- Indosat and F5 want to make managing enterprise AI a little easier
- Model sprawl, runaway costs and security risks are all in their sights
- A local edge could help telcos challenge hyperscalers’ AI dominance
Indosat wants to give its enterprise customers an AI easy button, teaming with F5 to package model access, traffic routing, security and consumption controls into a single service.
The Indosat Intelligence Suite pairs Indosat’s full-stack AI infrastructure spanning data centers, compute, connectivity and models with F5’s enterprise-ready platform. The goal is to address several enterprise AI pain points – including model selection and routing, tool sprawl, economics and infrastructure security and optimization – all in one place.
Ahmed Guetari, F5’s SVP of product management for service providers, told Fierce built-in routing selects models according to cost, quality and latency for each workload. Meanwhile, token usage is converted into credits that can be allocated across models, applications and teams, so consumption and spend are visible in one place.
There’s also a security layer that helps protect against the top large language model threats, data leakage and other emerging AI risks. This includes traceability for requests, which are authenticated, metered and logged, and tenant isolation and container security via F5’s BIG-IP Next for Kubernetes.
“The F5 framework provides AI Gateway capabilities for inspecting prompts and responses, sanitizing or redacting data, enforcing access controls, and detecting or preventing LLM attacks. It also identifies partner integrations for protections against threats such as prompt injection and jailbreaks,” Guetari said. “Together, these controls are intended to let enterprises move AI workloads into production with consistent security and auditability.”
A local alternative
Data residency is also central to the pitch, with local processing offered via the platform’s Sahabat-AI and Indosat-hosted tiers.
“There are regulated sectors in which data residency and security often decide whether AI can move beyond pilots,” Guetari said. “Banking and financial services, government, and telcos are important examples of sectors concerned with data residency, confidentiality and control over sensitive information.”
Interestingly, it’s this kind of local processing that can be a key differentiator for telecom operators looking to get into an AI game that is today dominated by hyperscale cloud companies. And it has the potential to be a big money-maker, too, given McKinsey forecast the sovereign AI market could be worth as much as $600 billion by 2030.
“Hyperscalers remain strong on scale and model variety; the local provider’s contribution is the in-market processing option, data sovereignty, and the integrated commercial and governance layer around it,” Guetari noted.
He added that its work with Indosat is a model other telcos could easily replicate to offer AI capabilities beyond basic compute hosting.