Data centres continue to be misrepresented and misunderstood, according to Digital Realty’s chief datacenter technology and engineering officer, Lex Coors.

Taking the stage for his closing keynote speech at Datacloud Energy Europe last week, Coors explained that the industry is currently wrestling with a “perception gap” and that, instead of megawatts, “digital value” should be the main message.

He shared the top things to remember about the data centre industry currently – where progress is being made and how it can be generated further.

Data centres: The story so far

The data centre industry prides itself on delivering essential services for banking, healthcare, social media, government, AI and more – but this, Coors said, is often overlooked.

“Societal function matters,” he said during his address. “Protestors are demonstrating against data centres, we should not ignore that nor blame the public.”

He added: “As an industry, we made a mistake ourselves. For many years, we proudly said: ‘we built another 200 MW data centre’ – but that is the wrong story. We are not just buildings; we are building the place where your [essential services] live.”

Coors explained how we are living through the construction of a totally new economic system, which means the responsibility lies with the data centre industry to enable predictability and consistency.

“Today, growth of the digital infrastructure is constrained. Not by demand – demand is unlimited, but by permitting, grid access and fragmented regulation,” he said. “Operators, hyperscalers and customers all say the same thing; we need decisions that are made in years, not decades.”

Coors’ six measures of success

Energy supply is another pressure point

According to Coors, the problem today is not total power availability but having power in the right place.

“Data centres cannot be built anywhere, they must be built where the fibre lands, where submarine cables arrive and where terrestrial fibre highways run,” he explained. “This is where highly connected campuses form the on-ramps and off-ramps of the hyperscale cloud.”

This, he explained, is why the EUDCA is co-leading the grid integration programme with ENTSO-E, supporting the data centre industry’s need to become grid participants.

“We are no longer just power consumers,” he added. “Future data centres may need to operate differently – providing flexibility, stability and instantaneous support through on-site storage.”

He added: “We also believe the first and most logical step for grid flexibility.”

Water usage

Water usage is another area where perception and reality are not always aligned in the data centre industry, Coors explained. He suggested that more than 90% of the data centre industry across Europe has committed to stay below 0.4 litres of water per kWh, under the terms of the Climate Neutral Data Centre Pact.

“That effectively eliminates traditional cooling towers and limits water use to very efficient adiabatic systems,” he said. “Many operators are going even further. Adiabatic cooling is only used during the hottest hours of the year – sometimes only a few hours per day in summer.”

Heat reuse

Often referred to as a missed opportunity for data centres, Coors said that heat reuse is another area where the industry is now working together to make it more of a reality. Under the EUDCA, alongside Euroheat & Power, Open Compute and the Net Zero Carbon group, he said the industry is development a marketplace model.

“Data centres make heat available. District heating networks can take it when they need it,” Coors said. “Not theory – real infrastructure cooperation.”

Efficiency and reporting

The European Union is now moving forward with new requirements – reporting rules and possibly a labelling scheme in this area. Coors suggested this could become a global model, with regional differences based on climate.

“It makes sense. A data centre in Finland should not be measured the same way as one in Spain,” he said.

Energy supply

When it comes to energy supply – nuclear and SMRs in particular, plenty of companies across the industries are starting to invest in these solutions as a sustainable alternative for the future.

However, Coors said there is a danger of progress being hindered because processes are too complicated.

“Europe is not helping itself. There are too many designs, national programmes and too much competition inside the European Union,” he added. “If we want success, we need one programme. One shared investment. Shared benefit.

“That is how Europe becomes strong.”

AI demand

AI continues to be transformative across the industry and disrupt the global technology industry. Just a few years ago, Coors said GPU cooling required 16 to 18 °C water supply temperatures. Today, the same vendors are asking for 35 to 40 °C supply temperatures.

“That changes everything,” Coors said. “It may reduce compressor cooling and increase efficiency and create the perfect conditions for heat reuse – because the outlet temperature now reaches 50 to 60 °C. Exactly where district heating needs it.”

Confronting perceptions and fuelling success

The data centre industry is one that faces many reality checks. But for Coors, this leaves space for real transformation.

“We are learning to explain what we do better,” he explained. “We are learning to design differently – working with the grid, rather than against it – and building infrastructure for society, instead of just ourselves.”

He added: “If we do that right, the data centre industry will not be seen as a problem, but as what it really is; critical infrastructure for the digital age.”

Related stories

Speed, sustainability and scale: How AWS seeks to redefine European data centre policy

Dame Dawn Childs on power, AI and mapping out Europe’s data centre future

Check out our new State of the Sector: Data Centre report

Data centres are evolving beyond their traditional role as passive energy consumers and instead becoming embedded in the broader energy ecosystem. As concerns mount over grid instability and the decentralisation of energy, the data centre industry is having to come up with new ways to be ‘good energy citizens’ and define a future that is more sustainable by design.

Europe’s energy system is currently undergoing significant changes. With the grid struggling with rising demand and AI consuming vast amounts of power, the onus is on data centres to act.

“Grids that traditionally were made up of fewer, larger generators now have millions of assets contributing to them,” said Katie Davies, head of energy and infrastructure policy at techUK, at Datacloud Energy Europe last week.

“Alongside decarbonisation, that brings system challenges around the intermittency of renewables and grid congestion.”

Confronting grid flexibility challenges

Grid congestion is a significant problem across Europe, as buildouts have not kept pace with demand and the rapid expansion of AI infrastructure. This is placing overwhelming pressures on ageing infrastructure and, with data centre electricity demand expected to double, operators and policymakers know they need to work together to keep everything running smoothly.

“In the last couple of years, I’ve seen a shift of focus. Now, when we talk about flexibility and demand response, there is a lot of focus on industrials,” shared Gilda Amorosi, senior programme manager at Microsoft. “However, as a sector, we are at a crossroads.”

She added: “[We need] an inclusive definition of flexibility that creates a menu from which operators can choose, not a mandated approach.”

The concept of on-demand flexibility is not new, but data centres haven’t traditionally been involved in it. At Datacloud Energy Europe, conversations very much turned to how the industry should be more active in supporting the grid.

“What we see with hyperscalers is that grid operators are asking them to disconnect from the grid on the day-ahead market with around two hours’ notice,” said Francesco Ciccola, regional sales director for Europe at Mitsubishi Power Aero LLC. “When that disconnection occurs, they still need power inside the data centre.”

The answer, he argued, lies in a combination of technologies – gas turbines capable of coming online in under ten minutes, battery storage for frequency response and AI-enabled system management – rather than any single “silver bullet”. However, he described himself as ‘clear-eyed’ about the limits of what can realistically be expected moving forward.

“It’s extremely ambitious to expect that, in the short term, we can have a backup or storage solution on site that can cover the full capacity and power needs of the data centre,” he added, given that backup diesel generators are not a sustainable solution and long-duration battery storage involves high cost.

For Neal Kalita, senior director of global power and energy at NTT Global Data Centres, flexibility comes with honesty. His company has participated in grid flexibility programmes in the US, but he said the experience wasn’t the most positive.

“We supported the grid during storm events and one of the things we noticed was that our scope one emissions went through the roof because of the diesel generation involved,” he explained. “Unfortunately, there was no mechanism to account for that or provide a penalty offset.”

Build the energy ecosystem of the future

Looking ahead, some data centre operators are pursuing an ambitious vision: becoming genuine nodes in a local energy ecosystem, generating, storing and redistributing energy in ways that benefit the communities around them.

Kalita described an initiative underway at NTT’s largest site in Germany that illustrates what this can look like in practice. He said: “We’re embracing an offshore wind farm being developed to the east of our site, connecting it up with a gas connection that we already have, fuel cells, and then combining all of that with our data centre load.”

He explained the plan involves an on-site electrolyser powered by behind-the-meter wind energy, producing green hydrogen to feed back through fuel cells. It then distributes the resulting heat to the local municipal network.

“They can take it not just one or two kilometres, but potentially 50 kilometres, because the quality and volume of the heat is significant,” he said. “That is the kind of integrated energy ecosystem we can actually create.”

