AWS said the problem started after temperatures rose inside a single data centre, blaming a cooling system failure while explaining that engineers brought extra cooling capacity online as recovery work continued.

The outage affected services in one Availability Zone, which is a group of connected data centres designed to operate separately within an AWS region. AWS said it redirected traffic away from the affected zone for most services to reduce disruption.

Later in the day, recovery was taking longer than expected because more cooling capacity was still needed before remaining systems could safely return online. AWS further added that it did not yet have a timeline for full recovery.

Coinbase confirmed its trading platform issues were directly linked to the AWS outage. Users reported problems accessing services and delays during trading, although the company later said all markets had been restored and trading resumed normally.

Meanwhile, CME Group reported login and latency problems on its CME Direct trading platform. In a notice to users, the exchange said it had completed “essential maintenance work” and confirmed customers could log back in. The company did not explain the cause of the disruption.

Neither AWS nor CME immediately responded to requests for additional comment outside normal business hours.

The incident again exposed how heavily large financial platforms depend on cloud providers such as AWS. Even a problem in a single data centre zone can spread quickly across trading services, apps and online platforms used worldwide.

AWS has faced similar problems before. In October last year, a major outage disrupted thousands of websites and apps, including Snapchat and Reddit. That disruption became one of the largest internet outages in recent years.

A month later, CME Group experienced another major interruption after cooling systems failed at a data centre operated by CyrusOne in the Chicago area. Trading across stocks, bonds, commodities and currencies stopped for several hours.

The latest outage also points to stress on data centres as companies expand artificial intelligence systems and use more high-density servers, which generate far more heat and demand stronger cooling systems.

AWS has advised customers to spread workloads across multiple Availability Zones and regions to reduce the risk of large-scale disruptions when outages happen.

Before now, Microsoft helped fund OpenAI’s rapid growth and used early access to its systems to build new tools across Windows, Office, Azure and other products.

Under the new agreement, Microsoft will remain OpenAI’s main cloud partner, but it will no longer have sole rights to host or distribute OpenAI technology.

That means OpenAI can now reach customers already using other cloud providers, instead of asking them to move to Microsoft Azure first.

Even so, OpenAI products are still expected to launch first on Azure unless Microsoft cannot, or chooses not to, support them.

Microsoft will also keep a non-exclusive licence to OpenAI’s intellectual property until 2032.

Another key change concerns money as Microsoft will no longer pay OpenAI a share of revenue from its own AI-related products. At the same time, OpenAI will continue paying Microsoft a 20% revenue share until 2030, though that amount now has a fixed upper limit.

Following the announcement, Microsoft shares fell about 1% in premarket trading on Monday, while Amazon and Alphabet, Google’s parent company, edged higher.

This means Microsoft has lost a strategic advantage, while its competitors now have a stronger path to offer OpenAI products through their own cloud services.

Still, aside from the exclusive deal, Microsoft is deeply tied to OpenAI. The software company owns about 27% of OpenAI, a stake estimated to be worth around $225 billion after OpenAI’s corporate restructuring last year.

The two companies have faced limitations in recent months, with reports in March saying Microsoft had considered going to court over a proposed $50 billion cloud deal involving Amazon and OpenAI that could have conflicted with earlier exclusivity terms.

There were also signs of stress inside OpenAI. Internal discussions reportedly revealed frustration that the old arrangement limited sales opportunities, especially among companies already committed to AWS or Google Cloud.

For OpenAI, the new structure provides room to expand faster and sell across more platforms.

For Microsoft, the deal removes exclusivity but secures long-term access to OpenAI technology while putting clearer limits on future payments.

For customers, it means more choice and less pressure to rely on a single cloud provider.

According to AWS’s official status dashboard update, two facilities in the UAE were directly struck by drones, while one facility in Bahrain sustained physical damage from a nearby drone strike.

The company stated: “These strikes have caused structural damage, disrupted power delivery to our infrastructure, and in some cases required fire suppression activities that resulted in additional water damage.”

This marks one of the first publicly confirmed physical military attacks on a major hyperscale cloud provider’s infrastructure.

