Telco AI partnerships point to importance of national data security

A crop of new AI partnerships demonstrates the unabated confidence of telco CEOs in the technology's capabilities to improve their businesses. It also highlights the importance of national data security measures when choosing - or competing with - technology partners.

Orange, for example, this week announced it is deepening its relationship with Google Distributed Cloud (GDC). Orange “sees enormous value in AI across every dimension of our business,” according to the company’s CEO Christel Heydemann in a statement.

The partnership will enable Orange to run sensitive network data and AI workloads on-premise or locally, in line with national regulatory demands. As a result, Orange expects “significant improvements in network planning and design through the automation of reporting, classification, and analysis,” which will help with capacity planning and root cause analysis.

Orange also hopes to improve operational efficiency by running Gen AI models on-premise in an environment that is “integrated into similar Vertex AI services on Google Cloud” (Vertex AI is Google's development platform for building and using Gen AI). And like other telcos it is tapping into AI to improve customer experience. In Orange's case this will include delivering personalized recommendations for phones and services, as well as potentially deploying Gen AI-based speech recognition, including in countries without Google Cloud presence.

At the same time, however, Reuters reports that Orange is one of 15 European companies to criticize a draft EU proposal that would allow US technology companies such as Google or AWS to bid for sensitive cloud computing contracts in the region without a European partner. Currently US technology companies need a joint venture or cooperation with an EU-based company to qualify for the highest EU cybersecurity label, notes the Reuters article.

Deutsche Telekom is also reported to have contested the draft proposal. Like Orange it has a large enterprise service arm and sees an opportunity to provide sovereign AI and GenAI B2B services. In March, for example, Deutsche Telekom announced it had won a B2B contract to provide access to a secure tool for Gen AI, which allows the customer to retain control of its data in accordance with GDPR and German data protection law.

EU regulation is among the most stringent when it comes to protecting citizens’ data. But operators in countries outside the region are also investing in sovereign data protection as AI usage takes off.

In Indonesia, for example, Indosat Ooredoo Hutchison (IOH) said in March it is ready to start using Nvidia infrastructure to develop national AI capabilities. This month the Indonesian government went further and stated that Nvidia and IOH plan to build an AI center this year in Central Java in 2024, at a cost of $200 million, according to reports.

Vikram Sinha, President Director and Chief Executive Officer of IOH underscored the importance of sovereign AI in the March announcement with Nvidia: “We are one step closer to our larger purpose of empowering Indonesia and leading the charge in sovereign AI, driving innovation, economic growth, and prosperity for future generations.”

The company plans to use Nvidia’s infrastructure to enable “cloud network acceleration, composable storage, zero-trust security, and GPU compute elasticity in hyperscale AI clouds.” IOH also said it will use NVIDIA Tensor Core GPU, based on the Blackwell platform, “to deliver a massive speedup to inference workloads, making real-time performance possible for resource-intensive, multitrillion-parameter language models."

IOH is one of a number of CSPs that are incorporating NVIDIA’s technology into their sovereign AI plans. Others include Reliance industries, the parent company of Jio, Swisscom, Singtel and France's Iliad. Like its European peers with large enterprise service units, Swisscom Group’s investment is targeting the market for the secure, regulated handling of data.

It makes sense for Nvidia to encourage telcos to use its processers and software when developing sovereign AI services. Cloud hyperscalers, however, can afford to go it alone, regulation permitting. Microsoft for example has announced it will spend US$2.9 billion on AI and cloud infrastructure in Japan. As part of the agreement, signed with Fumio Kishida, Prime Minister of Japan, Microsoft will provide AI training to more than 3 million people over the next three years, open a Microsoft Research lab, and collaborate on cybersecurity with the Japanese government.