However, it remains an ambitious model and Kalita acknowledged that green hydrogen at scale remains some way off.

“There isn’t a commercially available electrolyser on the market that can produce hydrogen at a truly industrial scale, and that is extraordinary,” he said. “No planet, no data centre, no digital. Simple as that.”

Across Europe, locations like the Nordics are able to provide a natural ground for this type of integrated approach. Ciccola advocated for the positive potential of Northern Europe, given its vast land, favourable weather, cold temperatures and renewable energy potential.

“That is potentially the perfect place to have a greater concentration of sustainable data centres,” he added.

As Europe seeks to be more competitive in the digital infrastructure space, ensuring grid security and stability is a vital mission. The defining question is whether regulators, policymakers, data centre developers and grid operators can align to turn ambition into infrastructure.

“We will never be able to compete with the Americans or Chinese on scale alone, but what we do have is an integrated grid,” Kalita argued.

“A continent that develops digital infrastructure without destroying the planet will not just have a competitive edge – it will have a moral one.”

Related stories

Collaboration, coordination and acceleration: Unpacking the European grid perspective

Datacloud Energy Europe: ‘Clarity and partnerships’ are the answer to sustainable innovation

Can Europe’s AI gigafactories close the gap with the US?

The event, hosted by Huawei alongside GSMA and GSMA Intelligence, centred on a clear theme: Building an Agentic Network through Device–Network–Business Synergy.

The discussion reflected a growing consensus across operators and technology providers that the 5G core network is entering a new phase. AI is no longer an experimental layer added on top of networks; it is becoming a foundational element shaping how services are delivered, managed and monetised.

The rise of the Agentic Core

During the summit, Huawei introduced its AgenticCore solution, designed to embed AI across the entire core network stack. This includes mobile data, voice services, operations and maintenance (O&M), and telco cloud infrastructure.

According to George Gao, president of Huawei’s Cloud Core Network Product Line, the goal is to equip operators with the ability to evolve both their networks and their business models.

AI enables three key capabilities: real-time experience awareness, global evaluation and coordination of resources, and intelligent interaction and execution. Together, these form what can be described as a “network brain”—a system that continuously analyses user experience and automatically adjusts network behaviour.

This creates the foundation for a closed-loop model in which user experience becomes measurable, manageable and ultimately monetisable. Operators can define service quality more precisely, tailor offerings to individual users, and ensure consistent delivery across different scenarios.

The release of the GSMA Intelligent Packet Core Foundry White Paper 2.0 at the event further underlined the industry’s commitment to aligning AI innovation with the evolution of the core.

This vision is already being translated into practical frameworks that define how intelligence is applied across the network.

AI-driven transformation across the core

The Agentic Core vision can be understood through four key initiatives, each applying AI to a different domain of the network.

Mobile Internet + AI focuses on transforming user experience and business models. By embedding intelligence directly into mobile data services, networks can dynamically adapt to user needs in real time. This enables more personalised, differentiated services and allows operators to move beyond traditional connectivity toward experience-based offerings.

Calling + AI redefines voice as a service platform. AI-powered capabilities such as real-time translation, transcription and contextual assistance can be integrated directly into calls, turning voice into a gateway for AI-driven services. This reflects a broader shift in thinking. Rather than treating voice as a legacy service, operators are beginning to see it as a platform for new forms of interaction and value creation. Industry analyst Andy Hicks highlighted that telecom operators have historically prioritised migration, moving from 2G and 3G to VoLTE and 5G, rather than reimagining what voice could become.

He argued that a converged core approach, where multiple generations of services operate within a unified architecture, should not only support legacy transitions but also enable future innovation. In this context, calling becomes a critical entry point for future digital interactions.

O&M + AI is driving the move toward autonomous networks. AI systems can detect faults, optimise configurations and manage performance through closed-loop automation. Huawei’s intelligent O&M framework represents a step toward Autonomous Network (AN) Level 4, where networks can operate with a high degree of independence.

The first phase introduces intelligent assistants capable of handling faults and configuration changes automatically. Later stages evolve toward hierarchical autonomy, where different network domains operate independently but in coordination. This reduces the need for manual intervention while improving efficiency, reliability and scalability as networks grow more complex.

AI-centric infrastructure underpins the entire transformation. Telco cloud platforms are evolving from cloud-native to AI-native, supporting GPU orchestration, real-time inference and generative AI workloads.

At the same time, infrastructure is being redesigned to integrate compute capabilities directly into the network, enabling AI processing at scale. This allows operators to host and run advanced AI services within the network itself, creating an environment where devices, applications and AI agents can interact seamlessly.

AI’s influence now spans both operations and infrastructure. Rather than simply transporting data, networks increasingly host, manage and orchestrate AI-driven processes. As autonomous capabilities mature and AI-native platforms scale, they will function as intelligent systems in their own right, coordinating interactions between users, services and machines in real time.

Preparing for the 6G horizon

Looking beyond 5G, the Agentic Core aligns closely with early visions of the 6G ecosystem.

In this future, networks will support ubiquitous connectivity for AI agents operating across devices, platforms and digital environments. The core network could act as a central exchange point, enabling seamless communication between machines, applications and users.

As 5G-Advanced continues to roll out, the role of AI in the core is expected to expand rapidly. The Agentic Core concept suggests that future networks will be defined not just by connectivity, but by intelligence embedded throughout the system.

For operators under pressure to unlock new revenue streams while managing increasing complexity, this shift may prove decisive in defining the next generation of telecom business models.

A big question telecoms operators face around the world today is how best to integrate AI into their networks. It is not about whether they will do this, but how efficiently and deeply they will manage it and incorporate AI into their practices.

Another question is how operators will ensure that AI is built to adapt to local realities and needs in different countries. While it is difficult to forecast the exact path ahead, insight is increasingly emerging from major operators on some of the steps they might take to meet these needs.

One of these is Indonesia’s Indosat Ooredoo Hutchison. At a round table event at this year’s Mobile World Congress in Barcelona, CEO Vikram Sinha discussed how the operator is undergoing a transformation from a traditional telco into an AI-driven technology company.

He reiterated Indosat’s belief that pushing for digital sovereignty rather than relying on foreign-built AI models is not optional, but a technological imperative for a country seeking to shape the future and empower its own citizens.

The company believes that this can also be a central prong in fast-tracking the country’s ‘Golden Indonesia’ strategy, which foresees the country becoming a high-income nation by 2045 and the world’s fourth-largest economy, enhancing productivity and driving up GDP through digital transformation.

The ‘great equaliser’

Setting the scene, Sinha cited how Indosat has come a long way towards its vision in just the past four years, since it merged with Hutchison 3 Indonesia and since Sinha took the reins at the company. Creating a huge operator, with around 100 million mobile subscribers today, the company’s AI strategy represents an interesting case study in a vast country with many remote areas and a population of around 290 million.

Sinha outlined how Indosat Ooredoo Hutchison’s goal is to become ‘AI-native’, making the technology a core element of the operator’s activities by “embedding AI in everything we do”. Forming part of this is Indosat’s belief that the technology should be for everyone and that it can help democratise intelligence. “We very strongly believe that AI is a great equaliser,” said Sinha.

A key part of Indosat’s journey is the development of app-based AI platform Sahabat-AI, which the company initially unveiled in November 2024 and has now officially launched in partnership with the government’s Ministry of Communication and Digital Affairs. The platform provides multi-model and multi-modal capabilities, allowing users to move easily from text to images and video, conduct smart searches, create content and receive coding assistance.

With an ethos that Sinha describes as “made in Indonesia for Indonesians” and aimed at strengthening the country’s digital sovereignty, the service is built as a digital companion to be open to everyone from students and creators to businesses and public institutions without its use requiring a technical background. “Our focus is on how this becomes a sovereign play for Indonesia,” said Sinha.

Local focus

To meet local needs and help overcome the digital divide, Sahabat-AI has been created with the use of an open-source large language model to support Indonesia’s multiple regional languages.

It also aligns with the country’s cultural values, social norms and ethical standards, bringing together research institutes, universities, government agencies and other partners in an open ecosystem to unlock new paths for socioeconomic growth and innovation, while helping to preserve local cultures and dialects amid the global surge in AI development.

