Cloud Computing and Data Centers

AWS, Azure, Google Cloud, cloud infrastructure, containers, and data centers.

124 Articles

SpaceX Gas Turbine Foundry Could Accelerate AI Infrastructure at the Cost of Pollution

Elon Musk says a secret foundry SpaceX is building in Bastrop, Texas, could accelerate the commissioning of gas turbines by roughly 18 months by manufacturing their blades in-house. But this solution to reducing energy bottlenecks at AI data centers comes amid lawsuits and health studies concerning turbine emissions.

2026-08-30

Cloud Computing and Data Centers

Nvidia’s AI Advantage Expands from GPUs to Managing the Entire System

The analysis argues that Nvidia’s competitive strength no longer depends solely on graphics processing units, but extends to the processors, memory, storage, and networking that organize data movement within hyperscale data centers. As AI workloads move into the gigawatt range, the efficiency of coordinating data has become a new competitive layer that may give Nvidia an advantage despite intensifying competition at the chip level itself.

2026-08-29

Cloud Computing and Data Centers

How to Get Ahead of Demand Spikes: Predictive Scaling for GPU Workloads on Kubernetes

Two Adobe engineers present a design for proactive GPU workload scaling on Kubernetes, intended to compensate for slow node provisioning that causes reactive scaling to lag behind demand waves. The design relies on a Bi-LSTM model, a spike detector, and a gradual scaler, with shadow tests showing 85% accuracy within a ±10% margin ten minutes ahead of demand.

2026-08-28

Cloud Computing and Data Centers

Meta Explains How Closed-Loop Liquid Cooling Helps Power AI Infrastructure

Meta uses closed-loop liquid cooling in most of its new AI-optimized data centers to handle rising heat while reducing water and space consumption. The company says a reinforcement-learning trial reduced cooling-fan energy consumption by 20% and water consumption by 4% at one of its data centers.

2026-08-27

Cloud Computing and Data Centers

How Cloudflare Reduced DNS Store Memory Usage by 56%

Cloudflare redesigned how DNS records are stored in the Big Pineapple platform, which powers 1.1.1.1 and other DNS services, reducing memory per entry from 953 to 420 bytes and saving approximately 100 terabytes across its fleet. The insertion rate also increased by 43%, while lookup latency decreased by 19%.

2026-08-27

Cloud Computing and Data Centers

AWS and NVIDIA Expand Their Partnership to Deploy Two Million GPUs and Build Infrastructure for Government AI

AWS plans to deploy an additional two million NVIDIA graphics processing units during 2027 and 2028, while also introducing Vera processors and supporting an architecture that combines Trainium and NVIDIA. The two companies also plan to build AI factories for the U.S. government containing 100,000 GPUs on a secure architecture.

2026-08-26

Cloud Computing and Data Centers

Amazon EC2 After 20 Years: From a Single Virtual Server to an Operating Layer for AI

AWS reviews the history of Amazon EC2 since the launch of its beta version in 2006, when it was limited to one server type and one Region, through to more than 1,200 instance types across 39 Regions. The latest milestones show how the service expanded from general-purpose computing to AI chips, Mac computers, deployment outside AWS Regions, and security isolation mechanisms supported by formal verification.

2026-08-25

Cloud Computing and Data Centers

OpenAI Data Center Executive Departs Amid Wave of Leadership Changes

Chris Malone, who led data center strategy at OpenAI, has left the company after approximately 18 months since joining it, as the company reorganized its infrastructure team. The departure is part of a series that has exceeded 13 executive exits during 2026, while OpenAI prepares for increased scrutiny ahead of a potential public offering.

2026-08-25

Cloud Computing and Data Centers

How to Learn Kubernetes Without Trying to Absorb the Entire Ecosystem at Once

Joep Piscaer suggests building an initial mental model of Kubernetes by understanding five pillars: desired state and reconciliation, separating the control plane from worker nodes, network layers, requests and limits, and the role of CNI and CSI plugins. The core idea is to defer advanced topics until a practical need for them arises.

