ARK Disrupt Issue 66: Bitcoin ETF, Payment Apps, Deep Learning

Please enjoy ARK Disrupt Issue 66. 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. The SEC Rejects The Bitcoin ETF, and Blockchain Asset Prices Take Off

This week was pivotal for blockchain assets. After the SEC disapproved the Winklevoss bitcoin ETF, the prices of many other blockchain assets like ether, dash, monero, augur, and zcash soared, many by 50 to 100% during the week ended March 17th.

Among the explanations for these exponential moves are a rotation out of bitcoin now that an ETF will not generate incremental demand, and an increased focus on the improving fundamentals supporting blockchain assets. Attracting the most attention, Ethereum benefited both from Bloq joining the Enterprise Ethereum Alliance and from increased institutional interest. The news on Bloq was particularly important because its CEO, Jeff Garzik, has been a long-time developer and core supporter of Bitcoin. That said, while convinced of Ethereum’s value proposition, we do believe that its recent trading volume and price spike point to an overheating market and warrant caution in terms of price projections.

Meanwhile, last Friday nearly 20 of bitcoin’s top exchanges released a contingency plan in the event of a Bitcoin hard fork. This plan is the result of continued debate, if not a polarizing battle, about the best way to scale bitcoin. While we are concerned about the optics of a hard fork, we remain firm in our belief that bitcoin is the most resilient and antifragile currency in the world.

2. Google Has Launched A New Peer-to-Peer Payment App to Compete With Venmo

Last week, Google launched a peer-to-peer payment (P2P) feature on the Gmail app for its Android customers in the US. Like Venmo, the Gmail app allows friends and family to split bills and transfer money to one another. Users can make payments with debit cards, bank accounts, and GoogleGOOG wallets…but at this time not with credit cards. Users can send and receive payments with attachments to their emails, much like sending documents and pictures with a tap on the attachment icon, as shown below.

This fee-free way to send money not only puts Google in direct competition with services like Venmo and SquareSQ Cash, but also enhances its strategic position with data from user transactions…giving it another shot at the giant in that space, AmazonAMZN. With growing competition in the P2P payment and transactions data space, ARK is focused on any changes at the margin.

3. Tesla Raises Capital, and Demonstrates the Power of Twitter to Drive Business

This week, in a much anticipated move, TeslaTSLA raised $1.2 billion in equity and debt: $350 million from equity, and $850 million from five year convertible notes with a 2.375% yield and a 30% conversion premium. Because the dilution to existing shareholders was less than anticipated and Musk bought almost exactly $25 million of stock, Tesla’s stock rallied on the news.

Elon Musk used Twitter to make some more headlines this week. After debuting the Powerwall 2 in Australia, former CEO of SolarCity, Lyndon Rive, told journalists that Tesla could solve South Australia’s power problems within 100 days. Musk backed him up with a tweet that Tesla would provide the solution for free if it failed to keep that timeline. Taking his cue from that exchange, another user responded to Musk and mentioned the Prime Minister of Ukraine, Volodymyr Groysman, who seemed interested and asked Musk to follow up with details, as shown below.

Is it possible that Musk will close more than $100 million in deals because of two or three tweets during the last week? If so, Twitter should figure out a way to get a piece of the action!

ARK Disrupt Issue 66 TweetsSource: Twitter, ARK Investment Management LLC

4. Is Oxford Nanopore’s Sequencing Technology Finally Ready For Prime Time?

In a webcast this week, Oxford Nanopore introduced a new portable DNA sequencer, the GridION X5, a benchtop system that can analyze up to five MinION flow cells, increasing the capacity of its pocket-sized portable device by five-fold. It explained that the system is ideal for labs requiring greater capacity than that offered by a MinION, while preserving the advantages of nanopore sequencing: “simple library preparation, real-time analysis, and new biological insights from long reads.” GridION also will allow labs to offer nanopore sequencing as a service. While the pricing for GridION is confusing relative to the MinION, we are encouraged by Oxford Nanopore’s progress in the long-read DNA sequencing space.

5. Was Intel’s Acquisition of Mobileye a Move on the Offense or Defense?

This week Intel bought Mobileye for $15 billion, nearly 30x forward sales or roughly 10x the price-to-sales ratio of the NASDAQ. IntelINTC had been collaborating with MobileyeMBLY and BMW on an autonomous car project while, at the same time, supplying processors to another partnership with Mobileye, Delphi, and Quanergy to develop an autonomous technology suite for a broad base of auto manufacturers. Clearly, Intel viewed Mobileye as an opportunity to gain some scale in the autonomous vehicle chip market, which ARK estimates will be worth $30 billion+ by 2025.

With almost half a century old x86 architectures accounting for the bulk of its sales, we think Intel bought Mobileye to gain access to new computing DNA, or computer vision. Mobileye’s EyeQ system-on-a-chip has become part of the active safety ecosystems at many if not most traditional automakers, attracting Intel’s interest as auto-related chips make up less than 3% of its revenues today. Qualcomm’s recent acquisition of NXP, the leading automotive semiconductor supplier, probably added more urgency to Intel’s strategic decision to acquire Mobileye.

While Mobileye’s expertise is in ADAS, or semi autonomous driving, fully autonomous vehicles will require much more powerful chips as well as R&D spending perhaps multiples of its current budget. At roughly half the cost of driving a personal car, fully autonomous vehicles will be highly disruptive to traditional auto manufacturers. As part of Intel, Mobileye will have access not only to significantly higher R&D resources but also to superior semiconductor manufacturing expertise and capacity to capitalize on the fully autonomous opportunity. That said, with their hardware orientation, Mobileye and Intel will be competing head on with NVIDIA, the leader in AI processors.

One last thought: Now that it has ended its relationship with Mobileye, Tesla is developing, or may have developed, its own system-on-a-chip for computer vision. In that context, Mobileye’s $15 billion valuation puts Tesla’s $40 billion valuation into an interesting perspective.

6. Deep Learning Compression Is 10x Better Than JPEG…When It Works

Deep learning (DL) has superseded an entire class of algorithms in image recognition, speech transcription, machine translation, and other applications. Thanks to Jack Clark’s excellent Import AI newsletter, we found a new paper suggesting that DL has the potential to replace JPEG and MPEG, the de facto algorithms used for image and video compression today.

The new algorithm, NCode, is based on Generative Adversarial Networks (GANs), a relatively new DL network that can create plausible images after being trained with examples. The pictures below compare the original image (row 1) to JPEG compression (rows 2-3) and NCode (rows 4-6) at various levels of compression. For most images, NCode achieves dramatically higher levels of compression and image quality.

ARK Disrupt Issue 66 Deep LearningSource: ARK Investment Management LLC

At this early stage of its evolution, NCode does have a significant weakness: the images it evolves, even if plausible, can be very different from the original one. Some of the images in the last row, for example, show a smile even though the original does not. Moreover, if the input image contains objects never before trained on the network, the output can differ dramatically from the input.

That said, DL compression has great potential. With a 10x improvement, it could enable low cost video streaming on wireless networks, expanding Netflix’sNFLX addressable market from ~1 billion fixed line broadband subscribers to ~4 billion mobile broadband subscribers.

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.