At the round table, Sinha cited uses such as the likes of healthcare, whereby Sahabat-AI can act as a co-pilot for medical staff in communication and patient administration. He noted challenges with the shortage of doctors and nurses in Indonesia that the platform can help to overcome, while the platform can aid health in remote areas as well.

In addition, it can aid in fields such as education, acting as a personal tutor and lesson planner for teachers and students, and in sustainable development. Future uses include applications in agentic AI and citizen services.

“We want to build something that solves real problems, and that helps with linguistics and culture,” said Sinha. “We also want to make sure that citizen services are as accessible to people as possible.”

‘North Star’

Sahabat-AI is one part of Indosat’s wider drive to integrate AI throughout its operations, enhancing Indonesia’s position to put itself at the forefront of the technology’s growth.

The company had earlier outlined this so-called “AI North Star” mission at its Capital Markets Day in 2024, citing three main pillars for empowering Indonesia through the technology and becoming a full-stack sovereign AI platform. These comprised becoming an AI-native telco, an AI tech co and an AI ‘nation shaper’.

As part of the drive to shape this mission, Sinha referred at the round table to the AI-RAN Research Center that Indosat established last year in Surabaya on the island of Java in collaboration with Nokia and NVIDIA.

The research centre, which the company describes as a “smart connectivity innovation hub”, is designed to develop AI-powered radio-access networks – housing research into Sahabat-AI as one of its projects, alongside work on physical and visual AI.

The associated AI-RAN infrastructure allows high-performance software-defined RAN and AI applications on the same infrastructure and connects to Indosat’s existing AI Factory, creating a distributed platform to bring AI closer to end users as workloads shift to the edge.

Through the research centre, Indosat and its partners are also seeking to set up an ‘AI Grid’, which will connect the AI Factory with new AI-RAN hubs across Indonesia that can host applications in any location from data centres to the distributed 5G network.

Training for the future

Furthermore, the research centre will see a new generation of AI and telecoms specialists being trained and upskilled through hands-on learning, mentorship and real-world experimentation, boosting knowledge transfer and innovation, and creating a workforce geared towards the future of advanced communications.

Indosat sees this focus on people as one of the key requirements for the country to develop AI into the future, with Sinha citing the need to invest in human resources to unlock the full potential of the technology. At the end of the day, all the developments in AI are “meaningless” if the company is not able to turn them into real value for Indonesia’s people, he said.

In a similar vein, Sinha emphasised the importance of fostering close partnerships to advance AI, which have been central to Indosat’s AI journey to date. These include the link-up with NVIDIA to power distributed AI infrastructure using its GPUs and Nokia on advanced RAN technologies.

In addition, the company has established a long-term partnership with Google Cloud to aid AI integration and a strategic collaboration with Ericsson and Google to consolidate its Business Support System onto a single cloud-based platform, helping accelerate AI adoption at scale.

Cross-sector push

Combining these partnerships with the cross-sector collaboration with government and research institutions, and businesses in multiple fields, it is clear that Indosat views the development of AI as a whole-society, ecosystem-focused endeavour. “I think one of the biggest risks any company can have is not having strong partnerships based on the right principles,” said Sinha. “We want to cultivate those partnerships.”

He added that Indonesia has favourable conditions when it comes to land, water and power, an abundance of which are essential prerequisites for AI to succeed in the world of today.

Data centre capacity in the country is also growing to support the surge in demand, with forecasts that it will rise from 0.3GW in 2024 to over 1GW by 2030 – by when AI workloads will account for almost a fifth of that capacity. However, investment in resources to accelerate this and to develop transformative use cases is still needed.

If that can be achieved, it may lay the ground for Indonesia to maximise its impact in ensuring AI sovereignty and AI for all. “We are in a place where we have started to make a real difference,” said Sinha.

RELATED STORIES

‘It’s not humans vs machines’: Amazon’s AI investment raises questions over jobs

Could Oracle cut 30,000 jobs to fund AI data centre growth?

Amazon confirms 14,000 job cuts as AI shifts corporate roles

Salesforce layoffs: The quiet job cuts behind a very loud AI boom

The company generated CNY880.9 billion in revenue and CNY68 billion in net profit, broadly in line with forecasts. Huawei invested CNY192.3 billion in R&D during the year, equivalent to 21.8% of annual revenue, bringing its total R&D spend over the past decade to more than CNY1.3 trillion.

“In 2025, Huawei’s overall performance remained steady,” said Sabrina Meng, rotating chairwoman at Huawei. “I would like to thank our customers for your ongoing trust and support. Thanks also to consumers for choosing Huawei, as well as suppliers, partners, and developers around the world for working with us.

“Of course, we couldn’t do any of this without the support of every Huawei employee. Thank you for your hard work, and also your families for their steadfast support.”

Huawei said the sustained investment is focused on strategic domains including connectivity, computing, cloud, devices and AI, as it seeks to reinforce its competitiveness across the full technology stack.

While its core connectivity business saw modest growth amid industry investment cycles, the company highlighted continued momentum in computing and AI, while cloud performance softened. The company also reported strong growth in its intelligent automotive solutions business.

Looking ahead, Huawei said it will continue to prioritise R&D to integrate AI and security into its products and networks, while building out ecosystem platforms such as Ascend, Kunpeng and HarmonyOS.

Meng added: “We are moving toward a future that is full of uncertainty, so we have to remain true to our strategy and maintain strategic focus. We will translate strategy to execution, keep cultivating the developer ecosystem, and pursue high-quality development.”

RELATED STORIES

Huawei AI chip gains orders from ByteDance and Alibaba amid China’s push for semiconductor self-sufficiency

Huawei’s vision for the future of data storage in the AI era

Wednesday marks the fiftieth birthday of Apple, a once unassuming technology organisation that quickly became one of the most recognisable industry giants in the world.

Apple’s rise to success has shaped both industry and culture, with its computers and smartphones quickly becoming embedded in the mainstream. Its success led to greater inventions such as the mobile phone app, Siri voice assistant and the Apple Watch.

But just how revolutionary was Apple? And where is the company going next?

From a garage to Silicon Valley: Apple’s story

Apple was founded on 1 April 1976 by Steve Jobs and Steve Wozniak in the garage of Jobs’ parents’ house in Los Altos, California. Although Wozniak later referred to the founding story as a “bit of a myth”, the early beginnings of Apple saw the garage being used for assembly, testing and initial operations of its hardware.

The company became well-known for its Mac computers, but perhaps its most memorable revolution came at the advent of the iPhone. Essentially a computer packaged as a phone without a keypad, the iPhone sent shockwaves across the technology industry, with some even believing it would never take off.

Developing additional versions of the iPhone, in addition to the iPod, Apple Watch, MacBook laptop, AirPods and more, Apple has well-established itself as a leader in the consumer technology market.

Setting new benchmarks in innovation, Apple also transformed the capabilities of modern computing – in addition to establishing a marketplace ecosystem for third-party developers to strengthen its position even further.

Overcoming the AI laggard reputation

While Apple remained ahead of the game for many years, it now faces the challenge of remaining dominant in a global market with a large appetite for AI. As competitors Microsoft, Google and Dell spend billions of dollars to ensure their AI leadership, Apple has often been criticised for its slower approach to the technology.

The iPhone company has struggled to make as much progress in AI, with analysts in the past arguing that they perhaps haven’t put as much effort into the technology. Apple is well-known for facing several significant challenges with the technology, including its delayed upgrades to Siri and the rollback of features like its AI-generated news summaries.

Notably, the ‘ChatGPT’ moment in late 2022 was reported to blindside Apple executives and force the company to rethink its approach. Speaking at the time, CEO Tim Cook said generative AI (Gen AI) had “great promise” but had some potential dangers.

To perhaps combat these struggles, Apple has shaken up its AI leadership in recent times. Notably, the company replaced its vice-president of AI with a top Microsoft executive and machine learning expert, Amar Subramanya, at the end of last year.

At the time, it was argued that perhaps Subramanya’s role could be to help Apple catch up and position the company more as an AI competitor on the global stage.

