AI4INDONESIA
Industry

Telkom CEO Says Most Generative AI Pilots Miss Profit Goals as He Maps Indonesia’s Future AI Industry

October 30, 2026 · 6 min read · Industry

Telkom CEO Says Most Generative AI Pilots Miss Profit Goals as He Maps Indonesia’s Future AI Industry

Telkom Indonesia president director Nanang Hendarno has argued that the country’s AI opportunity depends less on experiments and more on building the infrastructure and patient investment to turn them into real businesses, CNBC Indonesia reported. He was speaking on October 27 at the Danantara Indonesia Technology and Innovation Forum 2026 in a session moderated by Irwan Rei of Danantara, the state investment agency that oversees Telkom’s shareholding.

Nanang cited figures showing how hard it is to make generative AI pay. About 95 per cent of generative AI pilot projects show no impact on profit and loss, he said, only 25 per cent of companies hit their return-on-investment targets, and just 21 per cent have redesigned their workflows around the technology. At the same time, he noted that the cost of running an AI query has fallen by roughly 280 times since 2022, which changes what becomes affordable. He outlined five industries he expects AI to create or reshape in the years ahead.

He also set out what Telkom brings to that future. The group has 42.2 Gbps of satellite capacity, more than 211,000 kilometres of fibre and submarine cable, and 58 international points of presence. Its data-centre arm NeutraDC ran 48.4 megawatts of capacity as of the third quarter of 2026, with an aspiration to reach one gigawatt.

What it means in practice

Telkom CEO Says Most Generative AI Pilots Miss Profit Goals as He Maps Indonesia’s Future AI Industry — contextual photo

Telkom is packaging these assets into what it calls a sovereign digital technology initiative, made up of AI-ready data centres, an AI model hub, a national Internet of Things platform and a service called PaDiSera. The one-gigawatt target echoes earlier plans in which Telkom’s NeutraDC and Danantara mapped a 1 GW buildout, and a model hub could host local work such as GoTo’s open-source 27-billion-parameter LLM for Indonesian and regional languages.

The pilot figures also match what Indonesian businesses report about governance gaps, as in a Sumsub survey on how firms record AI decisions. Nanang stressed that AI infrastructure needs patient capital, since data centres and networks take years to pay back, which puts Danantara’s appetite for long-term returns at the centre of the plan.

What to watch next is whether Danantara commits fresh funding to NeutraDC’s expansion, which customers sign up for Telkom’s model hub, and whether Indonesian companies start moving more AI pilots into production.

← Back to AI4Indonesia