Billions in Data Centers and the First Global AI Agent Vulnerability — This Week's Roundup
World of Big Data Expo · weekly review · 2026-09-21
Weekly Roundup

Billions in Data Centers and the First Global AI Agent Vulnerability — This Week's Roundup

This week in AI, big data, and cybersecurity proves that the investment boom and new threats are growing at exactly the same pace. On one hand, we have record-breaking spending on data center infrastructure in Poland; on the other, a serious security vulnerability hitting the most popular AI tools for developers. Here's what happened.

01 · Data Center

Poland's Data Center Market Is Growing Faster Than Ever

Let's start with some impressive numbers. According to CRN Polska [1], spending on data center infrastructure in Poland jumped 92% year over year in the second quarter of 2026. This growth isn't being driven by a single factor — several trends are converging at once. First, prices for GPU-equipped servers have soared by 43.6%, which alone significantly inflates the value of investments being made. Second, a new generation of hardware — Nvidia Blackwell Ultra — is hitting the market, and data center operators are deploying it to keep up with growing demand for AI computing power.

This trend isn't limited to Poland, either. Dell'Oro Group, an analyst firm tracking the global infrastructure market, has raised its forecast for data center spending through 2030 from $1.7 trillion to over $3 trillion. Doubling that forecast in such a short time shows just how much the growth of artificial intelligence and big data is reshaping investment priorities across the entire IT industry — from cloud operators to hardware vendors to companies building their own compute resources to power their own models and services.

It's worth noting that this surge in spending isn't just good news for hardware vendors — it's also a signal to companies planning their own AI deployments that computing costs could keep rising over the coming quarters, and availability of the latest hardware will likely remain limited.

Close-up of a screen showing cybersecurity data and code — illustrating the analysis of a security vulnerability / system attack.
Source: Pexels · Author: Tima Miroshnichenko · License: Pexels License (bezpłatna, dozwolone użycie komercyjne)

02 · Cybersecurity

Plugin4Shell: A Zero-Click Attack That Bypasses Coding Agent Safeguards

Given how fast investment in computing power is growing, it's all the more striking how easily that power can be turned against the very developers building the software. Security researchers have disclosed a vulnerability dubbed "Plugin4Shell" [2] — described as the first attack of its kind targeting the supply chain of AI-based coding agents. The mechanism is especially concerning: the vulnerability bypasses a safeguard known as SHA pinning, which was meant to guarantee the integrity of downloaded code, and allows malicious code to execute without any user interaction whatsoever — hence the term "zero-click."

The issue affects four leading tools used by developers working with AI assistants: Claude Code, Codex, GitHub Copilot, and Gemini CLI. The good news is that Claude Code and Codex have already been patched by their creators. The bad news — GitHub Copilot and Gemini CLI, which Google is in the process of phasing out, remained vulnerable to the attack at the time the reports were published. It's a clear signal to the entire industry that as AI agents become part of everyday development work, a completely new category of software supply chain risk is emerging, and security teams need to treat these tools with the same vigilance as any other dependency in a project.

A symbolic juxtaposition of two sides of the same phenomenon: on one side, a growing data center and investment charts; on the other, a cracked cybersecurity shield.
Source: image generated by Cloudflare Workers AI (@cf/black-forest-labs/flux-1-schnell) · License: editorial material (AI image)

03

Two Sides of the Same Revolution

These two stories, though seemingly unrelated, capture the current moment in the industry well: capital is flowing into AI infrastructure faster than ever, but the pace of new tool adoption sometimes outstrips the pace of securing those tools against misuse. Expanding data centers and increasingly powerful coding agents are two sides of the same coin — the more computing power and automation we have, the greater the responsibility for security and thoughtful deployment.

Want to stay on top of trends like these, talk to data center infrastructure and cybersecurity experts, and see the latest AI and big data solutions in person? Join us at World of AI & Big Data Expo — the International Trade Fair for Artificial Intelligence and Database Technologies. See you there!

Invitation

Billion-dollar investments in data centers and vulnerabilities in AI tools for developers are just a preview of what you'll see at World of AI & Big Data Expo. Meet infrastructure providers, cybersecurity experts, and AI tool creators all in one place. Talk to them about how to safely scale projects built on artificial intelligence and big data. Reserve your spot today.

World of AI & Big Data Expo 2026
World of Big Data Expo · weekly review · 2026-09-21