The narrative surrounding Asia’s digital economy has fundamentally shifted. In 2026, the conversation is no longer solely about the promise of artificial intelligence (AI) but about the physical and digital infrastructure required to deliver it.
As the region moves beyond experimentation, the question for enterprises is no longer if they will use AI, but whether they possess the underlying infrastructure to make it work at scale.
This new reality is being forged by the convergence of three powerful forces: the urgent need for AI-ready infrastructure, the fragmenting landscape of data sovereignty, and the escalating demands of cybersecurity.
Across Southeast Asia, this shift is tangible. The 2025 edition of the e-Conomy SEA report noted that more than 4,600 MW of new data centre capacity is planned for the region – a 2.8-fold increase that outpaces the Asia-Pacific average.
This build-out is being driven by hyperscalers responding to a surge in compute-intensive workloads. However, rapid expansion brings real operational challenges. According to CBRE, AI-focused data centres now require more than double the power density per server rack compared with traditional facilities – a specification that much of the region’s existing infrastructure was not designed to meet.
This sets the stage for a new era defined by a search for strategic, high-performance connectivity. For Max Parry, VP of growth and emerging markets at Equinix Asia-Pacific, the answer lies in a fundamental shift away from the public internet.
He defines private interconnection as “a direct and private connection between two parties for data transfer,” contrasting it with the unpredictable nature of the public internet, where “organisations do not control the exact path their data takes.”
By leveraging these direct and private paths, organisations gain inherent privacy for data sovereignty compliance, superior reliability for high-bandwidth applications such as distributed AI, improved security without intermediary transport providers, and lower costs by reducing reliance on IP Transit services.
Data Sovereignty: The architect of fragmented networks
In a geopolitically fragmented Asia, data sovereignty has evolved from a compliance item into a core driver of IT strategy. Governments are tightening rules on data privacy and location, particularly for banking, healthcare and e-commerce.
As Parry notes, the regulatory requirements “especially relevant for banks and e-commerce players facing data sovereignty pressures… can be addressed using private interconnection platforms, avoiding compliance risks and operational penalties from inadequate connectivity.”
This regulatory pressure is accelerating what Gartner calls “geopatriation” – the relocation of workloads from global hyperscalers to local or regional sovereign options to mitigate geopolitical risk. Gartner forecasts worldwide sovereign cloud IaaS spending to reach US$80 billion in 2026 (with Mature Asia-Pacific recording 87% growth) and predicts that 20% of infrastructure cloud workloads will shift from global to local providers.
For multinational CIOs, Parry advises to think “locally, especially for multinationals: address individual jurisdictional requirements where operations exist by deploying distributed architectures tailored to each local regulatory environment.”
He asserts that cloud adjacency via private interconnection keeps paths short, reduces cross-border risks and supports an “applications-to-data” approach that keeps regulated data local while enabling compliant multi-cloud AI.
“Interconnection supports this by enabling distributed, cloud-adjacent setups that align local data residency with application needs, ensuring short paths for efficient multi-cloud AI without sovereignty violations.” Max Parry
The AI imperative: Low latency, high throughput
As AI workloads migrate to hybrid environments, the demand for performance is reshaping network priorities. AI services—from text-based chatbots to future video-based agents—have a low tolerance for delay.
Parry explains that to cut latency for real-time AI, “ultra-high-performance connections are essential... which requires shortening the effective distance between users, data, and the various technology platforms used to deliver the service.”
This requirement for proximity is driving the adoption of distributed AI architectures. IDC predicts that by 2027, 80% of enterprises will deploy distributed edge infrastructure to improve the latency and responsiveness of AI applications. The goal is to shorten the path between end users, workloads, and the data they interrogate.
Equinix Fabric, the any-to-any virtual private interconnection platform, delivers this by providing access to Equinix’s 10,000 customers, over 2,000 networks and over 3,000 cloud and IT service providers.
New capabilities, such as Fabric Cloud Router, enable dynamic multi-cloud routing, while Fabric Intelligence delivers AI-driven automation and natural-language configuration for low-latency, high-throughput flows between the edge, the core, and multiple clouds.
Subsea meshes and the rise of agentic AI
The physical backbone of this connectivity revolution is being laid beneath the waves. In an agentic AI world – where workflows accelerate with minimal human intervention – networking faces a step change in requirements.
Parry warns that “networking must become direct and private, less reliant on the public internet, which lacks the uptime reliability and performance guarantees needed for such workflows.” Networks will also integrate more AI-based technologies, becoming self-healing, self-optimised and self-provisioning.
This is driving massive investment in subsea cable infrastructure across Southeast Asia’s archipelagos. The U.S. Trade and Development Agency is supporting the SCNX3 cable system, a critical project connecting India with Singapore and other Southeast Asian hubs, designed to provide the resilient pathways needed for AI and cloud services.
Similarly, Telin and Singtel are developing the INSICA cable system between Singapore and Batam, slated for 2026, to support the surge in data centre traffic with “high bandwidth and low latency for commercial-scale and real-time applications”.
Parry notes that Equinix is closely tied to this market, with “much global subsea interconnectivity terminating in Equinix data centres.” This convergence of undersea capacity and on-land interconnection hubs is creating the dense, resilient meshes necessary for the region’s AI future.
A new model for network operations
As networks become more autonomous and AI-driven, interconnection itself is evolving. Parry outlines three major shifts by 2027: far greater use of private interconnection over public internet paths; decentralisation from a few mega-hubs to more local, smaller interconnection points beyond traditional metros; and AI-enhanced networks that are self-healing, self-optimising and self-provisioning.
These capabilities directly address the cybersecurity backdrop Gartner highlights for 2026: chaotic AI adoption, geopolitical tensions and regulatory volatility are accelerating threats, making pre-emptive, resilient architectures essential.
Private interconnection inherently improves security by eliminating intermediaries, supports regulatory compliance and delivers the agility banks, broadcasters and e-commerce players need to scale on demand.
For technology and operational leaders in Asia, the message is clear: private interconnection is no longer a nice-to-have technical feature but a strategic imperative. It delivers the low-latency performance, sovereign compliance, cyber resilience and on-demand agility required to turn AI ambition into competitive advantage in 2026-2027.
Those who act now – by embracing platforms and planning distributed, cloud-adjacent architectures – will be best positioned to thrive as the region’s digital economy accelerates.
