Industrial Robot Sales Topped 400,000 Units in 2018 

Follow Sam on Twitter @skorusARK

The International Federation of Robotics (IFR) released its World Robotics report, highlighting that industrial robot sales increased 11% to 422,000 in 2018, as shown below.¹

For the first time, the IFR broke out collaborative robots, which now account for 14,000 units or 3.3% of all industrial robot sales. Collaborative robot sales grew 27% year-over-year, more than double the growth rate of total industrial robots, validating ARK’s thesis that they will continue to gain share of total robot sales and fuel their growth.

While both ARK and the IFR forecasts for industrial robots were similar and close to the mark for 2018, now they are diverging considerably. Based on the slowdown in global auto sales and the US-China trade conflict, the IFR expects 0% growth for 2019, followed by 12% growth at a compound annual rate to 584,000 units in 2022. In contrast, we believe that powered by faster than expected cost declines and the deployment of collaborative robots outside the auto industry, ARK forecasts industrial robot sales growth of 25% at an annual rate to roughly 1 million units, nearly twice as many as the IFR is projecting, in 2022.

[1] Certain of the statements contained in this item may be statements of future expectations and other forward-looking statements that are based on ARK’s current views and assumptions, and involve known and unknown risks and uncertainties that could cause actual results, performance or events to differ materially from those expressed or implied in such statements.

What Is the Optimal Fee - or Take Rate - for Online Platforms? 

Follow Tasha on Twitter @TashaARK

In 2013, Benchmark Capital’s Bill Gurley wrote a piece stating, “High volume combined with a modest take rate is the perfect formula for a true organic marketplace and a sustainable competitive advantage.” Today, however, some successful online aggregators appear to be taking surprisingly high platform fees, as shown below. While platform fees differ by industry and offering, based on the company models we have examined the average fee is roughly 30%.

Some companies like Uber enter a market offering low platform fees to acquire users, only to increase them over time. Yet, pushback from business owners, or drivers in the case of Uber, as well as regulators can put a ceiling on platform fees. Interestingly, while Uber’s take rate has been fairly constant over time, Lyft’s have increased.¹ That said, because both firms went public recently, they may have been putting their financial statements in the best light, obfuscating their equilibrium take-rates. ARK is debating the best take rate strategy to maximize unit volumes and profitability. Send us your thoughts!

[1] Sources: ARK Invest, Company Reports. Note Lyft stopped reporting gross revenues after its S-1, so take rates shown are latest available.

Facebook Acquires Mind Control Startup CTRL Labs 

Follow James on Twitter @jwangARK

Hailed several years ago as the successor technology to smartphones, virtual reality (VR) and augmented reality (AR) have failed to gain mainstream adoption. A key obstacle is the user interface: it’s not intuitive. While PCs have the mouse and keyboard and smartphones a multi-touch display, VR and AR headsets have no real input device. To address this issue, Facebook offered nearly $1 billion this week for CTRL Labs—a startup that has designed a wristband that lets users control devices with their minds.

CTRL-Labs’ wristband uses a technology known as differential electromyography, or EMG. EMG captures electrical signals as muscles begin to activate when an individual intends to move. Already in developers’ hands, the prototype can translate actual or intended hand motion into digital actions, such as jumping over obstacles and shooting at asteroids. CTRL uses artificial intelligence to learn how different user intents map into unique electrical signals.

If successful, EMG-based control technology and Elon’s Musk’s Neuralink could close the gap between thought and input, paving the way for entirely new classes of computers that we can’t imagine today.

The Amazon Hardware Event Unveils Frames, Buds, Loop and More…

Follow Nick on Twitter @GrousARK

On Wednesday at its annual hardware event, Amazon announced more than 15 new devices. With products ranging from smart glasses to a 4-in-1 smart oven, it did not disappoint.

