CLF common divisor

AI Data Centers Drive U.S. Grid Expansion

AI computing infrastructure speculation has shifted from chips to power, and the hard bottleneck of grid equipment is an industry consensus. A $75 billion transmission expansion is underway.

Demand Side

U.S. electricity consumption: 4,097 TWh in 2024 → 4,283 TWh in 2026, with growth led by commercial and industrial sectors (primarily data centers). U.S. data center capacity is projected to reach 120 GW by 2030, about five times that of 2025. Tech giants have signed ratepayer protection commitments, locking in demand for power infrastructure.

Supply Side

765kV ultra-high voltage transmission grid construction. Texas, the Mid-Atlantic, and the Midwest have approved transmission expansion projects totaling approximately $75 billion. 765kV line length will expand from about 2,000 miles to 10,000 miles. The U.S. is entering its first major grid investment cycle since the 1970s. Supply and demand for high/ultra-high voltage equipment remain tight, with 765kV capacity scarce.

Bottleneck Side

Transformer lead times have stretched from the usual 4-6 months to 24-36 months. The U.S. faces a 30% supply gap for large transformers. Hardware (transformers, switchgear, steel) has replaced capital and software as the real bottleneck in AI data center construction.

CLF's Key Strategic Position: The sole U.S. grid material supplier, but supply and demand are severely imbalanced.

Through its AK Steel asset, CLF is currently the only company that can still produce grain-oriented electrical steel (GOES), the core raw material for transformers, domestically in the U.S.

GOES accounts for about 25% of transformer cost, and its loss performance directly determines transformer efficiency and key performance indicators. However, the supply-demand landscape is extremely severe:

• Huge Gap: In 2024, effective domestic U.S. GOES capacity is less than 280,000 tons, meeting only 45% of transformer demand.
• Supply Collapse: AK Steel's parent company CLF decided to permanently close three plants in 2020 due to long-term losses. Another U.S. producer, ATI, had already completely ceased GOES production as early as 2016.
• No Short-Term Fix: Currently, over 70% of U.S. transformers are over 25 years old. Demand from new energy grid integration and data centers is rising simultaneously. However, building a new domestic plant from scratch to production takes at least 3-5 years.

This means the more urgent the demand for grid expansion, the more prominent the GOES supply bottleneck becomes. AK Steel under CLF is the only domestic asset that can directly benefit.

Steel Main Business Under Pressure, Grid Benefits Not Yet Reflected

The boost in demand for GOES from grid expansion is not yet fully reflected in CLF's financials; current profits are dragged down by the steel main business.

• Severe full-year loss in 2025, with 2025 fiscal year EPS forecast at -$1.96. Although the Q1 2026 loss narrowed to -$0.42, it remains unprofitable, with a net loss of $1.2 billion over the past 12 months.
• High Leverage Risk: S&P downgraded its rating from BB- to B+, with the debt-to-EBITDA ratio soaring from 2.3x to 20.5x.
• GOES Not Yet Contributing Profits: AK Steel is the sole U.S. GOES producer, but it is currently positioned as a strategic asset to maintain capacity. The pricing power and profitability of this business segment are not yet fully recognized by the market.
• Product Mix Upgrade Potential: The company plans to shift towards higher-margin stainless and specialty steels. Coupled with infrastructure investment and new energy construction expansion, this could improve the product structure long-term, supporting higher gross margins and average selling prices.

The core of the current valuation game for CLF lies in whether it can achieve a profit reversal in 2026.

Overall, the market's pricing logic for CLF is betting on a return to profitability in 2026. Q4 2025 is the quarterly profit trough, and the macro backdrop will improve in 2026.

The grid business driven by GOES is not yet fully priced in. The pressure from the steel main business in the short-to-medium term is the main contradiction restricting stock price upside. But HDC has already hit a new high.

Solid Logic, But Catalysts Need Time.

The logic chain of AI data centers driving U.S. grid expansion is clear, and policies and funding are being implemented. This major trend will not reverse. CLF's strategic position as the sole U.S. GOES producer will become prominent in the future.

However, the reality that must be faced is:

• Short-term performance is still being dragged down by the steel main business; it takes time for grid benefits to flow through to the income statement.
• The capacity bottleneck is also its own bottleneck. The domestic U.S. GOES capacity gap is as high as 55%. CLF's existing capacity is far from meeting the surging demand. Whether it can seize this window to accelerate capacity expansion will be a key variable.
• Financial pressure from high leverage and low ratings limits CLF's flexibility when market conditions improve.

If one believes in the long-term logic of grid expansion, CLF is currently in a favorable strategic position, but patience is still required. This is a game of 2026 steel recovery + grid expansion, not the moment for the grid story to be immediately realized.