IBD Digital 2 months for $20 offerIBD Digital 2 months for $20 offer


Next Big Thing In Cloud Computing Puts Amazon And Its Peers On The Edge

Edge computing concept

The cloud computing revolution faces a reset as a new approach called edge computing puts data analysis closer to users and internet-connected devices. (Dennis Nishi)

Hangar Technology wants to shake up the drone market by getting rid of human pilots. But the bigger story might be how the startup uses the next big thing in cloud computing, called edge computing.

X

Edge computing deploys data processing, storage and networking close to sensors and where other data originate. The goal is to process and analyze data locally in real time rather than send it to faraway data centers in the internet cloud and wait for a response of, say, 150 milliseconds.

That sliver of time matters in some applications: A self-driving car that needs to detect a pedestrian or storm-damaged sign. A doctor who performs remote surgery using augmented reality tools. A video surveillance system equipped with facial recognition to identify a known shoplifter or terror suspect. Sensors that detect repairs needed on robotic arms in factories and oilfield pumps.

Edge computing is the next step that companies like Amazon.com (AMZN) are taking to expand the cloud. Incumbents mean to protect millions or billions of dollars in their revenue from poaching by edge upstarts. Companies new and old are developing relevant hardware and software.

"I'm convinced edge computing is here to stay," said Marco Argenti, vice president of technology for Amazon Web Services. "It's a natural evolution. Not all data needs to be transferred to the cloud to be processed. There might be bandwidth costs, cellular costs or it may be difficult to connect to the cloud. Or, you just need to react really fast, like in the case of a robot."

That's what Texas-based Hangar thinks. It has teamed with Vapor IO, which makes mini data centers, to bring cloud computing closer to drones, making them more nimble. The mini data centers will sit at the bottom of cellphone towers and other spots near where the drones fly.

"We're building our entire technology platform to operate in a future where we don't need people in the loop," said Jacob Rachniowski, Hangar's director of enterprise solutions.

Cloud Computing, Decentralized

Centralized cloud computing could soon be old hat. Today's internet "cloud" is made up of huge racks of computer servers packed into data centers. Some 200 warehouse-size, high-performance, "hyperscale" data centers span the U.S., says Synergy Research Group.

They're operated by Amazon, Microsoft (MSFT), Alphabet's (GOOGL) Google, Apple (AAPL), Facebook (FB) and other tech companies as well as data center firms like Equinix (EQIX).

A number of startups — including Vapor IO, EdgeMicro, Cloudflare, Packet and others — aim to challenge these giants and make edge computing a reality.

Many incumbents in the tech industry are investing in edge computing. Among them are Dell Technologies (DVMT), Hewlett Packard Enterprise (HPE), Crown Castle (CCI), Cisco Systems (CSCO) and wireless firms Verizon Communications (VZ) and AT&T (T).

How Can You Handle This Market? Start With These 3 Steps

Amazon, Microsoft and Google are the biggest providers of public cloud computing services. But powerhouse Amazon Web Services, or AWS, also offers edge computing software and services, primarily for web-connected devices on what is called the Internet of Things.

"Our philosophy is that edge is really an extension of the cloud, down to the tiniest device. There is no edge without the cloud — otherwise you're going back 15 years," Amazon Web Services' Argenti told IBD

Cloud Computing Revenue At Risk

But the idea is that data collected on the edge supposedly will be parsed for what's most important, and the rest will be sent to the cloud. The idea is to cut down on what's eventually sent to centralized clouds, saving time and money.

"By circumventing the need to access the cloud to make decisions, edge computing provides real-time local data analysis to devices, which can include everything from remote mining equipment and autonomous vehicles to digital billboards and wearable health appliances," said a report from consultancy McKinsey.

The big question is how much data processing will take place locally rather than being crunched at giant data centers. Revenue could shift to edge computing companies from the cloud's current leaders.

Amazon's cloud computing revenue is nearing a $25 billion annual run rate. And AWS revenue popped 49% in the June quarter.

Microsoft's Azure cloud computing services exceeds $9.8 billion in revenue while a similar Google venture tops $3.6 billion, says a Jefferies report. Data center tech providers — such as chipmaker Nvidia (NVDA) and equipment maker Arista Networks (ANET) — have a stake in how edge computing evolves.

Some expect the demise of centralized cloud computing. Peter Levine, managing partner at venture firm Andreessen Horowitz, predicted in a late 2016 video presentation that new edge apps "will obviate cloud computing as we know it."