While most of the use cases centered on the home, Alexa - Amazon’s voice assistant – arguably made the most notable leap, outside of the home. The initial announcements featured Amazon’s highly anticipated headphones: with a price tag of $129.99, Bose noise-canceling technology, and Alexa integration, its new Echo Buds now seem to be a formidable competitor to Apple’s AirPods. In a Steve Jobs-like “one more thing” moment, however, Amazon one-upped the earbuds announcement with two new products that no one expected: Echo Frames (Glasses) & Echo Loop (Ring). Through Alexa, they will allow users to stay connected to Amazon on the go, enabling it to recapture mobile users, an opportunity the Fire Phone missed.

Given Amazon’s success in the smart home market, perhaps competitors like Apple should be concerned. Amazon now owns a majority share of the smart speaker market and could penetrate 50% of US households by mid-2020. It also is enjoying success in other areas like home security (Ring) and entertainment (Fire TV). Amazon’s apparent dominance in the home bodes well for its new wearables business with Alexa the hub for all of its devices.

Amazon offered more perspective on Alexa’s success at Wednesday’s event. It now offers 100,000 skills, apps designed to work with the voice assistant, and is compatible with roughly 85,000 different devices.

How Will the SAFE Banking Act Impact Financial Services?

Follow George on Twitter @GeorgeOfARK

Marijuana sales are legal in some form in 33 states but most if not all marijuana-based businesses are not able to access basic financial services because the Drug Enforcement Administration (DEA) classifies marijuana as a Schedule I drug. This classification prevents banks, processors and sometimes local businesses such as plumbers and electricians from working with and in marijuana-based businesses, forcing them into cash-based transactions as a means of payment should they decide to do so. Last year, marijuana sales in the US were $11 billion, with estimates in the $30 billion range for 2024.

On Wednesday, the House of Representatives voted on and passed the Secure and Fair Enforcement (SAFE) Banking Act by a vote of 321 - 103. The bill now is heading to the Senate where it faces an uncertain future. If this bill were to pass in the Senate, the financial services industry would be able to serve the marijuana industry without violating any federal laws, likely expanding the market beyond $30 billion by 2024.

ARK's finch research has focused importantly on the un(der)banked population in the US. Rarely have we found an industry in the developed world that the financial services industry appears to have ignored entirely. We believe that cash based businesses are inherently inefficient. For example, they have to hire security and/or store physical dollars in safes and can’t use traditional payroll services. If the Senate were to pass SAFE, companies like Square, PayPal, and Intuit could benefit from a new market opportunity. As a point of reference, last year Square processed nearly $85 billion, so even capturing 10% of a potential $30+ billion market would have a meaningful impact.

Bitcoin’s Hash Rate Dropped 40%. Or Did It?

Follow Yassine on Twitter @yassineARK

Bitcoin’s 20% price sell-off is breeding fear, uncertainty, and doubt (FUD) once again. This week, news broke that Bitcoin’s hash rate had dropped more than 40%, raising the profile once again of blockchain’s “fundamental technical issues”.

Upon further analysis, the precipitous drop in Bitcoin’s hash rate seems to have been misinterpreted. An understanding of the hash rate calculation suggests why.

A common misconception is that the hash rate can be calculated with precision at any point in time. In reality, Bitcoin’s hash rate is an estimation based both on the mining difficulty and the mining frequency of previous blocks. Knowing the difficulty, one can estimate the hash rate needed to mine a certain number of blocks during a period of time.

Additionally, block times themselves are probabilistic and prone to a high degree of variance. Below is a graph courtesy of Nic Carter illustrating the time between blocks mined over the last thousand blocks. The outliers are notable.

This week, for example, 5 blocks took longer than 50 minutes to mine. Estimating the hash rate based on those 5 blocks would have suggested a significant drop in hash rate. While unusual, a high variance in block times is possible given their probabilistic occurrence.

As Jameson Lopp states, “It was just regular random fluctuations in block times. The longer time period over which you estimate hash rate, the more accurate your estimate is likely to be... and vice versa.”

