Maryland residents forced to shoulder $2 billion in grid upgrades: Cost pass-through from AI data centers

Core conclusion

In May 2026, U.S. federal energy regulators approved a decision that shocked Maryland residents: The state’s roughly 2 million electricity customers will be forced to shoulder the cost of up to $2 billion in grid upgrades, primarily to support out-of-state AI data centers. The ruling broke through the state's previous "user protection commitments."

Key Points

• When: Early May 2026, Federal Energy Regulatory Commission (FERC) approval
• Amount: $2 billion in grid upgrade costs spread over approximately 2 million customers
• Beneficiary: Out-of-state AI data center (non-Maryland local users)
• Core of Controversy: Who should pay for AI infrastructure? Is cost passing fair?

Background: Energy dilemma under the rapid rise of AI

The energy requirements for AI model training and inference are growing exponentially. The electricity consumption of a large AI data center is equivalent to the entire electricity consumption of a medium-sized city. Grid Strategies reports that total U.S. data center power demand is expected to grow from 21 GW in 2023 to 74 GW in 2030, an increase of more than 250%.

To meet these demands, grid operator PJM Interconnection plans to upgrade transmission infrastructure across the United States. Users in Maryland were allocated the lion's share of the cost of this upgrade - even though these data centers are mainly located in out-of-state states such as Virginia.

Key impact table

The truth about the cost of AI: Who’s paying?

This Hacker News trending post sparked a lot of discussion. Community users generally believe that the operating costs of AI giants are being socialized—that is, profits belong to the company and the costs are borne by the public.

Several points worth paying attention to:

• Cost passing is nothing new: Historically, grid upgrade costs have been passed on to ordinary users many times. But the scale of the AI ​​era is unprecedented.
• AI manufacturers should build their own energy: Comments suggest that AI companies should invest resources in new energy technologies such as nuclear fusion, rather than relying on existing user subsidies.
• Independent Grid in Texas: Some comments noted that the Texas ERCOT independent grid model could avoid such interstate cost shifting.

Impact on independent developers

What does this mean for WayToClawEarn’s readers—indie developers, content creators, and small teams?

1. The rising trend of cloud AI API costs is established: When power grid costs are passed on to data center operators, they will eventually be transmitted to end users through API price increases. OpenAI GPT-5.5 has recently increased in price by 100%, with energy costs being one of the important driving factors.
2. Local model competitiveness increases: Locally runnable models represented by DeepSeek V4, Gemma 4, and Qwen 3.6 are maturing rapidly. As cloud API prices continue to rise, the economics of the "local first" strategy are becoming more and more cost-effective.
3. Infrastructure decisions require a longer-term perspective: Choose cloud services or build in-house? Rely on API or local deployment? Cost expectations need to be calculated on an annual basis, not just on a monthly basis.

Adaptation suggestions

• Evaluate local model solutions now: Test DeepSeek V4 Flash, Gemma 4 and other models that can run on consumer-grade hardware
• Lock in existing API subscription price: If currently using OpenAI/Anthropic API, consider pre-purchasing the annual plan to lock in price
• Focus on energy-friendly AI tools: Choose models with higher reasoning efficiency (such as MoE architecture) to reduce energy consumption per unit token
• Strategic use of hybrid solutions: local models for simple tasks, cloud APIs for complex inferences

Tool entry

Tool brands mentioned in the text: OpenAI, ChatGPT, DeepSeek, Gemma, Gemini, Claude

Related reading

• Hacker News — In-depth community discussion
• Tom's Hardware

Internal link guidance

• Want to learn how to reduce costs with local models? See: Claude Code + DeepSeek V4 Building Tutorial: API Fees Dropped by 90% (15 Minutes)
• Real case: AI automation helps you save labor costs and cloud expenses: Indie Developer: n8n + OpenClaw Automation Workflow Earning $5,000/mo
• Want to know more? See: DeepSeek V4 vs Claude Code: 90% Cheaper, Same Quality