Could AI make your fridge explode? The impact of AI data centres on the grid and your appliances

AI-powered data centres are pushing electricity networks to their limits. But recent lessons from the US, Ireland and even Queensland show that, with the right planning, data centres could support rather than strain the grid.

Artificial Intelligence (AI) is driving a global surge in data centre construction and energy demand.¹ Unlike traditional cloud facilities, AI-focused data centres consume vast amounts of electricity and exhibit highly variable power usage patterns.² Their rapid growth is raising concerns about grid capacity and power quality.³ Grid operators and policymakers are now grappling with how to meet the energy needs of AI data centres without compromising reliability for other users.⁴

What are AI data centres?

AI data centres are high-powered facilities that house the technology and systems needed to run AI.⁵ They contain ultra-fast computers, high-capacity networks, and special cooling systems to handle the intense processing demands involved in developing and operating AI tools.⁶

A single facility can host tens of thousands of AI data servers working in unison, with total power consumption reaching into the hundreds of megawatts.⁷ In some cases, large AI data centres consume as much electricity as entire cities or small states.⁸

This high energy intensity, driven by constant operation and heavy cooling requirements, makes AI data centres one of the most power-hungry infrastructure ever built. AEMO predicts that by 2050, data centres will account for 12% of NEM demand.⁹

What AI data centres are doing to the grid and your appliances

Have you ever noticed your fridge or dishwasher sounding unusually loud? AI may be the reason.¹⁰ Unlike traditional data centres that draw power at a steady rate, AI data centres, powered by thousands of high-performance chips, consume electricity in short, concentrated bursts, especially during AI model training.¹¹ These sudden, high-density loads introduce new stresses on the grid, creating uneven and spiky energy usage patterns.¹²

These fluctuations cause what’s known as ‘bad harmonics’, a type of electrical irregularity that occurs when voltage and current waveforms stray from their usual pattern.¹³ Although harmonics are rarely perfect (a small amount of distortion under standard grid conditions is normal), studies have shown that areas within 80 kilometres of major AI data centres experience significantly higher levels of grid distortions.¹⁴ This leads to poorer power quality for everyday appliances with electric motors, like refrigerators, washing machines, and air conditioners, leading to potential malfunctions, overheating, and electrical fires.¹⁵

Case study: Data Center Alley

Ashburn, Virginia (often called ‘Data Center Alley’) hosts the world’s largest concentration of data centres.¹⁶ The rapid expansion of AI and cloud data centres in the region has pushed their combined electricity supply capacity to around four gigawatts — exceeding the total data centre capacity in London and Tokyo.¹⁷

In July 2024, a minor grid fault triggered an automatic emergency response at about 60 data centres in Northern Virginia, simultaneously disconnecting 1,500 MW of load from the grid.¹⁸ This sudden drop of power caused a measurable frequency spike on the Eastern U.S. power system.¹⁹ While no major blackout occurred, the incident highlighted how a concentration of data centres can amplify grid instability.²⁰

In response, local authorities have slowed approvals for new facilities until infrastructure upgrades catch up.  Grid regulators are also reviewing new rules to prevent future “mass disconnection” events.²¹ Meanwhile, community concern is growing as residents face the side effects from the concentration of data centres, including noise, land use pressure, and worsening power quality.²²

Addressing these challenges

Addressing the impact of AI data centres on electricity grids will require a combination of technology, infrastructure investment, and regulatory reform.²³ The core challenge is not just how much power these centres use, but how unpredictably they use it.²⁴

Solving these issues starts with improving how electricity is delivered and managed.²⁵ Technological solutions such as battery energy storage systems, dynamic voltage regulators, and harmonic filters can help smooth out erratic power flows and reduce the electrical “noise” that damages sensitive equipment.²⁶

Case studies: Microsoft and Quinbrook Infrastructure

Ireland has become a hub for data centres and is rapidly expanding its wind power capacity, introducing greater variability into the electricity supply.²⁷ To address both challenges, Microsoft is transforming its Dublin data centre campus into a grid asset.²⁸ The facility’s large banks of lithium-ion backup batteries (typically reserved for emergency power) have been upgraded and certified to help stabilise the grid.²⁹