Although slow to realise its potential, Apple shows signs of looking to grow its AI abilities. The company announced a multi-year partnership with Google in January 2026, under which the next generation of Apple Foundation Models will be based on Google’s Gemini models and cloud technology.

Speaking at the time, both companies said these models will help to power future Apple Intelligence features, including a more personalised Siri experience.

“After careful evaluation, Apple determined that Google’s AI technology provides the most capable foundation for Apple Foundation Models and is excited about the innovative new experiences it will unlock for Apple users,” their joint statement said.

Apple also remains one of the most valuable companies in the world, with Reuters reporting that continued strong demand for its iPhones will lead the company to post sales of US$465 billion in its ongoing fiscal year ending in September.

While its hardware remains a success story, time will tell if Apple becomes an AI leader.

Related stories

OpenAI poaches Apple’s director of data centre engineering

Why Apple is planning to increase AI and data centre spending

Apple co-founder Steve Wozniak pushes for AI transparency

While Nvidia has built its dominance on GPUs, the latest deal highlights growing pressure on data centre networks as AI workloads scale.

Bandwidth, latency and energy efficiency are emerging as critical constraints, pushing networking and optical technologies to the forefront of infrastructure investment.

The partnership will see Marvell’s custom silicon, optical interconnects and data centre networking capabilities integrated more closely with Nvidia’s platforms, including its NVLink ecosystem.

This is expected to improve how AI systems move vast volumes of data between processors, a key challenge in large-scale deployments.

As hyperscalers ramp up spending on AI infrastructure, demand is rising not only for compute but also for the underlying connectivity that links chips, servers and data centres. Marvell’s expertise in data centre interconnect and silicon photonics positions it as a strategic partner in addressing these bottlenecks.

The move also reflects a broader shift across the industry, where the ability to efficiently transport data is becoming as important as processing it. For operators and infrastructure providers, this trend is likely to drive increased demand for high-capacity optical networks and advanced interconnect solutions.

By investing in Marvell, Nvidia is extending its influence beyond processors and into the wider AI infrastructure stack, ensuring it remains central as the market evolves.

RELATED STORIES

Mistral raises $830m to build Nvidia-powered AI data centre

Nvidia passes $5tn valuation amid AI infrastructure surge

AT&T, one of the leading telecoms companies in the US, hopes to deliver $1 billion in cost savings and system upgrades in exchange for faster government approvals.

As reported by The Wall Street Journal, company chief executive John Stankey agreed to the new terms for the FirstNet programme last week, after months of talks with Commerce Secretary Howard Lutnick. The talks first started after a 2025 executive order by US President Donald Trump called for federal agencies to scrutinise all contracts.

“This agreement-in-principle … reflects AT&T’s ongoing dedication to our ⁠public-private ​partnership,” said AT&T’s president of public sector, Wes Anderson, as reported by Reuters.

AT&T was awarded the 25-year contract to build FirstNet in 2017, years after a federal commission recommended the setup of a first-response system in the wake of the 9/11 attacks.

FirstNet helps first responders, including medical personnel, firefighters and police officers, communicate vital information on a single network. It is currently used by roughly 31,000 US agencies.

The telco is responsible for maintaining FirstNet nationwide, even in areas where service isn’t commercially viable, to ensure first responders are able to respond to emergencies faster. It gives priority to emergencies when networks can become overloaded and deploys temporary infrastructure when cell towers and telecom cables can be destroyed.

Officials in the commerce department have requested greater oversight into how the programme is run and its partnership with AT&T, according to The Wall Street Journal report, suggesting that internal watchdog reports have criticised its management.

This news comes shortly after AT&T launched OneConnect, which it described as the first-ever single subscription for unlimited connectivity. It said this is the first and only connectivity provider to combine fast, reliable home internet and wireless together across as many devices as needed, under one subscription and one price.

The company said: “AT&T OneConnect runs on America’s largest wireless and fibre networks and America’s best and fastest home internet. Combined, it delivers the simple experience customers deserve and defines connectivity of the future.”

Related stories

Trump pauses Iran energy strikes, reducing risk to Middle East telecoms infrastructure

Google Cloud’s journey towards the ‘agentic telco’ to reshape network AI

Lumen’s big shakeup: Jim Fowler to replace Dave Ward as CTO

Microsoft is planning to invest US$5.5 billion in Singapore by 2029, as demand for AI skills and computing continues to grow across the region.

First reported by The Wall Street Journal, the investment involves the tech giant providing tools and training for tertiary students, teachers and nonprofits. Company vice chair and president Brad Smith said the investment will also go toward ongoing operations.

“Our ongoing investment in cloud and AI infrastructure reflects Microsoft’s long‑term confidence in Singapore as a global digital leader,” he said. “We’re focused on helping people and organisations use AI by strengthening skills, increasing cybersecurity and resilience and advancing trusted governance.”

Microsoft has committed billions across Asia in recent years, including in Indonesia, Malaysia and India. This particular announcement comes just a day after Microsoft said it plans to invest more than $1 billion in Thailand.

“As AI adoption accelerates, readiness remains uneven,” Microsoft said. “Many institutions and communities lack the skills, guidance, or capacity to adopt AI responsibly and effectively.”

This all forms part of Microsoft’s plans to double its data centre capacity in the next two years and spend more than it previously expected to on AI infrastructure. This includes data centres – with the company opening new data centre regions in countries such as Denmark to advance cloud and AI services.

Denmark East in particular has been designed to enhance digital resilience, innovation and economic growth. Microsoft customers in Denmark will be provided with local and secure cloud infrastructure.

The news also comes as Microsoft reshuffles its AI leadership, appointing Jacob Andreou to run its entire Copilot division.

​In a memo shared with the technology giant’s employees, Satya Nadella, chairman and CEO, said: “Progress at the AI model layer is more critical than ever to our success as a company over the next decade and is foundational to everything we build above it.”

Related stories

Could Microsoft sue OpenAI over $50bn AI deal with Amazon?

Nscale & Microsoft partner with Nvidia & Caterpillar to scale new AI data centre

Microsoft’s $360bn fallout: Is its OpenAI bet back-firing?

Iraq’s communications market stands on the brink of an inflection point set to position it as a key international transit hub in the near future, if remaining regulatory bottlenecks are broken down.

So says Asoz Rashid, CEO of the country’s iQ Group. It comes as his company moves to bolster Iraq’s standing further this year in the next phase of the Silk Route Transit fibre-optic project, run by subsidiary iQ Networks. This involves the construction of a secondary route across the country, providing resilience and redundancy for the 3,500km network previously launched in 2023.

Given the infrastructure both already deployed and planned in the country, Rashid believes the speed and quality of Iraq’s communications can rapidly come into line with those of other digitally advanced countries the moment restrictions lift.

“Once the government sorts out regulations and pricing structures, Iraq has the internet capacities for the market to transform overnight,” he says.

High-quality silk

Rashid points to the quality of infrastructure already available via iQ Networks, which claims the Silk Route provides the shortest terrestrial route connecting the Middle East and Asia to Europe – running northwards from the Gulf and reaching the borders of Kuwait, Saudi Arabia, Iran, Turkey and Syria.

“When more people started testing our cables, they realised how efficient our network is compared with the traditional Red Sea submarine cable route via the Suez Canal,” says Rashid. “The infrastructure is also essentially AI-ready.”

With that type of network, the country’s promise as an alternative terrestrial fibre route is surging, he says. It comes, too, as concerns about overreliance on the Red Sea pathway for traffic between Europe and Asia have risen following subsea cable cuts on the route over the past couple of years.

Remaining barriers

One remaining issue holding up a full-scale explosion of traffic in Iraq is the country’s internal regulatory environment, says Rashid, resulting in a fragmented and uneven market when it comes to fibre quality and pricing. He cites factors such as complexity with government policy and licensing, which is often carried out on a regional basis rather than by just a few carriers nationwide.

Sulaymaniyah in the north, for instance, where iQ Networks is headquartered, has speeds and latencies similar to other sophisticated networks around the world, says Rashid. However, the capital, Baghdad, to the south, is much further behind.

