Industry-Tailored Clouds Give Platforms More Marketplaces They Can Run

Their internet marketplaces under regulatory scrutiny, Google, Amazon, and others are building more ecosystems – but of a different kind.

Mark Ballard

August 24, 2020

9 Min Read
A Volkswagen plant in Zwickau, Germany
A Volkswagen plant in Zwickau, GermanyJens Schlueter/Getty Images

American tech giants that sell cloud services have begun turning their ubiquitous platforms into business ecosystems with the promise of disrupting whole industries.

Amazon, Google, and others have been trying to effectively replicate the success of their digital platforms in retail, media, and general commerce in the realm of enterprise technology. They are partnering with dominant corporations to develop bespoke services for their respective industry sectors, before offering them as a platform for the rest of the industry. 

With the countless companies that form industrial supply chains and fit out manufacturing workshops contributing data about their working processes, and with cloud providers contributing their powerful artificial intelligence and data analytics services, these platforms are intended to become places where the rules of industry, and its shape, are transformed the same way digital platforms have done elsewhere.

This trend is emerging amidst rising regulatory concerns about the extraordinary market power the tech giants gain from running their platforms. US regulators have been concerned that the platforms have abused their access to privileged data and their power to set the rules to suppress competitors and cheat the open markets their platforms create. Other regulators around the world, particularly in Europe, have warned that cloud providers could use bespoke technologies to lock whole industrial sectors into using their platforms. 

Related:How Google Cloud Plans to Win Enterprises from AWS and Azure

These varying international positions indicate different ideas about what makes a market corrupt. The question of lock-in has become so pressing in Europe, that banking regulators have even proposed to dictate the way cloud providers build their computer systems. Cloud providers themselves cite their use of open source software technologies to claim their platforms will not lock industries in. But the technological basis of the regulatory assumptions is subtle.

The US Congress House Judiciary Committee demonstrated American regulatory concern about digital platforms in late July, when it called chief executives of four big-tech firms to face accusations that they abused the power they got by running digital market platforms.

But the hearing also gave the execs and US representatives alike an opportunity to make a patriotic public defence of the cloud computing phenomenon, the entrepreneurs who made it happen, and the American political and economic system itself, as it proceeds with the "industry 4.0" revolution.

Related:How RBC Built Its Own GPU Farm for an Artificial Intelligence-Powered Banking Platform

Grand Visions

That transformation was evidently underway in recent months, with announcements of groundbreaking ventures by the big cloud providers, who had begun using the AI systems in their cloud data centers to redesign various industries’ processes.

Google pledged to "revolutionize the future of automotive manufacturing" with a cloud platform it built with the French carmaker Renault. Microsoft promised "unparalleled disruption" to the finance industry. Royal Bank of Canada (RBC) said a cloud platform it built with open source software house Red Hat would benefit none less than the whole AI industry and "the world."

Meanwhile, German car giant Volkswagen announced the launch of a business platform with Amazon Web Services, the heart of which was an app store intended to become a marketplace for industrial applications, driven by AWS cloud computing services and fueled with data collected from Internet of Things sensors spread throughout the VW manufacturing supply chain. Marketplace participants would redesign automotive manufacturing using data analytics and machine learning services provided by the platform. VW depicted an ecosystem of businesses contributing their own mashups and innovation back to the industrial app marketplace.

Neither VW or AWS would talk about it when asked by Data Center Knowledge. Amazon CEO Jeff Bezos had just then faced allegations before the House Committee that Amazon abused its exclusive access to data from its retail marketplace to outstep competitors it let operate there.

Bezos made a patriotic defence of American democratic capitalism. Hard work and American values led him from a humble immigrant upbringing to entrepreneurial success with a retail platform upon which 1.7 million small and medium companies traded.

"The rest of the world would love even the tiniest sip of the elixir we have here in the US," he said.

US and European Approaches Diverge

The chiefs of Google, Facebook, and Apple faced similar accusations. They made similar defences. Congressmen said that American democratic ideals were incompatible with concentrated markets and political control. Europe was pilloried for having more regulation and less success.

Such US assertions that the power to control a digital market creates opportunities to abuse it were being made while the US government was leading a campaign against Chinese economic imperialism.

Notwithstanding the fact that UK regulators have encouraged banks to rebuild their computer systems using cloud services, draft UK rules propose dictating that cloud computing systems exclude technologies that would make it tough for a bank to move its systems if it needs to. The Bank of England has thus declared a need for "substitutability" in cloud services. It implies opposition to banks building their complex systems using any cloud provider’s proprietary computer services, because they could not then easily shift their computing to some other provider. They would be "locked in." Other banking regulators concur. But US federal regulators have had a lighter touch. They have encouraged banks to act on their own judgment. "Proprietary" technologies were even an advantage to the banks that used them, the US Treasury said in December.