The incidents affected the AWS Middle East (UAE) Region (ME-CENTRAL-1) and the AWS Middle East (Bahrain) Region (ME-SOUTH-1), leading to outages and degraded performance for services including EC2, S3, DynamoDB, Lambda, and RDS.

AWS has described the recovery as prolonged and unpredictable due to ongoing regional instability.

Customers have been advised to activate disaster recovery plans and migrate workloads to other AWS regions where possible.

Impact on Businesses and Users

The attacks caused service interruptions for banks, delivery apps, government services, and enterprises across the Gulf that depend on these availability zones.

With data centres becoming strategic targets in modern conflicts, this incident highlights the physical vulnerabilities of cloud infrastructure, even for tech giants like Amazon.

AWS and local authorities have not reported casualties from the strikes.

For African businesses and developers heavily reliant on AWS; common for Nigerian fintechs, startups, and enterprises routing through global cloud providers, this serves as a reminder of geopolitical risks.

Experts recommend multi-region redundancy and regular failover testing to avoid similar disruptions.

Amazon has not released detailed images or extent of damage, citing security reasons.

Recovery efforts were reported as making progress in some areas, but full restoration timelines remain unclear.

Two availability zones in the UAE lost power, while Bahrain experienced connectivity issues. AWS said recovery for both regions could take “many hours,” advising customers to rely on services in other regions while it works to restore power and connectivity.

The UAE incident occurred after sparks and fire were triggered at one of the cloud provider’s zones, prompting an immediate shutdown of power in the affected areas. 

AWS did not confirm whether the outages were related to recent Iranian retaliatory strikes on neighbouring Gulf states, including the UAE and Bahrain, in response to U.S. and Israeli attacks on Iran.

The outages coincided with other challenges across regions, as Iranian strikes in the Gulf damaged infrastructure, including Dubai airport and the Burj Al Arab hotel, prompting questions about the possible link to AWS disruptions.

Separately, Abu Dhabi Commercial Bank (ADCB) reported technical issues affecting its platforms and mobile app users on Monday. 

While the bank did not confirm a direct link to AWS, the timing points to possible ripple effects from the data centre outages.

The newspaper said the incidents resulted from errors involving Amazon’s own tool, known as Kiro. In one case in mid-December, AWS customers experienced a 13-hour interruption.

Engineers had allowed the tool to carry out certain system changes. It then decided to “delete and recreate the environment”, the report said, which led to the disruption.

AWS disputed that account.

In an emailed response to Reuters, a company spokesperson said the disruption was brief and blamed it on user error. “This brief event was the result of user error-specifically misconfigured access controls, not AI.”

The spokesperson added that the interruption was “an extremely limited event” affecting a single service in one of AWS’s two mainland China regions. It did not impact compute, storage, database, AI technologies, or any other AWS services, the company said.

The December incidents follow an outage in October that disrupted Amazon’s cloud operations globally. That earlier failure affected Amazon’s own services and several high-profile apps, including Reddit, Roblox and Snapchat.

AWS is the cloud division of Amazon and supports a large share of the internet’s infrastructure. Because of that reach, even short interruptions can affect millions of users and businesses.

Both Amazon Web Services outages in December have drawn attention because they involved automation tools that can act with limited human input.

Cloud providers have been expanding the use of such systems to manage complex infrastructure. At the same time, customers expect stability and clear accountability when problems occur.

Competitors including Microsoft Azure and Google Cloud are also developing automated tools to manage their platforms.

AWS maintains that the December disruption resulted from misconfigured access controls, not from the coding tool acting on its own.

Market momentum remained stable, marking the fifth consecutive quarter in which growth has remained above 20% highlighting continued strength across the sector.

According to new research from Omdia, this performance reflects a significant shift in the technology landscape as enterprise demand for AI moves beyond early experimentation toward scaled production deployment.

As this transition accelerates, hyperscalers are increasingly redirecting competition away from the incremental gains in model performance and toward platform-level capabilities that support multi-model deployment and ensure the reliable operation of AI agents in real-world environments.

