Two years ago, Microsoft did something that left executives at Ericsson and Nokia sweating like soldiers before battle.

Within the space of a few weeks, Seattle's biggest export besides coffee had bought two small but irksome rivals to the Nordic vendors in the market for core network software.

The takeovers of Affirmed Networks and Metaswitch appeared to signal Microsoft's ambitions in the telecom sector. Fortified by Microsoft's money, the two companies threatened to be a lot more dangerous.

Taking the industry by surprise, the moves generated some awkwardness between Microsoft and the Nordic firms, both of which had been used to regarding the US software giant as more of a partner than a foe. This vision of Microsoft as a rival to traditional telecom vendors is one it has been subsequently trying to dispel.

"Ericsson and Nokia were very concerned, but at the end of the day what market share does Affirmed have?" said Rick Lievano, the chief technology officer of Microsoft's telecom business, during an interview at the recent Digital Transformation World event in Copenhagen.

"We didn't buy them because they have huge market share. We bought their expertise."

Figure 1: AT&T's operations center, still presumably under the control of AT&T.
(Source: Microsoft)

Yet what Microsoft did next was an even bigger shock to the industry. In June 2021, AT&T announced plans to run its entire 5G network on Azure, the name given to Microsoft's public cloud. As part of that arrangement, staff and assets would change hands, moving from AT&T to Microsoft.

Financial terms were not disclosed, and the deal was not even mentioned in AT&T's annual filing with the US Securities and Exchange Commission. But Microsoft had effectively bought the control center of the AT&T network and the expertise needed to run it.

These separate but related moves laid the foundations for what Microsoft now calls Azure Operator Distributed Services (AODS). The goal, as described in a blog earlier this year, is to give operators a "carrier-grade" cloud platform where they can feasibly run most of their workloads, including their core, radio access network (RAN) functions and business and operational support systems.

Microsoft is now in discussions with several operators about this yet-to-launch commercial AODS offer. Ultimately, Lievano talks about capturing at least 60% of the overall telecom market for public cloud services.

Different strokes

It is a markedly different strategy from the one AWS is pursuing, insists Lievano.

The world's largest public cloud has been far more visible in the telecom sector globally, signing up numerous operators as tenants and dozens of software companies as partners.

Dish Network, its flagship customer, has opted to run nearly all its workloads on AWS. Just last week, Japan's Rakuten revealed that its Symware app store of network goodies would become available through AWS, without even acknowledging the existence of Azure or Google Cloud, the distant number three.

"They have chosen to learn with their customer projects," said Lievano.

"AWS is going to learn a lot from Dish. Dish is going to pay a lot for AWS learning their business. We made a conscious decision not to do that. We are not going to charge customers for learning and becoming telco experts. We are ultimately going to acquire the right level of expertise."

That is necessary, according to Lievano, because the telecom sector cannot be treated like retail or hospitality.

"The reality is the telco sector has very explicit requirements that go beyond those of most industries, particularly around the network," he said.

With Affirmed and Metaswitch, plus the engineering expertise it bought from AT&T, it could work on bridging the gap to the cloud without putting customers at risk.

"We made an investment from an R&D perspective, not a sales-enablement perspective."

The main difficulty lies in the RAN, where even Lievano concedes the public cloud might not currently make sense economically.

"Remember how cloud providers work – you are not charged for uploading all that data to the cloud, but you are charged egress for getting data out of it," he said.

"If it's voice traffic being generated through your radio, it doesn't make sense to upload it to the cloud and then immediately egress it out to the destination. You'd be paying for all that egress."

Even so, Lievano's best estimate is that only about a fifth of telco workloads would run more efficiently in a private rather than public cloud.

Relatively few of them are hosted in a public cloud today. If the industry agrees with him about this 80:20 split, the opportunity for AWS, Microsoft and Google is huge.

Figure 2: Rakuten's Tareq Amin speaks to Light Reading in Las Vegas.
(Source: Light Reading)

One of the sector's most prominent figures seems broadly to concur.

"There are instances where the cost-ratio benefit of being on-prem versus public cloud maybe doesn't make sense," said Tareq Amin, the CEO of Rakuten Mobile and Rakuten Symphony, during a video interview with Light Reading at last week's Mobile World Congress Las Vegas.

"Today, if you ask me, [it is] the DU only," Amin continued, referring to the distributed unit (DU) responsible for baseband processing in the RAN.

"This is where I would say the cost structure associated with this DU to run on public cloud doesn't make sense. Public cloud cannot easily scale this out today. It doesn't mean tomorrow this problem cannot be solved."

Public cloud doubts

But others are not convinced. Software-as-a-service models still hold the greatest appeal for startups and smaller companies that lack the resources to build their own private clouds. As companies grow and gain scale, the public cloud becomes less economically attractive, according to James Crawshaw, a principal analyst with Omdia (a sister company to Light Reading).

Andreessen Horowitz, a venture capital firm, has also questioned the economic arguments about the public cloud, saying "the pressure it puts on margins can start to outweigh the benefits as a company scales and growth slows" in a paper published last year.

While he was sounding more amenable to the public cloud in Las Vegas, Amin last year said the public cloud providers were "years behind" his own company in running cloud mobile networks. The idea they could do it at lower cost was "completely wrong," he told reporters.

Cost is not the sole issue, either. For some telco executives, the bigger worry is a loss of control. Giving up network management as AT&T is doing is like "outsourcing a core competency," said Scott Petty, Vodafone's chief digital officer, during a conversation with Light Reading last year.

There is also the risk of vendor "lock-in," something operators are desperate to avoid in the network. Companies reliant on the public cloud have noted the difficulty and cost of moving workloads from one provider to another.

The more a company makes use of a public cloud's own services, the harder it can be to move, said Jerome Loridan of Orange Business Services. Most companies, said Andreessen Horowitz in its paper, "find it hard to justify moving workloads off the cloud given the magnitude of such efforts."

Want to know more about 5G? Check out our dedicated 5G content channel here on Light Reading.

But wariness about over reliance on one cloud could suit Microsoft in the short term. As a public cloud provider to all sectors, AWS is much bigger, catering to 34% of the market during the second quarter of this year, compared with the 21% served by Microsoft, according to data from Synergy Research Group.

Its relatively high profile in the telecom market so far might also prompt operators to think twice about handing AWS more business.

"In situations where AWS is the incumbent, you absolutely don't want to put all your eggs in one basket," said Lievano.

"We need to figure out what is the split. How can Microsoft be the preferred provider? How can we be 60% or 70% or 80% of general-use capability versus 30% or 40%?"

Figure 3: (Source: Synergy Research Group)

Microsoft, encouragingly, has been growing its share of the overall cloud market faster than any other big provider in the last five years, gaining about eight percentage points since late 2017, according to Synergy, mainly at the expense of smaller clouds.

And while Google Cloud has also advanced, it remains unprofitable. Losses there hit nearly $1.8 billion for the first six months of 2022, a 14% increase year-on-year.

The onus is suddenly on traditional vendors, including Ericsson and Nokia, to show their software can live in the public cloud – and perhaps not just one provider's public cloud.

Telco demand for multi-cloud capabilities, and the right to choose the public cloud host, could generate extra work for software developers trying to preserve cloud-agnosticism without sacrificing competitiveness. Love it or fear it, the public cloud looks unstoppable.

Related posts:

— Iain Morris, International Editor, Light Reading