The picture is compounded by the traditional view of Iraq as a postwar country with many security concerns.

Yet things have changed significantly in the past couple of years, with Rashid now far more confident that the market is truly ready to take off. “Regulatory bodies have started becoming more receptive to following an international market structure, at the same time as Iraq has gained more regional and international interest in its fibre transit market,” he says.

Iraq is also pushing to digitise its economy, with the country lying behind many others in the Gulf on AI readiness. Furthermore, it wants to reduce its dependence on cash and diversify outside the oil industry.

On the cusp

Meanwhile, there is significant recent evidence that the market is ripe for a surge. Internet penetration has grown to more than 80%, while there has been a flurry of activity to roll out new subsea cables connecting to Iraq.

In addition, iQ Networks has seen a big increase in demand for the Silk Route, surpassing the milestone of 1Tb of capacity in 2024. With the level of demand now from the likes of OTT players and hyperscalers, Rashid believes this will only continue to rise, and foresees financial companies migrating from Red Sea routes to the Silk Route as they see the latency advantages for connecting to European markets.

As the opportunity grows, iQ Networks hopes to take more of a role in managing data on the route, with a third phase of its Silk Route project later involving building data centres.

If such initiatives are combined with continued improvements in the regulatory environment, Iraq is on course to become a thriving hub for transit, believes Rashid.

“Internationally, demand is huge to enter the Iraqi market and use the transit,” he says. “If you have good and sophisticated digital infrastructure, it will bring big investment.”

Related stories

iQ Networks and GBI sign Iraq’s first dark fibre agreement

For countries aiming to ride the wave of digital transformation, maximising the opportunity means harnessing a number of intertwining trends. In Iraq, some companies are making tracks to help the country catch up with other Gulf nations and drive low-latency pathways between regions by pursuing several strands of the digital journey at once.

One of these is ISP Supercell Network, owned by SCN Holding, a group specialising in different verticals headquartered in Iraq and the UAE, with an additional presence in the US. This January, SCN signed a strategic partnership deal with global investment firm Otaiba Investments to accelerate the deployment of digital infrastructure in Iraq, as well as the UAE and the wider Middle East.

In Iraq, the two companies will focus on expanding fibre networks and data centre capacity to support advanced services, as well as developing new platforms for cloud computing, AI and IoT tailored to the needs of enterprises and government entities. In addition, Supercell hopes the momentum created by the partnership will help to attract international and local investors, supporting Iraq’s vision of becoming a regional digital hub.

“With moves like this, we’re working to optimise the digital transformation and preparing our infrastructure for the era of AI,” says Ehab Hussein, an executive board member at Supercell. “We also want to help coordinate development with local and global partners.”

Main pillars

One pillar of the partnership with Otaiba Investments is the construction of a transit route through Iraq that will connect the Gulf with Europe via Turkey. The aim is to complete the rollout in less than 18 months, serving as an attractive alternative terrestrial pathway to the Red Sea.

“From the Middle East and Far East to Europe, lots of people are looking at the Iraqi route because it can provide the shortest and lowest-latency pathway,” says Hussein. “The wider Gulf region is also booming, with developments in AI, smart cities and data transit. Our plan is to build a mesh network that can deal well with multiple issues like environmental impacts and fibre cuts.”

At the same time, this will complement the fibre-to-the-home network that Supercell has been rolling out in Iraq for residential and business users, today passing more than 2 million properties with high-speed connectivity.

As a further pillar of the deal with Otaiba Investments, Supercell plans to rapidly expand its data centre footprint in Iraq. “The data centre business is huge worldwide, but Iraq is still a very young market in this respect,” says Hussein. “We’re therefore seeking to help the country catch up with the rhythm of development and AI growth in the Gulf region.”

Green and inclusive

He sees the pathway ahead for Supercell as an exciting one, especially as regulations are becoming more flexible in Iraq to cater for digital growth. But Hussein emphasises that with potential rapid expansion of infrastructure and traction among customers ahead, the company has also set its sights firmly on green, sustainable and inclusive transformation.

On the environmental front, for example, Supercell is looking into power options such as solar panels to capture the high levels of sunlight in Iraq and building data centres near hydroelectric power plants. “We’re also working in parallel with hyperscalers to deepen understanding of the environment and carbon emissions,” adds Hussein.

Furthermore, the company has a big focus on inclusion of women in the business and on its role in the community, building education and awareness platforms to promote community responsibility and accessible digital learning. This goes hand in hand with its focus on customer-centricity through means such as having its own customer-experience centre.

Built on these foundations, Hussein expects Supercell to thrive in the next phase of its digital development. “One of the key success pillars for Supercell is our investment in people and expert teams,” he says. “That’s how we’ve built our reputation with vendors and technology partners, and are able to execute both in line with standards and a fast timeframe.”

Taking the world by surprise in 2025, the DeepSeek platform emerged onto the AI scene and caused a stir across global markets. Now, the AI landscape looks vastly different.

What made the Chinese AI app stand out was that its R1 model sought to undercut OpenAI’s ChatGPT – operating powerfully, but at a fraction of the cost.

One year later and DeepSeek services have been heavily adopted across global markets, particularly in Southeast Asia and Africa. Undoubtedly having shaped AI development and interest in 2025, its AI models continue to raise questions about the future of AI development. Companies continue to rethink how much they rely on expensive semiconductors and how high-performance computing (HPC) could evolve moving forward.

How DeepSeek thwarted global technology supremacy

Consequences of the R1 model’s arrival were monumental. Nvidia, whose chips continue to power the most advanced AI systems, had US$600 billion reportedly erased from its value. It was a blow to the US at the time, which had been seeking to slow China’s global AI development with chip export restrictions.

DeepSeek’s progress caught the US technology sector off-guard because it appeared that China could be just as powerful on its own, raising questions over the true impact of technology restrictions. Even US President Donald Trump remarked on the DeepSeek development as a “wake-up call” for US companies.

It also led to the race for AI dominance between China and the US intensifying in 2025, as AI leadership in the West felt threatened.

DeepSeek’s R1 model very quickly became a leading AI model, sparking a change that the most popular models were no longer majority developed in the US. According to Hugging Face’s blog One Year Since the “DeepSeek Moment,” much of 2025’s progress in the open-source community can be traced back to the release of R1.

“DeepSeek’s R1 model lowered the barrier to advanced AI capabilities and offered a clear pattern to follow, unlocking a second layer. Moreover, the release gave Chinese AI development something extremely valuable: time. It showed that even with limited resources, rapid progress was still possible through open source and fast iteration,” Hugging Face’s Adina Yakefu and Irene Solaiman wrote.

“This approach aligned naturally with the goals set out in China’s 2017 “AI+” strategy: combining AI with industry as early as possible, while continuing to build up compute capacity over the long term.”

Can the US stay ahead? Charting global interest in DeepSeek models

US technology companies will not want to be surprised by another ‘DeepSeek moment’. As AI interest and investment continue to surge, 2025 saw significant spending on infrastructure and large private funding for startup companies. In fact, Goldman Sachs predicts that AI companies may invest more than $500 billion in 2026, with major leaders like Microsoft and Nvidia leading capital expenditure (capex).

Amongst this backdrop, the DeepSeek platform keeps growing, with its surge becoming especially attractive in developing regions. This includes a two-to-four-times higher interest in Africa than in other parts of the world, as analysed by Microsoft in its Global AI Adoption in 2025—A Widening Digital Divide report.

In China, the platform now reportedly boasts a market share of nearly 89% among AI users.

2026: AI remains in the spotlight

DeepSeek models continue to disrupt the AI landscape, with The Legal Wire recently reporting that the Chinese government has granted conditional approval for DeepSeek to purchase Nvidia H200 AI chips.

Nvidia CEO Jensen Huang said the company had not yet received official confirmation of these orders, but such a move would follow similar authorisations by ByteDance, Alibaba and Tencent.

The US cleared H200 exports to China in January, but this DeepSeek transaction could spark further geopolitical tensions between the two nations, particularly after Nvidia came under political pressure in Washington after a senior lawmaker said the firm helped DeepSeek optimise its AI systems.