In Q3 2025, AWS, Microsoft Azure, and Google Cloud maintained their market rankings from the previous quarter, collectively accounting for 66% of global cloud infrastructure spending. Together, the three hyperscalers delivered 29% year-on-year growth.

AWS’s growth reaccelerated to 20% year on year quartering Q3 2025, marking its strongest performance since 2022.

Microsoft Azure and Google Cloud also maintained strong momentum, each delivering year-on-year growth of more than 35%.

As enterprise demand for AI continues to materialize, growth in the cloud market is shifting from early-stage experimentation and pilot projects toward the scaled deployment of enterprise-grade AI applications.

Backlog levels among leading cloud providers continued to rise, with AWS, Microsoft, and Google Cloud all reporting further increases in Q3 order backlogs, reinforcing the market’s underlying resilience and healthy demand environment.

Hyperscalers’ AI strategies are evolving from a primary focus on incremental model performance toward more platform-driven and production-ready approaches.

Enterprises are no longer evaluating AI platforms solely on model capabilities, but increasingly on their support for multi-model strategies and agent-based applications.

This shift is accelerating hyperscalers’ move toward platform-level AI capabilities. AWS, Microsoft Azure, and Google Cloud are integrating proprietary foundation models with a growing range of third-party and open-weight models, leveraging managed AI platforms and services such as Amazon Bedrock, Azure AI Foundry, and Vertex AI’s Model Garden to expand support for multi-model adoption.

“Collaboration across the ecosystem remains critical,” said Rachel Brindley, Senior Director at Omdia. “Multi-model support is increasingly viewed as a production requirement rather than a feature, as enterprises seek resilience, cost control, and deployment flexibility across generative AI workloads.”

Meanwhile, hyperscalers are stepping up investment in agent build-and-run capabilities, as real-world deployment continues to prove more complex than early experimentation suggested.

“Many enterprises still lack standardized building blocks that can support business continuity, customer experience, and compliance at the same time, which is slowing the real-world deployment of AI agents,” said Yi Zhang, Senior Analyst at Omdia. “This is where hyperscalers are increasingly stepping in, using platform-led approaches to make it easier for enterprises to build and run agents in production environments.”

Recent launches such as AWS AgentCore and Microsoft’s Agent Framework reflect this direction, providing standardized foundational capabilities that help enterprises more efficiently build, deploy, and operate AI agents in production settings

Amazon Web Services (AWS) led the global cloud infrastructure market with a 32% share and 20% year-on-year revenue growth.

This performance was supported by easing compute supply constraints, alongside incremental demand driven by its partnership with Anthropic.

AWS reported a total backlog of $200 billion by the end of Q3, underscoring sustained demand. Amazon Bedrock continues to evolve rapidly, expanding both model choice and platform capabilities, including support for Claude 4.5, 18 managed open-weight models, and enhanced Guardrails and Data Automation features.

At AWS re:Invent 2025, AWS also introduced the Nova 2 model family, Nova Act, and Nova Forge, strengthening its end-to-end enterprise AI stack from models to agents and automation.

In addition, AWS expanded its regional footprint with the launch of the AWS Asia Pacific (New Zealand) Region in September, adding three availability zones to support local data residency and low-latency workloads.

Microsoft Azure remained the world’s second-largest cloud provider in Q3 2025, holding a 22% market share and delivering robust 40% year-on-year revenue growth. In October, Microsoft renewed its partnership with OpenAI, further anchoring OpenAI’s AI development and deployment on Azure.

Azure AI Foundry continued to broaden its model ecosystem, supporting multiple frontier foundation models, including Claude Opus 4.5, Claude Sonnet 4.5, and Haiku 4.5.

The platform now serves more than 80,000 customers and provides access to over 11,000 models. In October, Microsoft also introduced the Microsoft Agent Framework, enabling enterprises to build and orchestrate multi-agent systems.

Customers such as KPMG are already applying the framework to enhance audit processes, with these production-grade AI deployments contributing to Azure’s overall growth momentum.

