TechnologyLumenPrint articleOct 9, 2026 – 8.31amThe first wave of enterprise AI has largely been about experimentation. Organisations have tested copilots, built proofs of concept and explored where the technology might create value. The next phase is proving to be more demanding. As AI moves beyond pilots, the limiting factor is increasingly not the models themselves but whether existing networks, cloud architecture and data environments can support them at enterprise scale.That shift is exposing the limitations of infrastructure built for an earlier generation of enterprise applications. Large volumes of information must move continuously between applications, cloud platforms, data centres and edge environments, while performance, governance and security have become business priorities rather than purely technical considerations.AI at enterprise scale depends on infrastructure that can move data securely and reliably across increasingly complex digital environments. iStockFrancis Thangasamy, managing director, APAC, at Lumen Technologies, says many organisations underestimate what successful AI deployment demands.“Many organisations are approaching AI as a layer they can add to existing technology environments,” says Thangasamy. “They identify the model, build a use case and move quickly into experimentation.”That approach can produce promising early results but does not necessarily prepare organisations for enterprise deployment.Advertisement“What has changed is not the appetite for AI, but the challenge of putting it to work at scale,” Thangasamy says.Francis Thangasamy, managing director, APAC, at Lumen Technologies. “In Australia, that means organisations need to think beyond the model itself and consider how data moves across clouds, applications and existing environments, and how that is governed and secured.”As AI scales, information moves continuously between applications, cloud platforms, data centres, partner organisations and, increasingly, edge environments. Infrastructure designed around stable traffic patterns and centralised architectures was never intended to support those demands.“The key point is that AI readiness is not only about choosing the right model,” Thangasamy says. “It is about whether the organisation has the digital foundations to move data quickly, securely and reliably across the environments where the business now operates.”Beyond successful pilotsMany AI projects perform well during early testing but encounter very different conditions once they move into production.“The most difficult stage is often the move from a successful pilot to a production environment,” Thangasamy says.At that point, AI applications must interact with operational systems, live information flows and multiple technology platforms simultaneously.One of the biggest obstacles is fragmented information spread across legacy systems, business units and cloud environments.“If that data cannot be accessed, moved, secured and governed consistently, AI models will struggle to deliver timely or trusted outputs,” Thangasamy says.The challenge grows as organisations adopt multi-cloud and hybrid operating models.Many enterprises now operate across public cloud providers, on-premises infrastructure, SaaS platforms and third-party ecosystems. Each environment has its own performance profile, security controls and operating model, making consistent governance increasingly difficult.Infrastructure performance therefore becomes a business issue rather than simply an IT consideration.“As AI becomes embedded in core workflows, infrastructure performance becomes a business issue, not just an IT issue,” Thangasamy says.“If data movement is delayed, visibility is fragmented, or security controls are inconsistent, the impact can flow directly into customer experience, operational decision-making and risk management.”“For Australian business leaders, the question is ultimately not whether the technology works in a pilot. It’s whether it can deliver business value reliably at scale, while managing security, governance and operational risk,” Thangasamy says.The infrastructure underpinning AI is also attracting greater government attention. In July, the federal government established an Office of AI within the Department of the Prime Minister and Cabinet and announced new national standards covering AI and large data centres.“Australian Standards for AI strengthen Australia’s framework to make sure AI investments are on Australia’s terms and strengthen our resilience, security and economy,” Industry Minister Tim Ayres said last month.“It’s not that the models aren’t working,” Thangasamy says. “It’s that the environment around them isn’t ready to support them at scale.”Connectivity becomes strategicThangasamy says data movement, latency and connectivity sit at the centre of enterprise AI deployment.“Data movement determines whether AI models can access the information they need across the enterprise,” he says.“Connectivity determines whether that movement can happen reliably across data centres, clouds, applications and partner environments. Latency determines whether the resulting insight is useful at the moment a decision needs to be made.”For many organisations, AI strategy is becoming inseparable from infrastructure strategy.Use cases such as fraud detection, automated supply chains, digital customer experiences and operational decision-making all depend on information moving quickly while remaining visible, secure and governed.“In these environments, speed and control have to work together,” Thangasamy says. “It is not enough to move data quickly. Organisations also need confidence that data is moving securely, consistently and in line with governance requirements.”That changes the role of enterprise networking, he says.“For Australian businesses, networks are becoming part of the operating fabric for AI,” Thangasamy says.“As compute, data, applications and users become more distributed, connectivity increasingly determines whether organisations can translate AI capability into business value.“Organisations that treat connectivity as a strategic layer for performance, visibility and resilience are better positioned to scale AI effectively.”Supporting AI at enterprise scaleFew organisations are likely to rely on a single technology environment. Cloud platforms have become the default for large-scale model training, while edge computing enables decisions to be made closer to where information is created. Hybrid architectures combine those capabilities while allowing organisations to meet operational, regulatory and security requirements.“The question is not whether to adopt cloud, edge or hybrid,” Thangasamy says. “It’s how to bring them together in a way that keeps performance consistent and risk under control as AI scales across the business.”Those increasingly distributed environments are also changing the cyber threat landscape.Black Lotus Labs, Lumen’s threat intelligence and operations group, has observed attackers adapting their techniques as enterprise environments become more complex. Increased targeting of internet-facing infrastructure, together with AI-assisted automation, is reducing the time available for security teams to identify and respond to malicious activity.Thangasamy says organisations need visibility well before threats reach their internal environments.“As AI adoption accelerates, security cannot be treated as a separate workstream,” Thangasamy says. “Australian organisations need to think about resilience, visibility and governance as part of the architecture from the outset, rather than addressing them after AI moves into production.”“For business leaders, the implication is clear: AI resilience is not only about protecting applications or endpoints,” he says. “It also depends on visibility into the networks and infrastructure that connect the enterprise.”The next milestone for organisations investing in AI today is no longer successful pilots but successful enterprise deployment. Achieving that will depend in large part on infrastructure capable of supporting AI securely, reliably and at enterprise scale.To find out more, please visit Lumen.This content is provided for informational purposes only and does not constitute legal, technical, cybersecurity, business, or other professional advice. While Lumen believes the information contained herein to be accurate as of the date of publication, Lumen makes no representations or warranties, express or implied, regarding the accuracy, completeness, or reliability of the information, and readers should independently evaluate its suitability for their circumstances. Any reliance on this information is at the reader’s own risk. Services are not available everywhere. Business customers only. Lumen may change, cancel, substitute, or modify products and services, or vary them by service area, at its sole discretion and without notice.All third-party trademarks, trade names, logos, brands, products, and service marks referenced herein, and any other third-party organisations identified in this article, are the property of their respective owners. Their inclusion is for identification and informational purposes only and does not imply sponsorship, endorsement, affiliation, or approval by such parties unless expressly stated. © 2026 Lumen Technologies. All Rights Reserved.Sponsored by LumenThis content has been funded by an advertiser and written by the Nine commercial editorial team.SaveLog in or Subscribe to save articleShareCopy linkCopiedEmailLinkedInTwitterFacebookCopy linkCopiedShare via...Gift this articleSubscribe to gift this articleGift articles to anyone you choose each month when you subscribe.Subscribe nowAlready a subscriber? LoginLicense articleRead MoreLumenFetching latest articles