TLDR IT 2026-08-24

Most enterprises have adopted AI tools, but they've sped up individuals without fixing how teams work. And that gap costs billions. Atlassian and AWS surveyed 500 executives to see what separates leaders closing the gapfrom those still paying the fragmentation tax. Inside:

• Why 55% of orgs are stuck in siloed work and how to break out
• The connected operating model that drives nearly 10X better collaboration outcomes
• Making the most of cloud and AI investments to drive lasting org-wide impact

Get the whitepaper

PJM is proposing rules that would make some new large-load customers, including AI data centers, interruptible when electricity supply gets tight. The proposal highlights how power availability is becoming a limiting factor for AI infrastructure growth.

Alibaba plans to raise roughly $10 billion to expand its AI infrastructure, chips, data centers, and model development. The raise comes days after the company reported a 75% jump in capex tied largely to AI.

Sam Altman discusses OpenAI's early non-consensus bets, scaling compute, AI adoption, and why society may absorb AI more slowly than model progress suggests. He also argues AI could enable a boom in small businesses while warning against concentrating control of the technology among too few players.

Anthropic lays out how AI changes software development beyond simply adding coding assistants, including planning, implementation, review, and operations. It argues teams need to redesign development workflows around agents rather than bolt them onto existing processes.

1Password's recent survey found that 47% of developers have seen their AI agents follow instructions embedded in untrusted content, and 74% have witnessed unintended consequences from agents.

Learn more about the security risks of agentic AI, and what developers can do to use agents securely.

Nvidia made a minority investment in Cloverleaf Infrastructure. The companies have partnered to develop AI factories across the US. Cloverleaf says its pipeline already includes multiple gigawatt-scale infrastructure projects.

Ramp has opened the internal model-routing infrastructure it uses across more than 100 AI use cases as a standalone product. Router can automatically choose models based on cost, quality, latency, and availability. Ramp says it cut its own LLM costs by roughly 30%.

The GitLab 19.3 update is aimed at helping enterprises scale agentic software development while retaining control over data, permissions, and spending. The release adds single-tenant deployment for the Duo Agent Platform, Secrets Manager support beyond pipelines, bulk SAST remediation, plain-language custom flow creation, and usage caps for agentic AI.

Enterprise data stacks were largely designed around human access patterns, but agents need reliable real-time access across systems. That is turning data integration and context infrastructure into a prerequisite for successful agent deployments.

Enterprises risk repeating the cloud cost-management mistakes of the last decade as expensive GPU infrastructure proliferates. AI teams increasingly need FinOps-style controls for utilization, capacity planning, and workload economics.

Anthropic plans to let enterprises keep required retained data inside their own cloud infrastructure rather than Anthropic's systems.

The release adds TLS certificate verification for vCenter connections during cloud-drive operations and includes several Kubernetes and disaster-recovery fixes.