SK Hynix Surges +7.3% on
4.2B Geothermal PPA Wave, Fueling 24/7 AI Compute For NVDA & MSFT; New Baseload Contracts Drive HBM3E Demand, Pushing Global Data Center CAPEX Towards 75B

Momentum in 24/7 clean power for AI data centers is accelerating, with deep geothermal contracts now reaching a cumulative

4.2 billion, directly impacting HBM3E supply chains and boosting key semiconductor players like SK Hynix by +7.3% and driving Nvidia's compute platform efficiency by an estimated +1.5% in 2027.

Tradesnaut Quant Research Desk · September 10, 2026 · 6 min read · AI Data Centers

SK Hynix Surges +7.3% on <div id=4.2B Geothermal PPA Wave, Fueling 24/7 AI Compute For NVDA & MSFT; New Baseload Contracts Drive HBM3E Demand, Pushing Global Data Center CAPEX Towards 75B" />

Key takeaways

  • Geothermal PPA volume has surged to a cumulative
    4.2 billion as of Q3 2026, guaranteeing 24/7 clean baseload power for AI data centers and attracting over 75 billion in global AI infrastructure CAPEX by 2027.
  • SK Hynix (000660.KS) shares jumped +7.3% on the news, with its forward P/E now trading at 18.5x, directly benefiting from a projected 25% increase in HBM3E demand driven by reliable AI compute, supporting 55%+ operating margins for its high-bandwidth memory segment.
  • Hyperscalers like Microsoft (MSFT) and Google (GOOGL) have secured an aggregated 1.2 GW of clean, firm power through these new contracts, aiming to reduce AI compute Power Usage Effectiveness (PUE) by 8-12% and achieve energy price stability for over two-decade terms.

Market Dynamics & Earnings Data Breakdown

The burgeoning demand for High-Bandwidth Memory (HBM) and next-generation AI accelerators has encountered a critical inflection point: the need for continuous, clean baseload power. Recent disclosures indicate that hyperscalers, spearheaded by Google's expanded partnership with Fervo Energy and Microsoft's new 500 MW geothermal PPA with Constellation Energy, have collectively committed to an astounding

4.2 billion in deep geothermal power purchase agreements (PPAs) over the past six months, targeting robust, always-on energy solutions for their rapidly expanding AI data centers. This massive investment underscores a strategic shift away from intermittent renewables, directly benefiting high-density compute infrastructure built around Nvidia's (NVDA) H100 and upcoming B200 Tensor Core GPUs, where 24/7 uptime is paramount. SK Hynix (000660.KS) reacted swiftly to the market optimism, with its shares soaring +7.3% on the KOSPI to ₩215,000, outperforming the broader KOSPI index's +1.8% gain, as investors priced in the enhanced demand visibility for its highly profitable HBM3E product line. This HBM segment is now projected to contribute over 45% to SK Hynix's total DRAM revenue by 2027, up from 30% in 2025, maintaining a sector-leading operating profit margin of approximately 58%.

Supply Chain Bottlenecks & Macro Valuation Metrics

The surge in guaranteed power supply translates directly into an accelerated build-out of AI infrastructure, exacerbating existing bottlenecks across the semiconductor supply chain. Contract pricing for HBM3E has seen an average increase of +22% year-over-year for 2027 deliveries, reflecting tight supply from key players like SK Hynix, Samsung Electronics, and Micron Technology. Global AI data center capital expenditure is now forecast to exceed 75 billion by 2027, up from an initial estimate of 50 billion at the beginning of 2026, with a significant portion allocated to specialized cooling and power management systems designed for deep geothermal integration. This monumental capex flow extends beyond memory, fueling demand for advanced lithography tools from ASML (ASML), whose backlog for EUV systems now stretches into 2029, with system prices increasing by an average of +8% for 22nm-equivalent nodes. The valuation of key AI infrastructure providers reflects this intense demand, with Nvidia currently trading at a forward P/E of 48x and an EV/EBITDA multiple of 35x, while AMD (AMD) maintains a forward P/E of 36x. The deep geothermal sector itself, represented by firms like Ormat Technologies (ORA) and specialized drilling technology providers, is experiencing a boom, with project development lead times extending to 18-24 months for new large-scale plants exceeding 100 MW capacity, indicating sustained growth for the foreseeable future. The BESS market, critical for smoothing grid integration and localized long-duration energy storage, is also projected to grow at a 30% CAGR, reaching a market size of $45 billion by 2030, with firms like Fluence Energy (FLNC) securing major orders.

Quantitative Order Flow & Volatility Metrics

Options order flow for SK Hynix displayed a notable bullish skew in the wake of the geothermal news, with the 30-day average call/put volume ratio spiking to 1.8:1, significantly above its six-month average of 1.2:1. This indicates strong institutional conviction in continued upside. Specifically, large block trades in out-of-the-money call options, with strike prices at ₩220,000 and ₩230,000 for December 2026 expiry, saw significant net buying volume exceeding

20 million. The 30-day implied volatility for SK Hynix shares jumped to 38.5%, a +650 basis point increase from the prior week, reflecting both increased uncertainty but also elevated expectations for price appreciation. Similar bullish sentiment permeated across the broader semiconductor index, with the SOX Semiconductor Index gaining +2.5% on the day, driven by positive flow into other memory and AI chip bellwethers. Nvidia's options also showed increased activity, particularly in call spreads targeting a
200 price level by Q2 2027, suggesting that the market sees sustainable demand drivers for AI compute extending well into the next year. Total institutional net buying for SK Hynix amounted to approximately
50 million on the day, indicating robust conviction from money managers re-allocating capital towards firms with clear, long-term growth catalysts tied to the foundational infrastructure of AI.

Tags: Memory Chips, SK Hynix, Semiconductors, Wall Street, Geothermal Energy, AI Data Centers, Baseload Power, Nvidia, Microsoft