Edge Computing 'Will Eat The Cloud'

Gartner Group, in a 2017 report, predicted that "The Edge Will Eat the Cloud."

"The edge will create some serious winners and losers, both in terms of vendors and businesses," said Thomas Bittman, a Gartner analyst, in a blog post.

But John Apostolopoulos, vice president of Cisco's enterprise networking business, sees plenty of growth ahead for both centralized cloud computing and edge computing.

"The pie is going to expand dramatically," he said. "The cloud is going to continue being important and it's going to grow. The edge, though, is going to grow much more than it has in the past and create a lot of value for customers. It's not going to be one or the other."

Apostolopoulos points to retailers like Walmart (WMT) as an example. With edge computing, he says they'll be able to bring cloudlike software applications to hundreds of stores "at the push of a button." The retail apps would process data from store cameras or sensor networks locally.

Walmart is building its own internal cloud network as its rivalry with Amazon heats up.

Traditional cloud services will keep growing as large companies move business workloads to AWS, Microsoft's Azure and other services, analysts say. Many emerging apps such as video surveillance will require the massive, data-crunching power of centralized cloud computing.

Hangar Technology's Drones

At Hangar, Rachniowski notes that a next-generation drone would still upload video at the end of each day to a traditional cloud provider, like AWS, for analysis.

Hangar plans to have its drones pop out of the mini data centers and fly off. With the help of precision location tools, they'll be controlled remotely, most likely with 5G wireless.

Taking drone pilots out of the equation may be a tall order. Federal regulators now require each commercial drone to have its own ground-based pilot who stays within that aircraft's line-of-sight.

Hangar's customers inspect construction and energy sites using conventional drones. With edge computing, the company hopes to take video data of construction sites, bridges or oil and gas pipelines and whisk them to mini data centers for fast processing and analysis. Pilots no longer will be needed to unload small data storage cards from drones.

It's an ambitious goal, aimed at improving drone economics. The drone market is tough. Well-funded drone startup Airware closed in September. It burned up $118 million in funding. Rachniowski insists the move to remote piloting is necessary.

"Digitizing the real world, deep data analysis, adjusting flight plans, will require edge computing to help us make educated decisions in real time," he said.

Cloud Computing Data Delays

Edge computing targets emerging apps that require real-time decision-making. Giant hyperscale data centers often sit tens or hundreds of miles from where data generates. It takes 150 to 200 milliseconds for data to go to a cloud services provider and back. That time span is called latency.

A 150-millisecond delay is no problem if you're searching for a restaurant on your smartphone. But cloud latency has been a hurdle for multiplayer online video gaming and other markets.

Edge computing aims to process data locally and cut the time to 5 to 10 milliseconds or below. On a factory floor or in a hospital bed, the difference can be meaningful.

"New economy companies and data in general is moving to the cloud and the cloud is moving to the edge," said Cole Crawford, founder and chief executive of Vapor IO. Its investors include cell tower operator Crown Castle and Berkshire Partners. Vapor aims to deploy 100 edge mini data center sites around the U.S. by 2020, up from two now.

"Deep learning analytics, data storage, all that stuff can live in hyperscale data centers for a very long time," he said. "I do believe that high-performance, high-touch apps that require low latency, low jitter are going to be located at the edge and there's going to be a lot of it."

Artificial Intelligence Fuels The Edge

Some technologies associated with edge computing — including artificial intelligence, augmented reality, 5G wireless, autonomous vehicles and the industrial Internet of Things — are still nascent.

Artificial intelligence is just starting to be used in cloud computing. AI software programs — made of computer algorithms — analyze huge amounts of data to identify patterns and predict outcomes. AI is being used in cybersecurity, health care and other industries. A big part of Nvidia's growth has come from selling AI chips to cloud data centers.

Tractica, a market research firm, forecasts that AI edge device shipments will boom to 2.6 billion units annually by 2025. That's up from 161 million in 2018. Edge devices include smartphones, smart home speakers, IoT cameras, drones, self-driving cars and manufacturing robots.

Google reportedly is developing a version of an AI chip used in data centers for edge applications. ARM, a chip designer now owned by Japan-based SoftBank (SFTBY), also is developing devices for edge applications.

Edge computing applications will require more computer memory horsepower to process AI, said a Guggenheim report in March.