The day after the precipitous drop, the estimated hash rate surged to all-time highs, another example of the skewing associated with a single day’s worth of data. One day does not a trend make.

Why Is There a Gap Between Illumina’s Genome List Price and the National Institutes of Health?

Follow Simon on Twitter @sbarnettARK

Over the past decade, Illumina (ILMN) has pushed the cost to sequence a whole human genome down from $100,000 to less than $1,000. Its goal is to drop the cost to $100 per genome in the early 2020’s. 

A recent update from the National Institutes of Health (NIH), however, suggests that in most high-throughput sequencing centers the cost is roughly $1,500 per genome in 2018. As a result, some analysts are wondering if Illumina is struggling to push costs down or if it is taking advantage of its dominant position in the market and holding prices higher than otherwise would be the case. 

While competition could lower prices more aggressively, ARK believes that currently Illumina is aiming to increase its platform utilization and guide budget-constrained research and academic users gradually to system upgrade cycles. Because many research labs do not have the budget to adopt NovaSeq, Illumina’s most efficient sequencing platform, a gap has opened up between Illumina’s list price of $1,000 per genome and the NIH’s realized price of roughly $1,500. Established labs are operating with older, less sophisticated workflows, elevating sequencing costs. Meanwhile, large clinical customers like Invitae (NVTA) are innovating on top of Illumina equipment, pushing sequencing costs below $1,000. With hardware upgrades, leading edge clinical customers could drive the cost per genome below $100 in the not too distant future, pushing the majority of Illumina’s installed base even farther behind the pace of innovation. Consequently, in recent quarters Illumina seems to have been prioritizing the conversion of a broader swath of existing customers to more sophisticated and productive sequencers by discontinuing the HiSeq X and discounting NovaSeq flow cells.

Illumina’s focus on system upgrades and consumables could be a function of the low capacity utilization of sequencing machines globally. Currently, at 80% utilization Illumina’s installed base could sequence roughly 12 million genomes per year,  five+ times more than ARK’s estimates for 2018 and corroborated by Illumina’s published consumables revenue. Starting from a system-wide utilization rate of just 21%, as its capacity utilization increases Illumina should enjoy a mix shift toward higher consumables as well as lower SG&A spending as a percent of sales. 

Illumina’s focus on platform utilization and consumables growth should accelerate revenue growth and increase operating margins. That said, to scale from roughly $4 billion in revenues to an estimated $20 billion during the next four to five years, Illumina will have to stimulate the clinical adoption of next-generation DNA sequencing. While increasing both utilization and system conversion might be a short-term growth strategy, in our view Illumina’s long-term success will depend on the commercialization of a $100 genome sequence and the exponential ramp of clinical sequencing. 

A New CRISPR Class Opens Up New Doors to Treat Diseases

Follow Manisha on Twitter @msamyARK

Investors increasingly understand that CRISPR is presenting transformational opportunities to treat, if not cure, disease. Most do not know, however, that Class 2 CRISPR systems such as Cas9, CPF1, and Cas12 account for only 10% of all CRISPR systems in nature: 90% are Class 1 about which researchers have gleaned very little, until now. 

Researchers at Duke University have demonstrated that Class 1 CRISPR–Cas3 systems can be expressed in mammalian cells and used to target DNA and control transcription. They repurposed type I variants of Class 1 CRISPR–Cas systems from Escherichia coli and Listeria monocytogenes which target DNA by way of an RNA-guided complex called Cascade. By tethering Cascade to activation and repression domains, like a dimmer switch, they have been able to modulate the expression of targeted genes in human cells. 

In other words, instead of turning a gene on or off completely, researchers now can control the expression of a protein-coding gene, both the time and the amount, to control disease. Because a gene responsible for a disease also can be crucial to another biological function, Cascade now is helping to cross the chasm.