According to a letter, the work focused on squeezing efficiency out of DeepSeek’s training processes, using fewer GPU hours, or computing cycles on specialised chips, to achieve performance rivalrous with leading US models from firms such as OpenAI or Google.

US officials have also previously said DeepSeek’s technology was aiding China’s military, though the company has not publicly confirmed this.

This keeps DeepSeek firmly in the spotlight as the platform prepares to launch its next-generation V4 model this month (February 2026). Although information about V4 is limited, the model is expected to be a coding-optimised model that could push the boundaries of what AI can do for software development.

Reported model testing shows V4 could outperform Anthropic’s Claude 3.5 Sonnet and OpenAI’s GPT-4o on coding benchmarks. However, at the time of writing, this remains unverified.

As the AI race keeps hotting up, what’s clear is that China’s input shouldn’t be underestimated. With infrastructure spending and data centres booming worldwide, China remains active in AI and still operates at the forefront of its development.

Related stories

Cisco CEO: AI is ‘bigger than the internet’, but warns of ‘carnage’ ahead

Marc Ganzi: Why Metro Connect USA matters more than ever in the AI era

ITW Asia: Subsea supply crunch, technology shifts, and the ‘AI question’

The strategic acquisition is designed to strengthen nLighten’s presence in a critical European data centre market.

Located in Émerainville, Paris, the data centre becomes nLighten’s eighth data centre site in France and adds to the company’s portfolio of more than 30 data centres across seven markets. It has been acquired from oXya, a leader in managed services and cloud solutions for business-critical applications.

“The acquisition of this Paris data centre represents a significant expansion of our French footprint and strengthens our position in one of Europe’s most dynamic digital infrastructure markets,” said Harro Beusker, CEO and co-Founder of nLighten. “Paris is a critical hub for regional connectivity, and this facility enables us to deliver enhanced capacity and resilience to our enterprise customers.”

The site is located in the eastern data centre cluster of Paris and is roughly one kilometre from nLighten’s existing PAR1 facility. It will continue serving anchor customer oXya under a long-term master services agreement, with additional capacity being made available to enterprise customers via channel partners.

The transition also enables oXya to focus on its core business of delivering managed cloud services and ensuring continuity for its infrastructure needs.

Beusker added: “The proximity to our existing Paris sites creates operational synergies while also allowing us to support dual-site deployments. This acquisition exemplifies our strategy of building smart, sustainable infrastructure that scales with customer needs and contributes to the digital transformation of European businesses.”

The facility is designed to support high-density and AI-ready configurations, in addition to providing scalable infrastructure that can evolve with customer requirements. nLighten is eager to focus on the delivery of sustainable, interconnected infrastructure tailored to enterprise needs, with seamless connectivity between its sites.

The news comes shortly after nLigthen signed an energy deal with Uniper to supply round-the-clock carbon-free power across all its data centre sites in Germany. Harnessing hydropower, the agreement supports nLighten’s renewable strategy of procuring green electricity and supporting its clients who require energy transparency.

For this site in Paris, nLighten is hoping to reinforce its position in one of the most important digital hubs in Europe and expand its portfolio of sustainable regional infrastructure solutions.

“This acquisition fully aligns with our ambition to build a distributed, locally operated digital infrastructure across France. By adding capacity in the Paris region through three interconnected sites, we provide our customers with the conditions needed to deploy more resilient architectures, better secure their data and meet growing requirements in terms of performance, service continuity, and digital sovereignty,” explained Anwar Saliba, managing director of nLighten France.

“Proximity is therefore becoming a concrete driver of sustainable digital transformation in France.”

Related stories

The GCC’s data centre revolution: Where sovereignty, sustainability & smart capital converge

nLighten adds senior executives as European data centre expansion continues

New atNorth metro data centre to boost Swedish digital transformation

While headlines often dazzle with tales of sleek Starlink satellites, the real backbone of our interconnected digital world is the vast network of subsea cables quietly stretching beneath the oceans. Unbeknownst to most, an intricate network of subsea cables carries 99% of the world’s international data traffic, silently enabling everything from streaming video calls to instant financial transactions.

“Not a lot of people understand subsea,” admits Carl Grivner, CEO of Flag. “They don’t realise that 99% of the traffic carried between countries travels via subsea.”

Grivner is undoubtedly a seasoned figure in the industry, clocking up over two decades of experience in digital infrastructure. Formerly CEO of Colt, he assumed leadership of FLAG nearly six years ago. During his tenure, the company transitioned from its previous identity as Global Cloud Xchange to the FLAG brand in the past year.

Grivner has overseen the acquisition of additional capacity on key subsea cable systems, including securing a fibre pair on the ECHO trans-Pacific subsea cable in partnership with Google. The company has also seen substantial investment growth across India and Asia. Under Grivner’s stewardship, FLAG secured a US$340 million refinancing package in 2025, enabling significant upgrades to its network, the development of future subsea projects, expansion into edge data centres, and the establishment of new cable landing stations.

Fast forward to 2026 and Grivner is looking to the future. Flag is powering ahead with a long-term strategy that puts data centre connectivity at its core, mapping network expansion to both present and future data centre locations.

“We’ve developed what we call Vision 2030, which is really about a route-co strategy. Think of us as an airline, flying as many routes as possible,” Grivner explained.

Hybrid connectivity is the future

New additions to Flag’s network stretch from Mumbai to Singapore, Mumbai to France, and Singapore to Guam, with a terrestrial extension into the Middle East on the horizon. And it seems the days of terrestrial networks playing second fiddle are over.

“Right now, we’re probably 90% subsea, 10% terrestrial. Over the next four years, our 2030 vision will shift that to about 60% subsea and 40% terrestrial. That balance is essential to reach data centres, ensure reliability, and provide redundancy across the network.”

Grivner is adamant that this hybrid model is the future: “The combination of both makes for a powerful story for customers… the terrestrial route may be faster, more reliable, or offer better performance metrics.”

He continues, “Going from London to Paris is one thing but combining terrestrial and subsea to take a customer from London to Tokyo – that’s where the magic happens.”

Following the trail

When it comes to growth Flag’s network expansion mirrors classic marketing tactics, Grivner draws a parallel with Burger King’s 1970s strategy: “It’s very straightforward for us – where data centres are built, connectivity will be required.”

He predicts a tenfold increase in capacity needs over the next four years. “Without connectivity, a data centre is just a big, dumb warehouse. The amount of data flowing through these centres is going to increase tenfold, it’s massive.”

Emerging global hubs

The global map of subsea connectivity is evolving fast. Grivner identifies intra-Asia routes, the Middle East, and Australia as new critical nodes. India stands out: “India is growing massively in terms of data centres right now… it’s underserved in capacity and infrastructure. It’s going to be a massive growth market.”

With hyperscalers leading record build-outs in 2025 and new Tier-II cities emerging as hotspots, demand is surging on both sides of the Atlantic. “At one time, we thought the Atlantic had enough capacity. Now, it’s at a point where more is needed – it’s being revitalised in terms of demand.”

The role of hyperscalers has shifted dramatically in recent years. No longer merely customers of wholesale capacity, these firms have become major investors and architects of new subsea routes, designing systems to meet the specific needs of cloud and AI workloads.

Flag positions itself as a neutral operator, “We’re like Switzerland. We don’t have a bias one way or the other. We aim to serve all of our customers globally,” Grivner said. Private cable builds are redrawing competition but also creating fresh opportunities. “We see private builds… they’ll lay five repairs but reserve some capacity for others. That ‘other’ could be us.”

The power conundrum

Grivner cites power or (lack of) as being one of the biggest challenges facing the digital infrastructure industry. The International Energy Agency cites that electricity consumption from data centre amounted to 1.5% of global electricity consumption in 2024 and has increased by 12% year on year over the last five.

Data centre energy consumption is projected to more than double by 2030, reaching around 945-980 Terawatt-hours (TWh) globally, driven primarily by the massive power needs of AI.

