Telcos with enterprise service divisions see B2B GenAI opportunities, but should they invest in new infrastructure to pursue them?
Telcos eye GenAI enterprise opportunities
Telcos are still in the early stages of formulating their GenAI strategies. Nonetheless, those with strong B2B and IT services divisions are weighing up how they can target enterprise customers with GenAI solutions.
Indeed, Swisscom Group believes the opportunity is large enough to warrant spend on generative AI full-stack supercomputers in Switzerland and Italy. Providing GenAI services to enterprises and government organizations is part of the rationale behind its investment in building what it calls a ‘trusted AI factory’ through a deal with Nvidia. The company intends to invest up to 100 million Swiss Francs into AI solutions over the next years.
Orange Business Services meanwhile has seen “AI services booming by +40% in 2023”, according to Aliette Mousnier-Lompré, CEO, Orange Business, in a LinkedIn post. And the company clearly expects GenAI to drive further growth as enterprises, and in particular small and medium-sized companies, “realize that off-the-shelf AI will not necessarily give the best outcome”.
Cracking the data conundrum
One of the major sticking points for enterprises, notes Mousnier-Lompré, is accessing, classifying and using business-specific data.
“The data is crucial, not just its quality but also its availability and its classification. We have seen quite a few customers whose move of data into the cloud has been accelerated by their first inroads into AI,” says Mousnier-Lompré. “I’ve also seen customers in tricky situations because they did not have the proper data classification and started to fuel their AI engines with confidential data.”
Swisscom also sees an opportunity to help enterprises overcome the complexities of data management. These include meeting regulatory requirements for data governance, as well as ensuring security, both of which are behind the growth of sovereign cloud services.
“A customer must be sure what happens with its data, where it happens and how it happens. Switzerland, with its many multinational organizations, needs trusted and unique sovereign building blocks for AI,” said Christoph Aeschlimann, CEO of Swisscom, in a statement.
"Data protection is a critical factor," agrees Ronnie Vasishta, VP of Telco at Nvidia. "Especially when there are regulations around how data can be accessed and moved ... and in many countries that data needs to be provided by a trusted source," he explains. "Telcos are ... trusted sources [and can play] a trusted role in ... AI infrastructure."
Extending data sovereignty
The idea of telcos providing sovereign AI may be quite new, but its cousin, the sovereign cloud, is not. A handful of European telcos with large enterprise service arms have been busy putting in place data sovereignty and sovereign cloud services in partnership with systems integrators and hyperscalers for some time.
IDC describes data sovereignty and the sovereign cloud as subsets of “digital sovereignty”, which refers to “giving data owners total control over how and where their data is managed, stored and processed by service providers”, including all the infrastructure and the support staff that can access the data and infrastructure, according to Rahiel Nasir, lead analyst for Worldwide Digital Sovereignty and Associate Research Director for European Cloud, at IDC.
The telcos best placed to take advantage of providing AI services to enterprises, sovereign or otherwise, are – like Swisscom – typically former incumbents with a systems integration arm and a large share of the local IT market, according to Tom Rebbeck, Partner, Research, at Analysys Mason.
“Middle Eastern operators," he adds, “are also typically the largest IT company in their country. They will be working with customers on exploring potential use cases for GenAI – almost always built on top of the LLMs [large language models] from Meta, OpenAI etc.. For these operators, it is really just a continuation of their existing business, working with customers on applications, cloud, security and so on.”
At the same time GenAI is creating new opportunities to partner. Mousnier-Lompré points out that “one astonishing thing about GenAI is how it swiftly created an almost brand new value chain." And as our recent report Generative AI: Operators take their first steps explores, when it comes to building commercial models, that value chain is still emerging (see graphic).
So far Swisscom is one of only a small handful of telcos that has opted to invest in national AI supercomputer infrastructure. Nvidia, which has also signed agreements with SoftBank in Japan and with Reliance Industries in India, expects more telcos to invest in AI infrastructure in coming months. In addition to developing AI sovereignty services, Vasishta expects telcos to provide and train AI foundational models in local languages.
Nation state AI
IDC agrees there will be a trend towards telcos and/or nation states building their own AI service infrastructure. While policy makers want their nations to benefit from AI, they also fear AI systems and infrastructures being taken offline, or being used by adversaries to act against their national interests, according to Chris Silberberg, Research Manager, EMEA Telco Insights, IDC. National incumbent operators are already trusted suppliers, making them a natural ally to national or transnational bodies, such as the EU, in developing sovereign AI infrastructures, he adds.
AWS, Azure, Google Cloud, Alibaba, and Oracle already hold a strong competitive advantage. However, "where global players have limited footprint, such as in Eastern Europe or the Middle East and Africa, there is greater scope for national or regional players like telcos to make the infrastructure landgrab to establish a leading position in these markets," notes IDC.
Nonetheless national - or even international - scale is not a guarantee of profits in the short- to mid-term. "Google Cloud which includes GCP and Google’s workspace initiative lost nearly $3bn in 2022, despite growing revenues to over $26bn, up $7bn YoY," states Silberberg. "There is the risk that national players in many markets simply won’t have the scale to square this as a competitive commercial offering without significant policy maker support and funding."
Some of the most ambitious telcos today are in South Korea, and they are also targeting business with other communications service providers.
"SK Telecom is building its own LLM and working with Anthropic to support telco use cases built on Claude. KT has its own model, MI:DM," says Rebbeck at Analysys Mason. "Both are trying to sell what they develop to others, including other telcos."
But telcos do not have the deepest pockets: Whereas SK Telecom has invested $100 million into Anthropic, AWS has invested an estimated $4 billion, notes Analysys Mason.
Other operators are likely to be less ambitious and use foundation models from Meta, OpenAI, Anthropic, or in the case of smaller telcos, sell solutions from companies such as Microsoft, says Rebbeck.