“The public proof set for operator AI is materially thinner than the partnership headlines imply.”

Telecom’s AI partnerships dominate the headlines, but how much value has actually been proven? This Appledore Research Note benchmarks the four hyperscale AI suppliers — AWS, Google Cloud, Microsoft, and NVIDIA — against the public record of operator deployments with named scope and quantified, realized outcomes. The findings are sharper than most coverage suggests. Microsoft leads on broad employee productivity and customer-care results, with Vodafone and Telstra disclosing weekly time savings, follow-up contact reductions, and a first-contact resolution jump from 70% to 90%. Google Cloud is strongest in data and AI-platform industrialization, evidenced by Vodafone’s AI Booster (proof-of-concept-to-production cut from 20 weeks to 2) and Telefónica’s BigQuery-backed analytics rollout. AWS has narrower but credible operational proof through BT’s Amazon Q developer rollout and Orange/MasOrange’s GenGraph root-cause analysis, which compresses diagnostic time from hours to seconds. NVIDIA’s public telco evidence remains pre-commercial, anchored by SoftBank’s AI-RAN field pilots.

The takeaway for vendors and operators alike: many large telco AI budgets are being allocated on narrative rather than disclosed returns. Buyers — and sellers — who insist on quantified, operator-originated outcomes will route capital toward the proof, not the press release.