In parallel, Microsoft continued to invest in regional infrastructure, announcing plans in November to expand its Azure cloud region in Malaysia and to launch a new Azure datacenter region in India in 2026.

Google Cloud remained the world’s third-largest cloud services provider in Q3 2025, delivering strong 36% year-on-year growth and increasing its market share to 11%. Growth was primarily driven by enterprise AI offerings, with quarterly revenue from this segment reaching several billion dollars.

As of September 30, Google Cloud reported a backlog of $157.7 billion, up sharply from $108.2 billion in Q2, underscoring strengthening demand visibility.

On the platform side, Vertex AI’s Model Garden continued to expand its large-scale AI model portfolio, adding new models such as multimodal variants from the Gemini 2.5 series, Kimi K2 Thinking, and DeepSeek-V3.2. In October 2025, Google

Cloud also launched Gemini Enterprise, an AI platform designed for enterprise customers that integrates the Gemini model family with enterprise-grade AI agents, no-code development tools, and security and governance capabilities.

Omdia defines cloud infrastructure services as the sum of bare metal as a service (BMaaS), infrastructure-as-a-service (IaaS), platform-as-a-service (PaaS) and container-as-a-service (CaaS) and serverless that are hosted by third-party providers and made available to users via the Internet.

The plan, which will begin construction in 2026, aims to add nearly 1.3 gigawatts of computing power, designed to enhance missions ranging from cybersecurity to drug discovery.

“This investment removes the technology barriers that have held government back,” AWS CEO Matt Garman said. “We’re giving agencies expanded access to advanced AI capabilities that will enable them to accelerate critical missions from cybersecurity to drug discovery. This investment removes the technology barriers that have held government back and further positions America to lead in the AI era.”

The new infrastructure will span AWS Top Secret, AWS Secret, and GovCloud regions, providing federal agencies with access to a suite of AI services including Amazon SageMaker for model training and customisation, Amazon Bedrock for deploying models and agents, Amazon Nova, and Anthropic’s Claude chatbot. 

The rollout will also integrate NVIDIA AI chips and AWS Trainium hardware, allowing agencies to run complex simulations, analyse massive datasets, and develop tailored AI solutions with outstanding speed.

AWS has been a long-standing partner of the U.S. government. Its first government cloud infrastructure dates back to 2011, followed by AWS Top Secret-East in 2014, the first air-gapped commercial cloud for classified workloads. 

AWS Secret Region was launched in 2017, accredited to handle all levels of classified data. Today, the company serves over 11,000 federal agencies, making it the largest provider of secure cloud services to the government.

While Amazon tops competitors in the cloud market, companies like Google, Oracle, and Microsoft have been expanding their AI products and services to federal customers. 

Google’s “Google for Government” and enterprise tiers from OpenAI and Anthropic have all been offered at nominal rates to encourage adoption.

Experts also noted the geopolitical implications. “The U.S. is in an AI arms race with China and will significantly increase its AI compute capacity to maintain its lead,” said D.A. Davidson analyst Gil Luria. 

AWS’s new infrastructure will allow agencies to convert previously fragmented intelligence, supply chain, and research data into actionable insights at an unprecedented scale, effectively compressing months of analysis into hours.

Combining high-performance computing with AI enables the project to achieve its mission of transforming how federal agencies operate, supporting missions across national security, scientific research, healthcare, and industrial innovation. 

For the government, this investment by Amazon will boost speed, security, and analytical capacity, strengthening America’s leverage in AI and technology.

The deal, announced Monday, gives OpenAI access to hundreds of thousands of Nvidia GPUs hosted on Amazon’s cloud and is a realignment in the power dynamics of the global tech industry.

The partnership allows OpenAI to immediately begin deploying workloads on AWS infrastructure, which will scale up through 2026 with room for expansion into 2027 and beyond. 

Amazon’s specialised EC2 UltraServers will connect Nvidia’s most advanced processors, H100 and Blackwell chips, through high-speed networks designed for low latency and maximum performance.

For OpenAI, the agreement is part of a goal to build 30 gigawatts of computing capacity, backed by a $1.4 trillion investment plan that highlights the sheer cost of sustaining frontier AI research. 

