Josh Builta - Research Director
Omdia’s AI Applications Software Market Forecast predicts that the value generated by artificial intelligence (AI) applications in telecoms will reach $8.7bn by 2027, increasing by a factor of eight from immediate pre-pandemic levels.
Customer experience automation has so far been the key sector representing as much as 78% of this total, being one that delivers immediate, measurable benefits on KPIs, such as churn, lifetime customer value, and subscriber acquisition cost, and also being one that suits the technology, as recommendation engines and natural language processing have been revolutionized by the transformer class of deep-learning models.
Omdia observes that CX-related AI apps—the CX, virtual assistant, and intelligent CRM categories in the chart—have fallen from being 78% of AI value generation in telecoms to just under 50% in 2022–23, before stabilizing at that level. When we say 50% here, though, we mean 50% of a fast-growing market—CX revenue will continue to rise rapidly, from $1.4bn today to $4.1bn by 2027.
However, the situation has begun to change. We expect AI providers at MWC to showcase how AI is moving from an add-on feature for CX applications to being part of the core business, as network optimization and IT operations automation ride up the adoption curve, and as AI technology increasingly forms part of the network itself.
As an example, NVIDIA’s Aerial 5G solution consists of a virtual 5G base station built around its Bluefield-2 SmartNIC and A100 GPU. This is intended both to provide hosting for edge computing applications that need true AI processing power, but also to support AI-driven approaches to core mobile functions such as beamforming, beam steering, and self-organizing networks.
Source: Omdia
No surprise, again, but the dominating force in mobile chip design is increasingly an AI company, with its new Ethos line of machine-learning accelerators and in-CPU acceleration in versions 8.2 and 9 of the core architecture. Expect announcements ranging from IoT to servers.
We’ve heard relatively little from this company in mobile since the end of the Tegra product line, but the increasing importance of AI has drawn NVIDIA back in. Expect more of the low-power AI acceleration technology developed for the Atlan automotive SoCs to turn up in mobile and IoT. We will be fascinated to see what applications operators and others find for the Aerial 5G’s mighty A100 GPU.
Invariably a big presence at MWC, Qualcomm is increasingly a force in AI generally after its CloudAI 100 chip did extremely well at 2021’s MLPerf Inference 1.1 benchmarking competition. The latest Snapdragon 8 Gen1 system on a chip (SoC) will be judged in part on its AI performance, especially as it is the first to use ARM’s v9 microarchitecture, but also expect announcements moving Qualcomm mobile AI technology beyond mobile.
You can meet with Omdia's expert team at the following events:
GroqDay (March 31, Online)
AI Summit London (June 15–16, ExCel, London)
IoT World (September 28–29, 2022, Santa Clara, CA)