“One new data centre I saw had power needs equivalent to Atlanta, Georgia. I don’t believe the necessary systems exist yet (to support the predicted level of capacity required) and that’s one of the unknowns moving forward,” Grivner warns. “Power is going to be one of the wild cards in all this.”

Geopolitics and resilience

Working in the Middle East and Asia means balancing diverse interests. “Different conflicts around the world affect what we do and that hasn’t changed, nor is it likely to.” Maintaining resilience is critical. Cable cuts, anchor drags, and supply bottlenecks expose the need for more fleet capacity, as Grivner points out, “It takes about five years to build a ship for this industry. There aren’t enough ships or service providers.”

Looking ahead, Grivner sees political unrest and infrastructure alignment shaping the next five years. “The challenge in the subsea space will be  navigating the geopolitical landscape and balancing assets with partners or competitors, especially on the terrestrial side.”

RELATED STORIES

Red sea risks highlight urgent need for subsea route diversification in the Middle East

Capacity Power 50: Middle East 2026 revealed!

Why the Middle East is attracting major data centre investment

Humans can watch what happens on the platform, but they cannot post or interact. Instead, autonomous AI agents create threads, reply to one another and upvote content, effectively socialising without human involvement.

It is easy to dismiss this as a novelty. In many ways, Moltbook probably is one. But from a connectivity perspective, it is still worth paying attention to, not because of what it is today, but because of what it quietly reveals about where network demand may be heading.

Strip away the novelty, and Moltbook becomes a simple demonstration of AI systems talking directly to other AI systems, continuously and at scale. There is no human pacing, no daily rhythm, no natural downtime, and that is where the implications for networks begin.

Global connectivity has always been shaped by human behaviour. Even as traffic volumes have exploded, from video streaming to cloud services, usage has remained broadly predictable. There are peaks and troughs, working hours and quiet periods, seasonal effects and geographic patterns. Networks, pricing models and capacity planning have evolved around those assumptions.

AI agents don’t fit into that model. They don’t sleep, nor do they follow time zones. And they definitely do not behave like regular users.

If platforms like Moltbook are an early glimpse of a broader shift toward autonomous systems coordinating with one another, then AI-to-AI traffic introduces a new type of demand: persistent, machine-generated and difficult to forecast. Activity may rise and fall based on optimisation cycles, training runs, or internal coordination, signals that are largely invisible to network operators.

Just as important is where this traffic sits. AI agents are not edge devices; they run almost entirely inside hyperscale cloud environments and data centre clusters. Their conversations generate east-west traffic, between data centres, across private interconnects and through internet exchanges, rather than traditional north-south flows between users and applications.

That has consequences for where capacity pressure builds. The growth story shifts further away from the last mile and deeper into metro fibre, inter-DC connectivity, IX capacity and private networking. For many operators, this is already familiar territory, but AI agents could accelerate the trend.

There are commercial implications, too. Machine-driven traffic is inherently harder to model than consumer usage. Long-term capacity commitments become more challenging when demand is less predictable, increasing the value of flexible, on-demand provisioning and shorter planning cycles.

Security and governance also deserve attention, given that AI-only platforms generate large volumes of automated traffic with limited human oversight. Feedback loops, runaway signalling or abuse could stress networks in ways existing monitoring tools were not designed to catch quickly. This is less about speculative AI risk and more about day-to-day operational resilience.

Moltbook itself may never matter at scale. Its real value is as a visible signal of a broader shift already underway: a growing share of network traffic generated not by people, but by machines talking to machines.

As AI agents become embedded in enterprise software, financial systems and automated operations, AI-to-AI traffic will move from edge case to material consideration. The question for the connectivity industry is whether networks are being prepared for that future now or whether the adjustment will only come once the traffic graphs start to change.

RELATED STORIES

AI now lies, denies, and plots: OpenAI’s o1 model caught attempting self-replication

AI fatigue in telecoms: When the hype outruns the hardware

For Metro Connect 2026 keynote speaker Marc Ganzi, the future of digital infrastructure is about more than building bigger data centres. It’s about anticipating demand, coordinating networks and creating systems that can scale efficiently and sustainably.

As CEO of DigitalBridge, Ganzi is focused on three priorities for 2026: power, partnership and pipeline. 

Reliable, low-carbon power has become a critical factor in deciding where and how to build infrastructure.  

Without it, even well-funded projects can face delays or limitations. Partnerships with hyperscalers, AI companies, carriers and utilities ensure infrastructure works as an integrated platform rather than disconnected assets and a robust pipeline of projects keeps the company ready to meet growing AI-driven demand. 

However, for Ganzi, success comes down to planning, collaboration and foresight, creating infrastructure that is ready before demand arrives. 

On SoftBank’s recent acquisition 

Recently, it was announced SoftBank signed a definitive agreement to acquire DigitalBridge, in a deal worth $4 billion. 

As a result, the acquisition will help SoftBank expand and finance the critical infrastructure needed for next-generation AI technologies. 

“As AI transforms industries worldwide, we need more compute, connectivity, power, and scalable infrastructure,” SoftBank chairman and CEO, Masayoshi Son, said.

“DigitalBridge is a leader in digital infrastructure, and this acquisition will strengthen the foundation for next-generation AI data centres, advance our vision to become a leading ASI platform provider, and help unlock breakthroughs that move humanity forward.” 

Ganzi added: “The buildout of AI infrastructure represents one of the most significant investment opportunities of our generation. “SoftBank shares our DNA as builders and long-term investors committed to scaling transformational digital infrastructure.  

“Their vision, capital strength, and global network will allow us to accelerate our mission with greater flexibility, invest with a longer-term horizon on behalf of our investors, and better serve the world’s leading technology companies as they scale their AI ambitions.” 

Misunderstood trends in AI infrastructure  

While AI continues to dominate industry discussions, Ganzi warns that certain narratives are oversimplified. 

“One misunderstood idea is that AI is only a data centre story. AI is a network story that stresses every layer of infrastructure, from subsea cables to street-level small cells. Underinvesting in any layer creates bottlenecks that limit overall value,” he notes. 

However, equally important is the misconception that AI infrastructure can scale without addressing power and sustainability. 

“Another misconception is that AI infrastructure can scale without rethinking power and sustainability. The reality requires disciplined, long-term coordination with utilities, regulators, and communities. That work is harder than the hype suggests, but it is where durable returns are created,” Ganzi explains. 

For operators and investors, this means that long-term planning and integrated platform thinking are no longer optional, they are essential. 

Priorities for 2026 

As a result, Ganzi reveals, lessons from 2025 have sharpened his priorities for this year, which includes power, partnership and pipeline. 

“Power means securing and optimizing it across platforms. Partnership means deeper collaboration with hyperscalers, AI companies, carriers, and utilities,” he says. 

“Pipeline means continuing to raise and deploy capital into the most compelling global opportunities. 

“Ultimately, our role is to ensure infrastructure is ready before demand arrives. 2025 reminded us that the window between concept and scale is shrinking, and 2026 will be about staying ahead of that curve,” Ganzi explains. 

Execution risk remains central. Even with abundant capital, permitting delays, grid constraints, and workforce shortages can stall projects, as a result, operators and investors who account for these constraints from the outset will have a strategic advantage. 

Lessons learnt from 2025 

Looking back, Ganzi identifies several lessons from 2025 that will define the next phase of infrastructure growth. 

“Execution risk, not lack of capital, is the primary limiting factor in this cycle,” he says. 

“It also reinforced the importance of diversification across data centres, towers, fibre, small cells, and edge assets, because AI demand does not concentrate in a single asset class,” Ganzi notes. 

However, according to the DigitalBridge CEO, innovation in 2025 was no longer linear, but compounding. 

“The speed of change in AI workloads, semiconductor architectures, and software forced a shift toward platform thinking rather than individual assets,” Ganzi states. 

“We also saw innovation in financing, including larger sector-specific funds and creative credit and securitisation structures, that enabled continued build-out despite higher interest rates.” 

“Ultimately, our role is to ensure infrastructure is ready before demand arrives. 2025 reminded us that the window between concept and scale is shrinking, and 2026 will be about staying ahead of that curve.” 