CEO Sam Altman stressed the scale of the challenge, saying, “Scaling frontier AI requires massive, reliable compute. Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.”

Amazon’s cloud chief, Matt Garman, described the move as a turning point for both companies: “As OpenAI continues to push the boundaries of what’s possible, AWS’s best-in-class infrastructure will serve as a backbone for their AI ambitions.” 

The announcement immediately lifted Amazon’s stock by 5% in premarket trading, adding roughly $330 billion to its market value, its biggest single-day gain in a decade.

The deal follows OpenAI’s recent restructuring, which ended Microsoft’s exclusive cloud arrangement and freed the company to engage multiple infrastructure partners. Although Microsoft still holds a 27% stake in OpenAI, the company has also signed cloud agreements with Google and Oracle, further diversifying its supply chain and reducing reliance on any single provider.

OpenAI’s new partnership with AWS also reveals the competition among the tech giants to control the world’s AI computing backbone. Companies are stockpiling GPUs, expanding data centres, and betting heavily on infrastructure to train and deploy ever-larger models.

Earlier this year, OpenAI’s models became available on Amazon Bedrock, bringing its technology to millions of AWS customers across sectors such as media, health, and data analytics. 

Monday’s announcement pushes that collaboration to an entirely new scale that could change how global AI systems are built and deployed.

With the world’s biggest tech firms spending at unprecedented levels, analysts say the OpenAI–Amazon partnership represents both ambition and risk. 

The scale of investment required to power AI systems is nearing levels once reserved for national infrastructure projects, leading to concerns that the competition could create the industry’s next financial bubble.

Still, for now, the partnership gives OpenAI what it needs most, raw computing muscle. And for Amazon, it offers a decisive edge in the most lucrative race in modern technology: the vision to power the intelligence of the future.

The company confirmed this on Tuesday, saying the layoffs are part of a restructuring aimed at simplifying operations and enhancing focus on its long-term priorities.

Beth Galetti, Amazon’s senior vice president of People Experience and Technology, said in a company blog post, “The reductions we’re sharing today are a continuation of this work to get even stronger by further reducing bureaucracy, removing layers, and shifting resources to ensure we’re investing in our biggest bets and what matters most to our customers’ current and future needs.”

The cuts will affect various divisions across Amazon’s corporate structure, including its human resources unit, known internally as People Experience and Technology (PXT), as well as operations, devices, services, and Amazon Web Services (AWS). 

According to people familiar with the matter, some managers were briefed on Monday and trained on how to handle employee notifications, which began rolling out Tuesday morning.

This new round of layoffs follows a series of job reductions that began in late 2022, when the company eliminated about 27,000 corporate positions due to slowing growth after the pandemic boom. 

At the time, CEO Andy Jassy described the move as necessary to “right-size” the company following years of rapid expansion.

In recent months, Jassy has doubled down on efforts to reduce what he calls “excess bureaucracy.” He previously launched an internal anonymous feedback line that led to over 450 process changes, all aimed at improving efficiency. 

The CEO also hinted in June that advances in automation and artificial intelligence could reduce the need for certain corporate functions, noting that Amazon was “still larger than it needs to be” following its pandemic-era hiring spree.

The company’s focus on automation and AI-driven productivity has already begun to change its workforce. Sky Canaves, an eMarketer analyst, said, “This latest move signals that Amazon is likely realising enough AI-driven productivity gains within corporate teams to support a substantial reduction in force.”

While the 14,000 layoffs represent less than 5% of Amazon’s 350,000-strong corporate workforce, internal projections seen by Reuters suggest the number could eventually climb to as high as 30,000, depending on business priorities and financial outcomes in the coming quarters.

Amazon currently employs about 1.55 million people worldwide, including warehouse and logistics workers. Despite these corporate reductions, the company still plans to hire about 250,000 seasonal workers ahead of the holiday shopping season, maintaining the same hiring scale as the past two years.

The announcement comes just days before Amazon’s third-quarter earnings report, scheduled for Thursday.

