able to overlay augmented reality (AR) filters on all videos and pictures taken with the glasses, a signature Snapchat feature that neither of the first two versions offered.


Despite all the hardware improvements, we are skeptical that the glasses will sell well. After the initial launch of Spectacles, Snap had to write down roughly $40 million in inventory after sales didn’t gained traction. Particularly concerning this time around is the hefty $380 price tag: Spectacles 3 is selling at more than twice the price of the first generation.


Whether or not the glasses take off, Spectacle 3’s hardware upgrades suggest that Snap is focused intensely on augmented reality (AR), which is not surprising. Snapchat already relies heavily on AR filters and advertising to capture new users.  What is new is its AR friendly hardware, suggesting that it is doubling down on that effort.




Wright’s Law Works for Mature Technologies Too

Follow Sam on Twitter @skorusARK


Last week ARK wrote about Wright’s Law and theorized that Tesla could increase its gross margin on the Model 3 to more than 30% by the end of 2020. To recap, Wright’s Law states that for every cumulative doubling of production, costs fall by a fixed percent. For the auto industry, that percent has been roughly 15%.


“Cumulative production” is a key tenet of Wright’s Law, explaining why costs do not drop to $0 in a mature technology. As shown below, the cumulative production of internal combustion engine (ICE) powered vehicles has been roughly 2.5 billion, compared to just 3.4 million for electric vehicles (EVs). As a result, even if global auto production were to stabilize at ~90 million vehicles annually, a cumulative doubling would take 29 years.


Given the ramp in production and the smaller production base for EVs, a cumulative doubling should occur in little more than a year. Wright’s Law applies to both technologies but the difference in time to achieve a cumulative doubling implies the average ICE vehicle will decline in cost by ~0.5%, 24 times slower than the ~12% annual cost decline for an EV.


Stay tuned for a blog that goes deeper into Wright’s Law as it applies to the auto industry.


SNL 08.18.19 2

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.



3 East 28th Street, Floor 7
New York, NY 10016 United States

You received this email because you are subscribed to Research Newsletters from ARK Investment Management LLC.
Unsubscribe from ARK emails or choose the types of emails you want to receive. Unsubscribe from all.

This Newsletter is for informational purposes only and does not constitute, either explicitly or implicitly, any provision of services or products by ARK Investment Management LLC (“ARK”). Investors should determine for themselves whether a particular service or product is suitable for their investment needs or should seek such professional advice for their particular situation. All content is original and has been researched and produced by ARK unless otherwise stated therein. No part of the content may be reproduced in any form, or referred to in any other publication, without the express written permission of ARK. All statements made regarding companies, securities or other financial information contained in the content or articles relating to ARK are strictly beliefs and points of view held by ARK and are not endorsements of any company or security or recommendations to buy or sell any security. By visiting and/or otherwise using the ARK website in any way, you indicate that you understand and accept the terms of use as set forth on the website and agree to be bound by them. If you do not agree to the terms of use of the website, please do no access the website or any pages thereof. Any descriptions of, references to, or links to other products, publications or services does not constitute an endorsement, authorization, sponsorship by or affiliation with ARK with respect to any linked site or its sponsor, unless expressly stated by ARK. Any such information, products or sites have not necessarily been reviewed by ARK and are provided or maintained by third parties over whom ARK exercises no control. ARK expressly disclaims any responsibility for the content, the accuracy of the information, and/or quality of products or services provided by or advertised on these third-party sites.