In a first-of-its-kind arrangement, these batteries will respond to grid signals and discharge electricity when there’s a shortfall in wind or other generation.³⁰ This supports the local transmission operator in maintaining reliable service during fluctuations in renewable output and reduces reliance on fossil-fuelled generators for reserve power.³¹ Importantly, it reduces reliance on fossil-fuelled generators for providing reserve power.³² Microsoft’s Dublin project shows one innovative solution: turning data centres’ existing battery infrastructure into a buffer for the grid, creating a win-win for both reliability and sustainability.³³

In South-East Queensland, the “Supernode” project is taking a similar approach on a larger scale. Developer Quinbrook Infrastructure is building an 800 MW/ 2000 MWh battery energy storage system to support a planned $2.5 billion data centre campus near Brisbane.³⁴ The mega-battery will ensure a stable power supply for the data centre and also offer dispatchable services to the grid, such as feeding electricity back in during peak periods or emergencies.³⁵ Stage one is under construction with the final stage (stage four) due to break ground in 2027.³⁶

By firming up renewable energy capacity and acting as a safeguard against outages, the Supernode battery is designed to reduce the grid impact of the new data centre.³⁷ This project reflects a more proactive model: instead of just plugging data centres into the existing grid and hoping it holds, operators are investing in on-site resources to help balance supply and demand.³⁸

Balancing the power demands of AI data centres

As artificial intelligence reshapes the digital economy, AI data centres are becoming some of the most demanding elements of modern electricity networks. Their intense and unpredictable power use is placing new pressure on infrastructure, degrading power quality, and affecting household appliances. While often invisible to consumers, these impacts are prompting growing concern.

Lessons from Ireland and Queensland demonstrate that, with proper planning, data centres can support rather than strain the grid. By investing in local energy storage, renewable generation, and improved coordination with grid operators, AI data centres can enhance – rather than undermine – system resilience.

---

¹Austrade, ‘AI and Data Centres’ (Web Page) <https://international.austrade.gov.au/en/do-business-with-australia/sectors/technology/ai-and-data‑centres>.

²IBM, ‘What Is an AI Data Center?’ (Web Page, 21 February 2025) <https://www.ibm.com/think/topics/ai-data-center>.

³McKinsey & Company, ‘AI power: Expanding data center capacity to meet growing demand’ (Web Page, 30 October 2024) <https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/ai-power-expanding-data-center-capacity-to-meet-growing-demand>.

⁴Digital Realty, ‘How AI Is Changing the Data Center Landscape’ (Web Page, 12 March 2024) <https://www.digitalrealty.com/resources/articles/data-center-ai>.

⁵What is an AI Data Center? (n 2).

⁶Ibid.

⁷All‑About‑Industries, ‘AI Data Centers Strain the Quality of the Power Grid’ (Web Page, 14 January 2025) <https://www.all-about-industries.com/ai-data-centers-strain-the-quality-of-the-power-grid-a-34f3ce06f4ce3ea4c02f4ff04cb3ebad/>.

⁸Ibid.

⁹Australian Energy Market Operator, AEMO’s Updated Forecasting Methodology Targets Rapidly Growing Electricity Loads (Web Page, 26 July 2024) <https://www.aemo.com.au/newsroom/news-updates/aemos-updated-forecasting-methodology-targets-rapidly-growing-electricity-loads>.

¹⁰Wendover Productions, “How AI Is Ruining the Electric Grid” (YouTube, Nebula, 2025) <https://www.youtube.com/watch?v=3__HO-akNC8\>.

¹¹What is an AI Data Center? (n 2).

¹²Bloomberg, ‘AI Needs So Much Power, It’s Making Yours Worse’ (Web Page, 27 December 2024) <https://www.bloomberg.com/graphics/2024-ai-power-home-appliances/>.