It was amidst this in July that IBM said it was introducing a financial cloud ecosystem in Europe, in a data center it operated in Paris for the French bank BNP Paribas, intended to help banks overcome regulatory fears of cloud. It wanted to convene a council of banks to agree common rules to govern their data and services, so it could use them to fashion financial computing services that were regulatorily sound. It had designed the rules already, in partnership with Bank of America. It now wanted banks in Europe and other jurisdictions to adapt them.

This was necessary not because regulators and regulations weighed heavy on cloud computing, Hillery Hunter, VP of IBM Cloud, told DCK. It was because banks interpreted the rules conservatively and as a result were reluctant to use public cloud providers. This would typically lead banks to vet any potential cloud computing process with a manual procedure that took 18-24 months. 

Yet they all wanted the same rules implemented in roughly the same way. Banks typically had lots of processes they might move to the cloud. The ecosystem promised rapid innovation. Public, shared cloud services, as opposed to banks’ own private systems, had advantages. But banks wanted to be very sure they could answer any regulatory probe about public cloud processes as well as they would if it concerned their own computer systems. 

"It's highly inefficient. In some cases, with sensitive data, approval has been impossible," Hunter said. If on the other hand they agreed to a common way to implement the rules in a cloud system, it would make financial cloud services more feasible.

But regulatory concerns about lock-in raised the question whether the rules banks agreed to with IBM were proprietary to the vendor. Hunter would not say.

Open Source as Lock-In Prevention?

Lock-in is a sensitive subject. Dominik Wee, managing director of global manufacturing for Google Cloud, told us that Renault used Google because its cloud largely relied on open source software and would therefore not lock it in. Wee said competing cloud services were typically proprietary, but wouldn’t name them.

The basis of this difference, Wee said, was Kubernetes, a technology that has revolutionised software by abstracting it from its computing environment, so that it becomes portable; and atomising it, so that it becomes adaptable and ready for shared use, in mash-ups, in unanticipated innovations, by teams not even considered, for purposes not yet conceived; all of it made widely available through an open API in, for example, a cloud ecosystem. 

Google developed Kubernetes and open sourced it, so there were no commercial hindrances to its use. Other cloud providers adopted it. Their enterprise customers have apparently been using it to break their monolithic business applications into cloud services. But other cloud providers implement Kubernetes in a way that locks customers in, said Wee.

Locked in or not, the result is business applications being built using cloud services. Renault, for example, took a machine learning service that Google developed originally to make its data centers more energy efficient and applied it to car paint shops. Google now plans to offer it as a specialised service for all car manufacturers as part of its cloud platform.

Others have used Kubernetes to develop private cloud services, in their own data centers, for their sole advantage. RBC, ExxonMobil, and the Israeli Ministry of Defence spent the last two years working with Red Hat to adapt Kubernetes to use Nvidia GPUs used in supercomputers, according to Tushar Katarki, a Red Hat product manager. 

The work was made open source, so any cloud provider could use it to build services. The services these organisations developed, however, are private, formed into an ecosystem for their own developers. Conversely, public cloud providers, who by necessity make their services open to anyone, use proprietary technology to lock customers in, said Katarki. He would not say which cloud providers, if any, do not. Red Hat, like Google, sells its own adaptations of open source software to other cloud providers.

A business would become locked in even to Kubernetes as soon as it built a complex application with it, former Google Cloud technical director Gregor Hohpe has written. Lock-in is not black and white, but a trade-off between the cost of moving a computer system from its environment, and the rent required to stay. Even Google's own implementation of Kubernetes is so idiosyncratic that it might lock a customer in. Whereas AWS's proprietary S3 cloud storage service is so popular that all industries have adopted its API. Whether, however, the automotive supply chain was at risk of being locked into using its highly specialised AI ecosystem AWS would not say.

Management consultants say ecosystems demand "a new way of thinking about business." Large companies often fancy running them, thinking they can control the market. But even the largest firms in any sector lack the power. The way to get by in an ecosystem, the theory goes, is not with controlling power but with influence. The mode is collaborative competition, unanticipated innovation, coevolution. The means is ever-shifting and overlapping networks of participants, carrying flows of data, services, and money. It might seem counterintuitive, they say. An exemplar of this adaptive industrial organization is Chinese cloud provider Alibaba. Some take it for granted that the platform itself is not biased.

Read more about:

Google Alphabet

About the Author

Mark Ballard

Mark Ballard is an award-winning journalist who has been writing about technology and related business, law, and public affairs since 1993. Details of his work can be found here.

@MarkJBallard

Subscribe to the Data Center Knowledge Newsletter
Get analysis and expert insight on the latest in data center business and technology delivered to your inbox daily.

You May Also Like