2026-08-25

Cloud Computing and Data Centers

Furukawa Electric to Invest Approximately ¥100 Billion to Increase Optical Fiber Production Linked to AI Data Centers

Furukawa Electric announced an investment of approximately ¥100 billion to strengthen the production of optical fiber and fiber-optic cables, responding to the expansion of data centers and demand related to artificial intelligence. Additional production is scheduled to begin gradually in fiscal 2028, raising total production capacity to approximately twice its fiscal 2025 level.

2026-08-25

Cloud Computing and Data Centers

How Cloudflare Tested the EmDash Platform Under Production Load Before Migrating Its Blog

Cloudflare migrated its blog to EmDash, an Astro-based content management system, using multiple caching layers, load tests, and a gradual rollout to reduce migration risks. The company reports early results including handling up to 850 requests per second, as well as an MCP experiment for accessing and managing content through agents.

2026-08-24

Cloud Computing and Data Centers

How Atlassian Is Building an Automated Root Cause Analysis System for Cloud Incidents

Atlassian presents a root cause analysis methodology based on linking metrics, logs, and traces temporally and through a service dependency graph. The goal is to turn large volumes of monitoring data into ranked, verifiable hypotheses while preserving the engineer’s role in validation and decision-making.

2026-08-24

Cloud Computing and Data Centers

AWS Bundles Weekly Updates Including Student Rewards and a Local Zone in Las Vegas

AWS announced the Student Rewards program for university students and launched a new Local Zone in Las Vegas, along with updates to Bedrock, Glue, CloudShell, and Auto Scaling services. The roundup also includes the general availability of AgentCore payments and expansions to the AWS regional infrastructure.

2026-08-24

Cloud Computing and Data Centers

AWS Launches Glue 6.0 at 30% Lower Cost with Full Apache Iceberg v3 Support

AWS announced the general availability of AWS Glue 6.0, with prices reduced by 30% compared with previous versions and full support for Apache Iceberg v3 capabilities. The release is based on Apache Spark 4.1, Python 3.12, and Scala 2.13, and adds improvements for semi-structured data processing, PySpark function development, and single-digit-millisecond-latency streaming.

2026-08-21

Cloud Computing and Data Centers

GitHub Reveals Causes of August 17 Outage and Announces Measures to Strengthen Platform Reliability

GitHub said the August 17 outage lasted 7 hours and 47 minutes and was primarily caused by a critical component at a United States data center being unable to keep up with peak demand, disrupting the platform’s services worldwide. The company plans to increase capacity, isolate critical systems, and strengthen the management of retry operations and responses to sudden load spikes.

2026-08-20

Cloud Computing and Data Centers

Can Urine Really Be Used to Cool Data Centers? The Joke Leads to a Real Solution

A marketing joke about using urine to cool artificial intelligence data centers has opened a discussion about using recycled water. Experts point out that the viable solution is treating wastewater, not pumping urine directly, but expanding this approach requires substantial infrastructure and investment.

2026-08-20

Cloud Computing and Data Centers

Three Principles for Limiting Cascading Failure in Distributed Systems

Sam Newman explains how a small failure in one component can turn into a widespread collapse in distributed systems, citing the Ronan Point incident, an AWS outage, and a problem at a used-car sales website. He proposes three practical strategies: reducing risks, strengthening components, and limiting the coupling between them.

2026-08-19

Cloud Computing and Data Centers

Cloud-Native Platform Sovereignty Begins with a Multi-Level Architecture, Not Just Region Selection

A CNCF article analyzes how control, workload execution, observability, and workflows can be separated across independent levels and clusters to improve the demonstrability of cloud sovereignty. OpenChoreo presents a practical model based on outbound connections and declarative configurations, while legal and operational constraints remain.

2026-08-18