Earlier this month, something unthinkable happened. Iranian drones struck three Amazon Web Services data centres in the Gulf, two in the United Arab Emirates and one in Bahrain. The strikes knocked two of the ME-CENTRAL-1 region’s three availability zones offline and disrupted more than 109 services in a matter of hours.
Emirates NBD and First Abu Dhabi Bank reported intermittent banking app outages, whilst taxi platform Careem, payments companies Alaan and Hubpay, and enterprise software provider Snowflake, were all hit with service disruptions.
AWS told customers, “We continue to strongly recommend that customers with workloads running in the Middle East take action now to migrate those workloads to alternate AWS regions. Customers should enact their disaster recovery plans, recover from remote backups stored in other Regions, and update their applications to direct traffic away from the affected Regions”.
The company that had sold the Gulf the idea that the cloud was indestructible, told its own customers to migrate their workloads out of the Middle East immediately.
The strikes were not merely a dramatic news story. They were a forcing function; the moment the data centre industry’s quiet anxieties about physical security in the Gulf crystallised into something far more concrete. Within days, the question every hyperscale operator, cloud architect, and infrastructure CTO was asking was the same: what happens when a war zone swallows a cloud region?
Gambling on the Gulf
Between 2021 and 2024, the Gulf states made a spectacular bet on becoming the world’s data infrastructure crossroads. The logic was sound on paper. The UAE and Saudi Arabia sat at the geographic midpoint between European latency requirements and Asian demand, enjoyed sovereign capital in quantities that commercial markets could not match, and had constructed business environments designed to attract the world’s largest hyperscalers with a combination of land concessions, energy subsidies, and tax policy that bordered on the generous.
According to Research and Markets, the Middle Eastern data centre market was valued at USD 5.57 billion in 2023 and projected to reach USD 9.61 billion by 2029. Saudi Arabia’s HUMAIN programme and the UAE’s AI corridor with France were pulling rack densities above 100 kW and driving the adoption of immersion and two-phase liquid cooling at a scale most Western markets had not yet achieved.
In October 2025, Khazna Data Centers announced a new 20-hall AI-optimised facility in Ajman and new data centre services regulations in Saudi Arabia were framed around Vision 2030 – an ambitious blueprint which aims to transform the Kingdom into a global digital hub.
The physical risk to this infrastructure, however, was almost absent from site selection models. As analysts at Omdia have long observed, geopolitical and country risk should be the first, not the last, item on a site selection checklist.
The Russia-Ukraine conflict provided a warning that the industry absorbed theoretically, but possibly not operationally. The Gulf’s sovereign financing and its premium power concessions made it easy to discount the region’s position in a geopolitically volatile neighbourhood. The attacks of 1 March ended that reasoning.
For the first time, the prospect that hyperscale data centres could be drawn directly into geopolitical conflict is no longer just theoretical.
The Iranian Revolutionary Guard Corps claimed the Bahrain facility was struck because of its alleged role in supporting US military and intelligence operations, a claim AWS has not confirmed. However, when a data centre is in the physical proximity to military infrastructure (Bahrain hosts the US Navy’s Fifth Fleet), it inherits a portion of that facility’s risk profile, whether its operators intend it to or not.
The industry has, of course, survived outages before. AWS’s redundancy architecture is specifically designed to tolerate the failure of a single availability zone. What it was not designed for was a coordinated physical attack on multiple zones within the same region simultaneously.
The Gulf’s investment pipeline now faces a far more serious deterrent: the undeniable fact that physical infrastructure can be taken out during active conflict.
Could the cloud drift to India?
India’s emergence as the region’s data centre alternative is not a post-conflict solution. Instead, India is a quietly growing story that has been building for several years, and the geopolitical catalyst could hit the accelerator.
The numbers are impressive. India’s IT load capacity stood at 1.4 GW as of the second quarter of 2025, and that figure is expected to double within two years. The city’s fibre infrastructure, proximity to international submarine cable landing stations, and established enterprise ecosystems means it functions as the natural anchor for any operator seeking low-latency coverage of South and Southeast Asia.
The investment commitments also make for interesting reading. Since 2020, more than $14.63 billion has been committed to Indian data centre development, with a further $20 to 25 billion expected by 2030. Global cloud service providers (AWS, Microsoft Azure, Google Cloud, and Oracle) all have growing commitments to Indian regions. OpenAI announced plans to build a 1 GW capacity data centre in India, and Tata Consultancy Services (TCS) is reportedly close to additional deals with several technology giants.
Construction costs are competitive, and policy is moving decisively in the right direction. The draft National Data Centre Policy proposes up to 20 years of conditional tax exemptions, 100 per cent electricity duty exemption and the designation of Data Centre Economic Zones with pre-allocated land near IT corridors.
India’s position within the Asia-Pacific time zone is also relevant here. For Gulf enterprises and governments migrating workloads away from the damaged AWS regions, India offers the closest viable alternative with comparable latency to the Middle East, which is certainly more appealing than routing traffic through European regions or the US.
Meanwhile, the submarine cable infrastructure connecting the Gulf to Mumbai’s landing stations means that the network path already exists. The migration logic practically writes itself.
Carl Grivner, CEO of FLAG commented, “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.”
The growing significance of India’s data centre industry was highlighted at the recent India AI Impact Summit 2026, which drew international luminaries such as former Meta AI scientist, Yann LeCun. Although the summit saw last-minute withdrawals from Bill Gates and NVIDIA’s Jensen Huang due to unforeseen circumstances, the event highlighted India’s rising stature and the world’s keen interest in its digital future.