Metro Connect USA through the years 

Commenting on the Metro Connect USA event, which Ganzi is keynoting at, he says: “Metro Connect started as a niche gathering for carriers and fibre operators.  

“It has evolved into the premier strategy event for the entire digital infrastructure value chain- and has done so against a backdrop in which digital infrastructure increasingly underpins economic competitiveness and technological innovation on a global scale. 

“Today, hyperscalers, private equity, infrastructure funds, credit investors, tower and data centre operators and policy leaders are all in the same room,” he says. 

“The content has evolved as well, moving from bilateral deals and route maps to discussions around AI workloads, power markets, securitisation, ESG, and now the geopolitics of connectivity. It has become the place where the industry takes stock of the year and sets the agenda for what comes next,” Ganzi adds. 

Milestones in Metro Connect 

For Ganzi, three milestones stand out in the event’s growth, which includes the scale, the shift from telecom to infrastructure and the prominence of AI. 

“The event has grown from a few hundred attendees to well over 2,000, mirroring the growth of digital infrastructure as an institutional asset class.  

“Metro Connect now reflects a converged ecosystem where towers, fibre and data centres are financed and operated as one interconnected platform.  

“It’s where investors, operators, and customers compare notes on what actually happened in the prior year, pressure-test assumptions, and recalibrate capital allocation,” Ganzi explains. 

As the conference occurs early in the year, its discussions have an outsized impact. 

“The themes that emerge here, such as AI infrastructure, open-access models, wholesale fiber pricing, and cost of capital, tend to shape boardroom and investment committee discussions for the rest of the year,” he says. 

“A few years ago, it was something people kept on their radars. Today, it is central to nearly every session and boardroom discussion, from power planning to network architecture to M&A,” Ganzi concludes. 

Related stories

Marc Ganzi: Why power & AI will redefine the data centre industry 

DigitalBridge CEO Marc Ganzi to deliver keynote at 25th Metro Connect

Marc Ganzi: ‘Power, not GPUs, is our biggest challenge’

On January 8, 2026, Iranian authorities imposed a near-total internet shutdown in response to widespread anti-government protests, cutting international access to almost zero and severely limiting domestic communications.

Monitoring groups reported that internet traffic plunged to minimal levels, with international links effectively severed for much of the population.

By late January, evidence of traffic recovery emerged. Cloudflare Radar data showed a notable increase in HTTP requests, with overall connectivity rising compared with the shutdown period, particularly via mobile networks operated by major carriers such as MCI and Irancell. However, full restoration remains far from complete.

Despite these gains, access is highly restricted and inconsistent. Many global platforms, including popular search engines and social media, remain blocked or only sporadically available in different regions.

Officials have issued conflicting statements on when normal internet access might fully return, with some suggesting services could normalize by the end of the week, comments that have yet to translate into universal connectivity.

The blackout and its aftermath have exposed structural shifts in Iran’s digital landscape. Analysts now see evidence of “whitelisting”, a tiered access system that allows limited domestic services while blocking most global content. This approach indicates a broader strategy toward digital isolation, embedding dependency on a state-controlled National Information Network.

Economically, the shutdown has inflicted significant damage. Reports estimate daily digital economy losses in the tens of millions of dollars, with wider impacts on business operations, e-commerce, and financial services.

RELATED STORIES

Iran’s internet blackout: Nation remains offline as protests and economic strain deepen

Trump says he will talk to Musk about restoring internet in Iran amid nationwide blackout

TD Cowen estimated that Oracle’s OpenAI deal could require $156 billion in capital spending, with layoffs potentially freeing up $8-10 billion in free cash flow.

The bank’s research note suggested that, in addition to potential job cuts, Oracle could sell its health unit Cerner to support the financing of its AI data centre endeavours. It said finding equity and debt investors are questioning how Oracle plans to finance its data centre building programme to support its US$300 billion five-year contract with OpenAI.

It said: “both equity and debt investors have raised questions about Oracle’s ability to finance this build-out as demonstrated by widening of Oracle credit default swap (CDS) spreads and pressure on Oracle stock/bonds.”

TD Cowen also said US banks had pulled back from Oracle-linked data centre project lending, while private operators leasing to Oracle were struggling to secure financing

Oracle has been building large AI data centres rapidly, including its 4.5-gigawatt (GW) initiative with OpenAI. Taking place in Abilene, Texas, the upcoming AI data centre will use Nvidia GB200 GPUs for advanced AI training, in addition to featuring liquid cooling technologies.

TD Cowen has estimated the OpenAI build-out alone would require roughly three million GPUs and other IT equipment. Oracle is also building for Meta and Nvidia in a total commitment worth $523 billion.

Oracle sent shockwaves through the global tech markets at the end of last year, as its shares tumbled over 10% after its results missed revenue expectations. The company has accrued significant debt over its AI build outs, raising questions over if heavy investments will gain good enough returns to justify financial risk.

The company said 2026 would be a pivotal year for advancing AI in the US and Oracle would be playing a key role.

“We’re building the AI infrastructure needed to unlock new cures for diseases, enable scientific breakthroughs and drive significant economic growth,” the company said via its online blog in January. “Already, we have AI infrastructure projects underway … they are playing a critical role in our country’s future, fortifying American leadership in AI for generations to come.”

The news also comes amid job cuts peaking again in the technology sector, with Amazon confirming a new wave of global job cuts, in addition to Ericsson, Meta, TCS as each company looks to streamline processes and prioritise AI.

Related stories

Cisco CEO: AI is ‘bigger than the internet’, but warns of ‘carnage’ ahead

iXAfrica to host Oracle’s first public cloud region in Kenya

Intel CEO ‘disappointed’ as 13% share drop indicates AI data centre struggle

As businesses look to scale AI adoption, Tata Communications said its new platforms can help bring together network, cloud, cybersecurity and other offerings. Together, the solutions aim to remove silos, simplify operations and embed a greater level of trust in the digital space.

The suite comprises three independent, complementary offerings – Tata Communications IZO+ Multi Cloud Network, Tata Communications Edge Distribution Platform and ThreadSpan.

Tata Communications IZO+ Multi Cloud Network posits AI-enabled connectivity as a strategic advantage. The platform removes cost-related frictions and gives organisations greater control over how data moves and costs – prioritising intelligent policy, built-in optimisation and a single view of performance so that AI can be scaled faster.

Tata Communications Edge Distribution Platform enables faster responses and real-time digital experiences by bringing intelligence closer to where data is created. The company said it combines content delivery, security and compute at the edge to reduce latency and boost resiliency, therefore creating the foundation needed to run real-time AI applications globally.

ThreadSpan offers a unified, single-pane view across hybrid and multi-vendor networks, combining visibility, manageability, security and automation across network, cloud and security domains. Its use of AI enables businesses to enforce policies and identify potential issues much sooner to improve security readiness.

“Digital infrastructure is becoming increasingly complex and AI is amplifying that challenge. With our new suite of AI-ready offerings and our Digital Fabric, we are bringing together a secure, unified and intelligent foundation that simplifies how enterprises design and run their digital environments,” said A.S. Lakshminarayanan, MD & CEO at Tata Communications.

“This enables our customers to reduce complexity, operate with confidence and focus their energy on innovation and scale AI securely.”

The new solutions arrive as AI is becoming more embedded in business decision-making, particularly as businesses recognise their infrastructure is not AI-ready. As AI interest continues to soar, organisations are faced with the reality of rising costs and security gaps.

According to Tata Communications, enterprises need infrastructure that is designed end-to-end for AI to best optimise the technology. The company said its new solutions address this and form a cohesive platform that helps businesses with these complex environments and enables them to better deploy AI at scale.

“By bringing connectivity, edge intelligence and predictive observability into a unified AI-ready ecosystem, Tata Communications is helping enterprises unlock real business value from AI, with infrastructure that is simpler to manage, faster to deploy and built for the demands of tomorrow,” the company said in its press release.

Related stories

US $3.2m subsea grant to link India, Singapore & SE Asia

AMD and TCS partner to drive enterprise AI adoption at scale

Tata Communications integrates new TGN-IA2 subsea cable into global network