¹³Ibid; see also Data Center Dynamics, ‘AI Data Centres Causing “Distortions” in US Power Grid – Bloomberg’ (Web Page, 6 January 2025) <https://www.datacenterdynamics.com/en/news/ai-data-centers-causing-distortions-in-us-power-grid-bloomberg/>.

¹⁴Ibid.

¹⁵Tom’s Hardware, ‘AI data centers reportedly cause power problems in residential areas – decreased power quality in homes near data centers causes reduced lifespan for electrical appliances’ (Web Page, 28 December 2024) <https://www.tomshardware.com/tech-industry/artificial-intelligence/ai-data-centers-reportedly-cause-power-problems-in-residential-areas-decreased-power-quality-in-homes-near-data-centers-causes-reduced-lifespan-for-electrical-appliances>.

¹⁶Lightyear, ‘Ashburn colocation: The Data Centre Alley’ (Web Page, 12 November 2021) <https://lightyear.ai/blogs/ashburn-colocation-data-center-alley>.

¹⁷Goldman Sachs Research, AI to Drive 165% Increase in Data Center Power Demand by 2030 (Web Page, 4 February 2025) <https://www.goldmansachs.com/insights/articles/ai-to-drive-165-increase-in-data-center-power-demand-by-2030>.

¹⁸Grid Status, ‘Byte Blackouts: How large data center loads are surfacing new issues’ (Web Page, 18 April 2025) <https://blog.gridstatus.io/byte-blackouts-large-data-center-loads-new-issues-pjm/>.

¹⁹Ibid.

²⁰Ibid.

²¹The Washington Post, ‘Data-centre protest sparks community resistance’ (Web Page, 5 October 2024) <https://www.washingtonpost.com/technology/2024/10/05/data-center-protest-community-resistance/>.

²²Data Center Knowledge, ‘Why communities are protesting data centers – and how the industry can respond’ (Web Page, 13 June 2024) <https://www.datacenterknowledge.com/data-center-construction/why-communities-are-protesting-data-centers-and-how-the-industry-can-respond>.

²³Hydrogen Fuel News, ‘AI Data Centers Are Placing a Serious Strain on the Power Grid’ (Web Page,22 March 2024) <https://www.hydrogenfuelnews.com/ai-data-centers-power-grid/8568880/>.

²⁴Steve Hanley, ‘Demand for Power at Data Centers is Disrupting Harmonics in Nearby Electrical Grids’ (Web Page, 30 December 2024) <https://cleantechnica.com/2024/12/30/demand-for-power-at-data‑centers‑is‑disrupting‑harmonics‑in‑nearby‑electrical‑grids/>.

²⁵Ibid.

²⁶Ibid.

²⁷Microsoft, “Microsoft datacenter batteries to support growth of renewables on the power grid” (Web Page, 7 July 2022) <https://news.microsoft.com/source/features/sustainability/ireland-wind-farm-datacenter-ups/>; Associated Press, ‘Ireland embraced the AI boom. Now its data centers are consuming too much of its energy’ (AP, 19 December 2024) <https://apnews.com/article/6c0d63cbda3df740cd9bf2829ad62058>.

²⁸Microsoft (n 27).

²⁹Ibid.

³⁰Ibid.

³¹Ibid.

³²Ibid.

³³Associated Press (n 27).

³⁴RenewEconomy, ‘Quinbrook seals record finance deal to build huge “Supernode” data centre battery’ (Web Page, 22 January 2025) <https://www.reneweconomy.com.au/quinbrook-seals-record-finance-deal-to-build-huge-supernode-data-centre-battery/>; Quinbrook Infrastructure Partners, ‘Supernode BESS’ (Web Page) <https://www.quinbrook.com/quinbrook-project/supernode-bess/>.

³⁵Ibid.

³⁶David Scaysbrook, ‘Powering Big Data’ (Panel Discussion, Australian Clean Energy Summit, 29 July 2025).

³⁷RenewEconomy (n 35).

³⁸Ibid.