Amdocs builds with Cloudera, launches aOS for agentic AI

Amdocs waited two years to earn a return on its Cloudera migration, which began with an Oracle data crisis in 2014. The multi-billion-dollar telecommunications software and services provider could not scale its on-premises Oracle databases. Database capacity fell short of the volume Amdocs was ingesting. Management costs and complexity rose with every increase.

A broader market shock drove that growth. Mobile devices, social media platforms, online services and Internet of Things sensors generated new data volumes, while legacy data-preparation systems, databases and warehouses lacked the necessary scale.

Why did Amdocs choose Cloudera in 2014?

Amdocs required large-scale processing, plus security and governance for personally identifiable information (PII). It selected Cloudera’s Hadoop-oriented HBase platform for large workloads. Cloudera already had six years of history, while Databricks and Snowflake remained nascent. Robust alternatives remained scarce, and Boaz Rubin, Amdocs’ head of network software research and development, said the company chose Cloudera because it needed a very robust solution.

Rubin identified lower total cost of ownership (TCO) as the main objective. Oracle was very expensive and lacked a good scaling solution then. Oracle Exadata later improved the vendor’s offering.

Why did the migration take two years?

Amdocs used the platform both operationally and to build products for communications customers. Those customers applied exacting controls to PII and other sensitive data. Exposure could trigger regulatory violations and severe financial penalties. Cloudera also proved less user-friendly than some competing platforms.

Amdocs first built a generic tool that stored customer data as JSON files in Cloudera’s database. Customers rejected that workflow because they needed to run SQL queries. Amdocs then built a tool that mimicked its Oracle implementation. Customer acceptance followed that redesign.

Rubin described the beginning as very hard and the Oracle-to-Cloudera move as very complicated because customers performed strict work with Oracle, and two years passed before Amdocs began receiving a return on its Cloudera investment.

Gainwell Technologies, a healthcare technology provider, struggled after becoming a Cloudera customer about six years ago. Its chief innovation and modernization officer, Paddu Vedam, said user experience still needs work.

He wants easier deployment and front-end capabilities comparable to Databricks, enabling business users rather than only technically inclined development teams. Cloudera has built many of those capabilities, but Vedam sees room for improvement.

What scale did Amdocs reach with Cloudera?

After gaining experience, Amdocs built a robust reporting and data-extraction solution over the following five years. With operations still on premises in 2014, Amdocs later became one of the first Cloudera customers to use the platform in the cloud, expanding storage capacity. Cloud computing had become ubiquitous by then.

Cloudera also became the foundation for products used by communications customers managing more than 200 million subscribers.

Rubin said: “The telco market is very strict regarding TCO, so we were able to provide a very competitive solution with the right TCO thanks to Cloudera.” He later compared TCO under Cloudera with estimated costs using database capabilities from Databricks, MongoDB, EnterpriseDB and DataStax. Some offered fancier features, but none matched the TCO Amdocs required.

How does Cloudera fit into Amdocs’ agentic AI plans?

Amdocs now builds agentic AI-powered tools alongside its established data-management products. Two obstacles dominated: ensuring that data guiding AI agents remained accurate, and predicting development and deployment costs across locations, clouds and platforms. Rubin warned that satisfying differing environments can become very expensive.

By February 2026, Amdocs had overcome the initial barriers and delivered aOS, an AI-powered operating system for telecommunications customers. Cloudera still does not participate in Amdocs’ AI workflow because of regulatory, compliance and cloud-data access concerns. Amdocs instead builds and deploys AI and enables customer use of aOS inside its controlled environment.

Cloudera recently introduced Cloudera Anywhere Cloud, a hybrid data and AI platform for building and deploying agents and other applications across cloud, on-premises and hybrid settings. The product allows Amdocs to begin using Cloudera to build and manage AI.

Rubin said: “Only Cloudera provides a solution for on premises.” He believes Cloudera can serve as the on-premises foundation as enterprises react to expensive cloud services and AI tokens. He expects more companies to return to on-premises infrastructure.