Summary Bullets:

• BT’s partnership with AWS gives internal BT divisional stakeholders a flexible path to generative AI (genAI) that should lead to customer-facing benefits and operational efficiencies.

• There is potential for the model used for this partnership to be adapted into an enterprise genAI solution with BT playing an aggregator role.

BT has announced the launch of BT GenAI Gateway, an internal genAI and large language model (LLM) platform delivered based on AWS technology. The solution has been built in partnership between BT and AWS, is hosted in AWS infrastructure, and makes use of AWS’ Amazon Bedrock (API access to genAI) and Amazon SageMaker (machine learning tools) solutions. Via the Bedrock platform, BT will have access to genAI foundation models (FMs) from companies such as AI21 Labs, Anthropic, Cohere, Meta, Mistral AI, Stability AI, and, naturally, Amazon itself.

The struggle to become more digital and more automated is existentially critical for service providers and any platform launch or partnership that makes the transition easier is to be welcomed. The partnership route is necessary for telcos that are not realistically able to become genAI platform providers in their own right. BT has chosen a strong partner in AWS as its aggregation model, as BT notes, means that it is not tied to a single genAI provider. GlobalData’s research, including its new LLM Competitive Landscape Assessment (Large Language Models (LLM): Competitive Landscape Assessment) has highlighted that different genAI/LLM platforms offer different strengths and weaknesses. Being able to select the most appropriate platform is vital to BT’s efforts to create the best possible internal user and external customer experience.

Google, Microsoft, and OpenAI are clear absentees from the list at present. Given their competitive positioning it is unlikely that they will become available via the Amazon Bedrock APIs, but the opportunity is there for BT to expand the scope of its portal and make available APIs to a wider set of genAI/LLM providers. If BT chooses to take this route, there are also the glimmerings of a potential pathway to greater relevance for BT (and other service providers) in the genAI market.

BT’s role will, as noted above, not be as an LLM platform player. However, as with its multicloud strategy, there is the potential for BT to become a multi-genAI player. Aggregating LLM APIs in conjunction with access and cloud/data security and data sovereignty management services may be a way for providers such as BT to capitalize on the reluctance of the major AI players to cooperate. Connectivity to public cloud and digital infrastructure providers can be combined with service providers’ own edge facilities, which are likely to become relevant to more latency-sensitive use cases.

However, connectivity and security on their own may not offer sufficient value add (although they are both a clear value add). Where BT may also be able to offer further relevance and market perception is as a large-scale user of genAI in internal and customer-facing platforms. Helping businesses to understand what is possible and how to manage the costs of using genAI services (e.g., data egress) and then providing a pathway to those services can be a new route to market for service providers urgently seeking ways to increase their presence in the AI value chain.