"While most of today's machine learning takes place entirely inside the cloud, executing more (machine learning) at the edge has implications across the tech supply chain," said Robert Cihra, a Guggenheim analyst.

Edge Computing And 5G Wireless

AT&T and Verizon still test next-generation 5G services using high radio frequencies. They see 5G improving mobile video and providing "fixed" broadband to residential customers.

AT&T and Verizon also are testing 5G for enterprise applications that could use private wireless communications. As part of that, AT&T and Verizon are deploying new network software for flexible, "intelligent edge" services.

5G wireless will help make the cloud ubiquitous, says Ihab Tarazi, chief technology officer of startup Packet. Tarazi has also been the tech chief at Equinix and a vice president of engineering at Verizon.

"Wireless architectures are evolving and will be huge enablers," Tarazi said. "The cloud is going to extend, filling in all the empty gaps to the edge."

The startup's investors include SoftBank, Dell Technologies Capital and Samsung Next. Packet is teaming with another SoftBank-owned firm, wireless services provider Sprint (S), and Ericsson (ERICY) to offer managed services for the Internet of Things.

Packet on Nov. 6 said it was partnering with  SBA Communications (SBAC) to deploy its first Boston-area edge data center location at a cell phone tower.

Packet positions itself as an alternative to AWS. The startup has put cloud hardware in 18 smaller data centers located in metro areas. It plans to add 50 more. Customers can customize their cloud hardware and use open source software to develop edge applications.

Cloud Computing Economics, Pricing

Enterprise customers will wait for edge pricing to be attractive, as they have in the case of cloud computing. Amazon, in its early days as a cloud services provider, continually slashed prices to attract customers.

Cloud computing titans prospered under a shared infrastructure model. Clients rent computing resources via the web. It's unclear whether a multitenant business model will emerge for local, mini data centers.

Schneider Electric, EdgeConneX and EdgeMicro are among providers of mini data centers. The data centers, about the size of a shipping container, cost around $280,000 to deploy in edge sites, not including servers, switches and other equipment, says a Credit Suisse report. Aside from power and cooling, the mini data centers provide wireless antennas and fiber-optic network connections.

According to an IDC survey, large companies have an interest in delivering new digital services closer to customers. But they worry about the upfront costs in edge computing.

"We've seen a few vendors exit the micro data center space because of the level of customization and not a lot of demand made it unprofitable," said IDC analyst Jennifer Cooke.

Operators of large data centers, such as Equinix and Digital Realty (DLR), have yet to jump into local data centers.

"The engineering case for deploying equipment closer to the edge of the network makes sense to us; the economics remain a big question mark," said MoffettNathanson analyst Nick Del Deo, who follows the data center industry. He says it's unclear whether wireless firms want to invest in edge infrastructure themselves.

Dell, HPE Eye Edge Computing

Microsoft, though, has disclosed plans to invest $5 billion in IoT and the edge over the next four years. HPE has bet $4 billion on intelligent edge technology.

For computer server makers like Dell and HPE, edge computing could be a chance to rebound. Corporate spending on internal data centers has dropped as more business workloads shift to the public cloud.

Many tech industry incumbents are eyeing the rise of edge computing.

"It's an opportunity to serve a converged infrastructure," said Jim Davis, analyst at Edge Research Group. "Some of these companies have broad product lines and they're looking at a new environment."

VMware, Cisco, Nokia Moves

VMware (VMW), a Dell subsidiary, in September announced new software for edge computing. Cisco is building specialized servers for the edge. Telecom gear makers Ericsson and Nokia (NOK) are developing edge computing gateways.

Analysts expect next-generation content delivery networks, from startups Cloudflare and Fastly, to handle edge applications.

Wells Fargo, in a September report on edge services, said it expects the cloud computing titans to prosper.

"Although edge computing is still in its early days, we view Amazon and Microsoft as having the most comprehensive portfolio of edge-to-cloud technologies and a generally available edge platform — followed by Google," the report said.

Scale Computing, a startup that makes servers and data storage systems for edge computing, in late October raised $35 million in a funding round. China's Lenovo is a new investor.

YOU MAY ALSO LIKE:

Cloud Computing: Find Top Cloud Stocks And Track Industry Trends

Looking For Potential Breakout Stocks? Start With These Lists

Improve Your Investing Results With This Simple 3-Step Routine

Sell And Take Profits Or Hold? Here Are Several Guidelines To Follow

Chart-Reading Basics: How A Buy Point Marks A Time Of Real Opportunity