The US Is Making Data Centers Pay For Grid Access — But Power Bills Are Still Going Up

bgr.com Jun 25, 2026

The US Is Making Data Centers Pay For Grid Access — But Power Bills Are Still Going Up

Curated by Shubham · TechCapsules · Jun 25, 2026

AI-summarised brief · reviewed before publication

The US Federal Energy Regulatory Commission issued orders to speed up the integration of large energy users, such as AI data centers, into regional grid operators. This decision aims to prevent cost shifting, where data center owners underpay for electricity costs, forcing other customers to pay the rest. The orders require energy companies to explain their tariffs and provide enough energy for data centers and other customers. The move comes as AI-driven electricity bills spike, despite companies behind AI footing most of the bill. The orders are subject to change, with energy companies given 60 and 30 days to respond.

💡 Why It Matters

· FERC's orders hold data centers accountable for their energy usage, ensuring fairness in electricity cost distribution.
· This shift in policy addresses the financial burden on ordinary consumers.

Read full article on bgr.com

How space weather could bust the AI boom

America's electric grid is facing unprecedented strain due to the rapid expansion of power-hungry AI data centers. Utilities are struggling to keep up, and the increasing energy demand is causing consumer electric bills to rise. A lesser-discussed risk is the vulnerability of the grid to space weather, which could cripple portions of the grid by damaging high-voltage transformers. Scientists have warned of the potential consequences of a severe solar storm, and investment in space weather forecasting and preparedness has been inadequate.

💡 The tech giants behind the AI revolution must prioritize space weather forecasting and preparedness to protect the electric grid and AI systems critical to national security and society. By articulating the importance of space weather forecasting to lawmakers and investing in grid resilience, they can mitigate the risks of catastrophic damage from severe solar storms.

windowscentral.com · Jun 24

NVIDIA claims its ‘next‑gen AI infrastructure’ offers a fix for data centers’…

NVIDIA has announced its next-generation AI infrastructure, claiming it can solve the water consumption challenge for data centers. The system uses a recirculated mix of water and propylene, similar to automotive antifreeze, to cool data centers, reducing the need for additional chilling equipment. This allows data centers to operate at higher temperatures, making them more reliable in hotter environments. According to NVIDIA's Chief Sustainability Officer, Josh Parker, the water consumption challenge for data centers is "largely solved." The new infrastructure is expected to reduce both operational costs and environmental impact.

💡 The success of NVIDIA's next-gen AI infrastructure could set a new standard for the industry, forcing companies to reevaluate their approach to data center cooling. This could lead to a significant reduction in water consumption and energy usage, mitigating the environmental impact of data centers and potentially averting costly disputes with local communities.

fool.ca · Jun 24

Why Data Centre Stocks Could Be the Smartest Buy on the TSX

Electricity producers and utilities in Canada are well-positioned to benefit from the data centre boom, as these facilities require a significant amount of power to operate. Capital Power, ATCO, and TransAlta offer different ways to invest in this trend, with Capital Power providing a direct route into data centre electricity demand through its power generation assets and contracts. ATCO offers exposure to regulated utility infrastructure, while TransAlta is a higher-volatility option with potential for growth. These stocks may provide a unique investment opportunity, driven by the increasing demand for reliable power from data centres. Key details include Capital Power's recent earnings growth and contracting success.

💡 Data centres' insatiable demand for power creates a steady revenue stream for these utilities, supporting their dividends and growth prospects. By investing in these stocks, investors can tap into the data centre boom without directly investing in tech companies.

semiconductor-digest.com · Jun 23

Addressing Liquid Cooling Gaps in AI Data Centers

xMEMS engineers Tom Tarter and Henry Neynavaee highlight the limitations of liquid cooling in AI data centers. Liquid cooling effectively manages heat from high-power processors but leaves secondary components exposed, leading to temperature-related issues and throttling. To address this, the engineers propose localized, solid-state micro-cooling that targets specific hotspots with minimal power consumption. This approach aims to maintain thermal balance across the entire server, rather than just cooling the highest-power chips.

💡 Closing thermal gaps in AI data centers can significantly reduce operating expenses, particularly for hyperscale operators. By minimizing the need for high-power fans, data centers can lower their energy consumption and extend the lifespan of critical components.

teslarati.com · Jun 23

Tesla just trademarked MEGAPOD: here’s what it is

Tesla has trademarked 'MEGAPOD' with the USPTO, listing it as modular data center hardware systems for artificial intelligence computing. The application describes MEGAPOD as self-contained units integrating servers, AI processing hardware, networking components, power distribution, and cooling systems. This filing aligns with Tesla's AI compute strategy, which includes deploying AI agents on parked vehicles and dedicated compute units at Supercharger stations. The modular hardware platform could support scalable, rapid deployment of distributed compute resources, complementing Tesla's other AI infrastructure efforts.

💡 The MEGAPOD trademark application suggests Tesla is preparing to launch a modular AI computing platform, potentially enabling edge or regional AI inference without relying on centralized data centers. This move could accelerate Tesla's AI ambitions, particularly in the context of its Digital Optimus project, which aims to develop AI agents capable of handling complex digital tasks.

teslarati.com · Jun 23

SpaceX confirms third massive compute deal at Colossus data center

SpaceX has confirmed its third massive compute deal at the Colossus data center in Southaven, Mississippi. The deal, worth approximately $6.3 billion, grants Reflection AI immediate access to NVIDIA GB300 chips at SpaceX's Colossus 2 data center. In return, Reflection will pay SpaceX $150 million per month, starting July 1, with either party able to terminate the agreement after 90 days' notice. This deal highlights SpaceX's strategy of commercializing its Colossus supercomputing infrastructure, originally developed for Elon Musk's Grok AI models.

💡 The Reflection contract underscores SpaceX's growing influence in the AI infrastructure market, where demand for advanced training and inference capacity is surging amid chip shortages. By transforming its data centers into flexible revenue sources, SpaceX is diversifying its income streams beyond traditional rocket launches and Starlink operations, positioning itself as a leader in the lucrative AI sector.