The white elephant in the (server) room
Enthusiasm for India’s potential cannot ignore the fact that power reliability remains one of the most significant challenges. India’s grid is improving, but it remains unreliable in ways that matter at hyperscale. Diesel backup at the scale required is expensive, polluting, and increasingly incompatible with the ESG commitments providers have made to their shareholders (labelled as India’s ‘dirty secret’ by KPMG analysis). No amount of tax exemption fixes a 200 MW facility running on generators during a grid event.
Water stress is also serious and under-reported. Mumbai, Chennai, and Bengaluru, the three anchor markets, all face acute water pressure. As rack densities climb above 100 kW for AI workloads, the cooling water consumption becomes a genuine constraint, not just an operational nuisance.
Regulatory unpredictability is a real risk. The draft National Data Centre Policy is exactly that, a draft. India has a history of ambitious policy frameworks that stall in implementation, get revised mid-cycle, or create compliance complexity that wasn’t in the original document. Operators who have built in markets like Singapore or Ireland know what stable regulatory environments look like; India is not there yet.
Geopolitical risk doesn’t disappear completely either; it just shape shifts. India-Pakistan tension is a permanent feature in the region. The India-China border dispute flared seriously in 2020, and tensions remain. Admittedly, these are not equivalent to the Gulf’s current exposure, but they are not zero either.
One sceptic notes that construction timelines veer on the side of optimistic. The 4-5 GW by 2030 projection assumes planning, permitting, grid connection, and supply chain execution all proceed roughly on schedule. India’s track record on large infrastructure projects is mixed.
Connectivity infrastructure also matters. India’s submarine cable links to Europe, the Middle East, and Southeast Asia are well-established, but the internal terrestrial network requires continued investment. Latency for certain low-tolerance workloads routed from Mumbai to end-users in more distant Indian markets can be a consideration for operators designing multi-tier architectures.
What are the alternatives?
Singapore remains the most operationally mature market in Asia, with excellent connectivity and political stability. The US Trade and Development Agency (USTDA) recently signed an agreement with SubConnex Malaysia to attract investment for a system expected to serve nearly 1.85 billion people and expand capacity for AI and cloud services.
USTDA is supporting the use of US technology in the proposed SCNX3 submarine cable system, an infrastructure project linking India with Singapore and key data centre hubs across Southeast Asia. The proposed cable will connect Chennai, India, with Singapore, with additional potential landings in India, Malaysia, Thailand and Indonesia.
The problem is that the Singaporean government imposed a moratorium on new data centre construction from 2019 to 2022, specifically because of land and power constraints, and those constraints haven’t gone away. It is a hub, not a growth market.
Malaysia, meanwhile, is mopping up enormous hyperscale investment because it offers Singapore-adjacent connectivity at significantly lower land and power costs. Microsoft, Google, and ByteDance have all committed substantial capacity there. Digital Realty has also unveiled plans to enter the Malaysian market. It doesn’t have India’s domestic demand story, but for workload hosting, it is a credible alternative.
Poland and Central Europe are quietly becoming the most interesting story for operators thinking about the Eastern Mediterranean and Gulf corridor. Warsaw in particular has seen significant hyperscale investment, offers EU legal certainty, and sits within an acceptable latency range of both European and parts of Middle Eastern enterprise customers. For Gulf enterprises with European operations, this has practical appeal.
Saudi Arabia itself is the counter-intuitive option. The argument here is that the HUMAIN and Vision 2030 capital is so substantial, and the sovereign commitment to becoming a data infrastructure powerhouse so deep, the Kingdom will invest in hardening and redundancy in ways that a commercial operator would not. Sovereign risk looks different when the sovereign is also the largest customer and the largest investor.
South Africa and Kenya are longer-term plays for operators thinking about the African connectivity corridor, with Nairobi increasingly cited as ‘one to watch’ as emerging hub. However, neither is ready to absorb Gulf-scale workload displacement in the near term.
The death knell for commercial neutrality
One analyst observed that physical attacks on digital infrastructure are “only going to become more common as AI becomes more and more significant.” That framing, AI infrastructure as a strategic target, demands a radical rethink from the data centre industry.
The industry spent the better part of the past decade constructing a global map built on a single, quietly held assumption: that commercial neutrality was a form of protection. A hyperscale facility was not a military target – it was a utility. The events of 1 March have retired that assumption.
The industry must price that risk, plan for it, and in many cases, move away from it. Not because the Gulf is finished as a digital infrastructure market (sovereign capital and sovereign ambition will ensure it is not), but because the calculus of where to place workloads has changed.
Cloud infrastructure that hosts government data, military contractor workloads, or critical national services is now at the mercy of armed conflict – regardless of whether its operators intended it to be.
Related stories
Iran-US war puts subsea cable network on a knife-edge
Red sea risks highlight urgent need for subsea route diversification in the Middle East
US $3.2m subsea grant to link India, Singapore & SE Asia
Data centres in the line of fire? What the US–Iran war means for the Cloud
Datacloud Global Congress
Industry heavyweights from the data centre sector will convene at this year’s Datacloud Global Congress, to explore strategies for managing power and energy, ensuring operational resilience, and planning for geopolitical disruptions. Attendees can expect in-depth discussions on how recent initiatives are reshaping the economics, operations, and global strategy of hyperscale and edge data centres.
To register for the event, click here.