Analysts expect modest growth in AWS, the company’s most profitable arm, after it reported a 17.5% year-over-year increase last quarter, lagging behind competitors Microsoft Azure and Google Cloud.

The disruption, which originated from AWS’s US-EAST-1 data centre in Ashburn, Virginia, lasted several hours and left millions of users unable to access services such as Snapchat, Reddit, Venmo, Zoom, and even Amazon’s own Prime Video and Alexa. 

According to outage tracking site Downdetector, more than four million users globally reported problems during the incident.

By 3:01 p.m. PT (10:00 p.m. GMT), Amazon confirmed that “all AWS services returned to normal operations. Some services such as AWS Config, Redshift, and Connect continue to have a backlog of messages that they will finish processing over the next few hours.”

Root Cause: Network Health Monitor Failure

AWS identified the source of the failure as a malfunction in a subsystem that monitors the health of its network load balancers, a key component responsible for distributing traffic across multiple servers. 

The fault within its Elastic Compute Cloud (EC2) network also triggered a Domain Name System (DNS) error that prevented apps from locating DynamoDB, one of AWS’s most critical database services.

The outage, Amazon confirmed, “originated from within the EC2 internal network,” once again placing focus on the US-EAST-1 cluster, the same region linked to similar breakdowns in 2020 and 2021. Despite repeated failures in this location, the company has yet to explain why the data centre is still a recurring weak point.

Global Impact: From Banks to Gaming Platforms

The outage exposed the scale of global dependence on AWS. Major banks, telecom firms, and government agencies across Europe and North America reported downtime. In Britain, Lloyds Bank, Bank of Scotland, Vodafone, BT, and even the UK tax and customs authority (HMRC) experienced service interruptions.

For consumers, the impact was immediate and across-the-board. Social platforms like Snapchat and Reddit went dark, while fintech platforms such as Venmo, Robinhood, and Coinbase froze transactions. 

Gaming networks, including Fortnite, Roblox, Clash Royale, and Clash of Clans, were also affected. Lyft users in the United States were unable to book rides, and airline check-in systems at LaGuardia Airport in New York temporarily failed.

Artificial intelligence startup Perplexity, cryptocurrency exchange Coinbase, and trading app Robinhood were among those confirming that AWS was at the root of their service disruptions. Even Signal, the encrypted messaging platform, was hit. “This outage once again highlights the dependency we have on relatively fragile infrastructures,” said Jake Moore, global cybersecurity advisor at European security firm ESET.

Experts Warn of Fragile Cloud Dependency

The scale of the outage has reignited talks about the world’s overreliance on a handful of cloud providers, Amazon, Microsoft Azure, and Google Cloud, which collectively power a vast portion of global digital infrastructure.

“This was the third major AWS outage in five years linked to the same region,” said Ken Birman, professor of Computer Science at Cornell University. “When people cut costs and cut corners to try to get an application up, and then forget that they skipped that last step and didn’t really protect against an outage, those companies are the ones who really ought to be scrutinised later.”

Ryan Griffin, U.S. cyber practice leader at McGill and Partners, added that “for major businesses, hours of cloud downtime translate to millions in lost productivity and revenue.”

Recurring Weakness in US-EAST-1

The Ashburn-based US-EAST-1 cluster is AWS’s oldest and largest region, usually set as the default for many of its services. This makes it a single point of failure, and experts argue that the region’s recurring issues demonstrate the risks of centralised cloud design.

Nishanth Sastry, director of Research at the University of Surrey’s Department of Computer Science, said the incident was predictable. “The main reason for this issue is that all these big companies have relied on just one service,” he noted.

Market Reaction and Next Steps

Despite the global disruption, Wall Street remained largely indifferent. Amazon’s shares closed 1.6% higher at $216.48 on Monday. The company has promised to release a detailed post-event summary but has not clarified whether structural changes will be made to prevent a repeat of the incident.

The Amazon Web Services (AWS) global outage, the largest since last year’s CrowdStrike malfunction that crippled hospitals, banks, and airports, tells us that the cloud’s convenience comes with fragility.