The operations stories enterprise IT leaders actually read.
Volume 342, Issue 4 · 8 articles
The recent AI Infra Summit had a lot of great presentations on innovations happening at the chip, memory, system, network, and storage levels. But there’s also quite a bit happening further down the five-layer cake, in the underlying power and cooling that’s needed to make the racks of AI infrastructure work.
Alex Cordovil, the research director at the Dell’Oro Group, presented a lot of good information during his 2026 AI Infra Summit session titled “The Market Outlook – What the Last 12 Months Tells Us About the Next 24.”
Breakthroughs in AI have led to a mad dash to build massive data centers all over the world. Just five years ago, adding 5GW of data center capacity per year was considered bold. Now Dell’Oro Group is projecting 60GW per year of new data center capacity by 2030, which is near the top of projections of other industry analysts, including Bain, McKinsey, JP Morgan, and BCG.
There are headwinds to the buildout, including the rising cost of borrowing as well as community backlash against data centers, Cordovil said. While many data center projects are projected to go off the rails, the overall trend appears to be accelerating growth.
Rupert Goodwins argues cloud resilience fails against real-world threats like war, weak backups, and an AI-driven storage crunch.
Columnist Rupert Goodwins argues cloud infrastructure's reputation for resilience breaks down against real-world threats. He cites AWS permanently losing data in its Bahrain region due to military damage, showing that resilience within one region isn't the same as an independent offsite backup.
He also points to the UK's air traffic control system failing without adequate backup, reportedly because redundant storage was deemed too costly.
Separately, AI datacenter expansion has consumed global storage capacity, with Western Digital exhausting its entire 2026 hard-drive production by February, making backups more expensive.
Goodwins urges organizations to price backup risk properly instead of treating resilience as someone else's problem.
Gartner's 2026 Magic Quadrant ranks Everpure, HPE, and Huawei as leaders in consumption-based infrastructure services.
Gartner's 2026 Magic Quadrant for Infrastructure Platform Consumption Services evaluates vendors offering storage and compute on a consumption-based, as-a-service model.
Everpure leads the Leaders quadrant, followed by HPE GreenLake and Huawei, which is new to the top tier this year after Everpure overtook HPE for first place.
Dell, Hitachi Vantara, IBM, Lenovo, and NetApp sit in the Challengers quadrant, while Fujitsu is a niche player.
Gartner projects consumption-based storage services will replace 50% of on-premises enterprise storage capex by 2029, up from 20% in 2026, as the market consolidates around integrated, SLA-backed offerings.
A Register columnist recounts a seven-hour Windows 11 update ordeal after months of laptop inactivity.
Register writer Richard Speed describes a roughly seven-hour update process on a Windows 11 laptop left idle for months, requiring numerous component updates, firmware patches, driver updates, and OS upgrades with repeated restarts.
He criticizes Windows' longstanding inaccurate progress bars, which stall at 65% for an hour before jumping to 98% and then hanging again.
The piece contrasts Microsoft's marketing of Windows 11 efficiency with the reality of lengthy, unpredictable update sessions, arguing the episode reflects a broader industry habit of shipping fixes later ('fix it in production') rather than ensuring quality before release.
As AI racks exceed 250kW, near-total liquid heat capture is turning fanless server cooling from niche to standard.
As AI server racks exceed 250 kW of power density, hybrid liquid-air cooling becomes impractical because the remaining 30% air-cooled load still equals 75 kW, forcing costly parallel cooling systems. CoolIT's approach captures nearly all heat in liquid, cutting air cooling below 1% of the load so servers can run fanless. The challenge extends beyond processors to memory, networking, storage and power components, each neding tailored thermal handling, unified through a modular coldplate design proven across six generations.
As rack densities climb toward 1 MW, the piece argues near-total liquid heat capture will move from a premium option to the industry standard for flagship systems through 2028.
Deferred IT maintenance has become a compounding business risk that AI adoption accelerates rather than solves.
Author Syed Ali argues that IT maintenance backlogs - including overdue patches, ownerless devices and integrations, and undocumented configuration changes - have evolved from a technical nuisance into material business risk.
Citing an Uptime Institute survey where 57% of respondents reported outage costs exceeding $100,000, Ali warns AI automation can't fix maintenance debt and actually accelerates its consequences when ownership records are incomplete.
He recommends CIOs shift from tracking task completion to assessing business risk exposure and reporting exposure rather than completion rates.
A Basel Action Network report warns AI infrastructure could generate 395-617 million tonnes of e-waste by 2050.
The Basel Action Network estimates AI infrastructure buildout from 2025-2050 could produce 395 to 617 million tonnes of electronic waste, enough to fill 15-23 million shipping containers that laid end-to-end would circle Earth about six times.
The report says prior estimates undercounted waste by ignoring power, cooling, and networking gear, which make up roughly 87% of datacenter equipment weight versus GPUs alone.
It assumes aggressive replacement cycles, such as AI accelerators every 2-3 years versus 5-7 years for traditional servers.
The group warns no hyperscaler, government, or international body has published a plan for handling AI-driven e-waste at this scale, and current recycling infrastructure is inadequate.
TechTarget reviews eight customer support knowledge base platforms, from Document360 to Zoho Desk, with pricing and features.
A customer-facing knowledge base serves as a pivotal tool for modern businesses aiming to scale customer support, enhance customer experience (CX), and improve operational efficiency.
By providing centralized, easily accessible documentation—such as FAQs, setup guides, troubleshooting articles, and product manuals—organizations empower customers to resolve issues independently.
This shift toward self-service significantly reduces incoming ticket volumes for support teams, shortens resolution times, and lowers operational support costs.
Choosing the right knowledge base software requires balancing content creation capabilities, search functionality, software integration, AI automation, and overall usability.
More from Operations · All IT section newsletters · This week