Amazon Web Services has an Achilles' heel: hybrid cloud. Competitors Microsoft and IBM are light years behind AWS in market share, but Microsoft and IBM have long histories working with enterprises in on-premises computing, with relationships going back decades. Amazon has its public cloud, and that's all, leaving it in the odd position of being both a market leader and a newcomer.

Amazon Web Services Inc. is looking to change that with its announcement last week of new Greengrass software and Snowball hardware, designed to get AWS cloud capabilities out of the data center and onto the customer premises.

Greengrass is designed to run on the network edge, on an emerging generation of Internet of Things connected devices that need to store data and perform computations locally, rather than sending information to a remote data center and wait for the compute to be performed and results returned over the Internet.

The software puts a "mini AWS" -- a select set of AWS capabilities -- inside connected devices, Amazon says. It runs on "almost any device with a general-purpose CPU that runs Ubuntu or Amazon Linux, and supports ARM and x86 architectures." It runs Amazon's Lambda service, which orchestrates workloads on virtual machines, and can run tasks locally, keep device data in sync, and communicate with other devices, as well as connect to the AWS cloud.

Greengrass already has users. As we reported earlier today, Technicolor (Euronext Paris: TCH; NYSE: TCH) is integrating Greengrass into its set-top boxes, gateways and tablets, to monitor connection quality and automatically troubleshoot problems. And NASA's Jet Propulsion Laboratory is using Greengrass on an educational robot to demonstrate programming and test its robots. (See Technicolor Brings AWS Home to Gateways.)

Amazon made another hybrid cloud effort earlier this month, partnering with VMware Inc. (NYSE: VMW) to allow enterprises to run VMware workloads -- a popular platform for on-premises computing -- on AWS. That will allow enterprises to run the same applications on premises that they do in the AWS public cloud, and shift workloads between the two as needed. (See Enemies No More: Amazon & VMware Partner on Cloud.)

Amazon introduced its own device to run Greengrass -- the AWS Snowball Edge, which Amazon describes as a "data transfer appliance" -- a more businesslike way of saying it's a truck, which is what it is. It has a storage capacity of up to 100TB, and can transfer data to and from Amazon Simple Storage Service (Amazon S3), cluster with other Snowball Edge devices to form an on-premises storage pool, and run Lambda to process and analyze data.

Infrastructure manufacturers including Intel, Qualcomm, Annapurna and Labs are integrating Greengrass into their platforms, and Amazon is also making the software available for download.

Philips is using Snowball Edge to extend its HealthSuite connected health device ecosystem, and the Hatfield Marine Science Center is using Snowball Edge for oceanic and coastal research.

Amazon faces competition in appliances for hybrid cloud and edge computing. Startup Vapor IO is building self-contained miniaturized data centers, including racks, cooling and management software, in a cylinder nine feet wide and nine feet tall. And startup CloudZero provides general-purpose compute devices that are deployed on-premises but managed over the cloud. (See Vapor IO Builds Wee Data Centers.)

Amazon announced Greengrass and Snowball Edge at its AWS re:Invent conference in Las Vegas last week.

Also at the conference, the cloud provider introduced three new cloud AI services: Lex, a version of its Alexa technology, to allow developers to build conversational user experiences for the web, mobile and connected device apps; Polly, which "transforms text into lifelike speech, enabling apps to talk with 47 lifelike voices in 24 languages," and Rekognition, for image recognition, Amazon says.

Want to know more about the cloud? Visit Light Reading Enterprise Cloud.

AI is a hot area of competition for cloud providers; it's advanced technology that's difficult for enterprises to implement on their own, but cloud providers look to make it simple. Microsoft Corp. (Nasdaq: MSFT), Google (Nasdaq: GOOG) and Amazon all announced new AI and machine learning technologies earlier this month. (See Hypercloud Guys Make Big AI Push.)

Amazon also used the conference to tout a couple of customer wins. Amazon announced that Matson, a Pacific Ocean shipping carrier, closed all four of its on-premises data centers and migrated to AWS cloud.

And Amazon said that Workday, which provides enterprise cloud applications for finance and human resources, will use AWS as its preferred public cloud infrastructure for customer production workloads. Capital One Financial Corp. said it also plans to use AWS as its primary cloud infrastructure provider, migrating core applications to AWS over the next five years.

— Mitch Wagner, , Editor, Light Reading Enterprise Cloud