SK Hynix Surges +14.7% on Record HBM3e Demand, Propelling NVDA Market Cap Beyond

Key takeaways

SK Hynix reported Q2 2026 operating profit soared 195% YoY to .95 billion, exceeding Street estimates by 18.2% on HBM3e revenue growth of 135% YoY to $4.7 billion.

Major hyperscalers like Microsoft, Amazon AWS, and Google Cloud collectively projected 2026 capital expenditures exceeding 80 billion, a +28% increase from 2025, primarily allocated to AI compute infrastructure.

The SOX Semiconductor Index recorded a new all-time high of 7,345, rising +1.8% intraday, driven by a 0.76 standard deviation shift in options call volume for top-tier AI suppliers.

Market Dynamics & Earnings Data Breakdown

The semiconductor sector, particularly high-bandwidth memory (HBM) and AI accelerator segments, witnessed extraordinary momentum following robust Q2 2026 earnings reports. SK Hynix led the charge, announcing an operating profit of .95 billion, a staggering 195% year-over-year increase, significantly beating consensus estimates of .34 billion. This impressive performance was underpinned by a 135% YoY surge in HBM revenue, reaching $4.7 billion, driven by insatiable demand for HBM3e from key AI accelerator manufacturers like Nvidia. Shares of SK Hynix (000660.KS) rallied +14.7% on the KOSPI, closing at ₩268,500, pushing its market capitalization to approximately

42 billion. Concurrently, Samsung Electronics (005930.KS) also saw its HBM division grow revenue by an estimated 95% YoY, though overall memory profit growth was a more modest 110% YoY, reflecting its broader product portfolio. Analyst revisions post-earnings showed average 2027 HBM revenue estimates for the sector increasing by 22.5%, highlighting a re-rating of future growth potential.

Nvidia (NVDA) continued its dominance, with its data center revenue projected to exceed

00 billion in FY2027, up from an estimated $70 billion in FY2026. The stock surged +3.2% to

,405.60, catapulting its market cap above

.5 trillion as investors factored in increased HBM supply and stronger-than-expected AI chip demand. Microsoft (MSFT), Amazon AWS (AMZN), and Google Cloud (GOOGL) all reaffirmed aggressive AI infrastructure build-out plans during their respective earnings calls, with combined Q2 2026 cloud segment revenue growing at an average of 26.8% YoY. These hyperscalers collectively reported Q2 2026 capex spending reaching $71 billion, a +31% increase from the prior year, underscoring the massive investment wave in AI that is directly translating into orders for Nvidia's GPUs and SK Hynix's HBM. The market's immediate, sub-second reaction to these data points was observed across algorithmic trading platforms, indicating rapid parsing of multi-modal information streams, from earnings call tonality to SEC 8-K disclosures.

Supply Chain Bottlenecks & Macro Valuation Metrics

The burgeoning demand for HBM has tightened supply considerably, pushing contract prices for Q3 2026 HBM3e memory up by an average of +22.5% quarter-over-quarter, with some spot market transactions even higher at +25% for immediate delivery. This premium pricing is a direct result of the complex manufacturing process and the limited number of suppliers capable of meeting stringent quality and performance requirements for advanced AI chips. Major memory manufacturers are investing heavily, with SK Hynix projecting 2026 capex of

4.2 billion, up from

0.5 billion in 2025, primarily dedicated to HBM expansion. Micron (MU) also announced a similar capex trajectory, aiming for

2 billion in 2026 to scale up its HBM production, albeit with a slight lag behind its Korean competitors. The overall semiconductor industry capex, excluding memory, is estimated to reach 50 billion+ in 2026, driven by foundry leaders like TSMC (TSM), which reported a 100% utilization rate for its N3 and N2 process nodes, with lead times for advanced orders extending to 18-24 months.

Valuation metrics for key AI beneficiaries reflect this aggressive growth outlook. SK Hynix is currently trading at a forward P/E of 19.8x based on revised 2027 earnings, representing a 35% premium compared to its 5-year historical average of 14.6x, yet still below Nvidia's forward P/E of 38.5x. The EV/EBITDA multiple for the top three HBM suppliers (SK Hynix, Samsung, Micron) now averages 12.4x, a significant expansion from 8.8x just 12 months prior. Hyperscalers’ unprecedented capex commitments – Microsoft, AWS, and Google planning a combined 80 billion+ in 2026 – underscore the structural shift in enterprise IT spend towards AI. This massive capital flow is corroborated by satellite imagery analysis showing accelerated construction and heat signatures consistent with expanding data center footprints globally, providing an early signal that the AI investment cycle is far from peaking. Operating profit margins for HBM segments are now comfortably in the 50-55% range for leading players, a significant uplift from traditional DRAM margins of 25-30%.

Quantitative Order Flow & Volatility Metrics

Market microstructure analysis reveals heightened activity across options markets, particularly for firms at the forefront of AI innovation. Nvidia (NVDA) experienced a substantial bullish skew in its options order flow, with the 30-day average call/put ratio reaching 1.85, significantly above its historical mean of 1.2. The volume of out-of-the-money (OTM) call options for NVDA with October 2026 expiry trading at strikes of

,500 and

,600 saw a 0.76 standard deviation increase in daily volume, indicating strong institutional conviction in continued upside. Implied volatility for NVDA remained elevated at 58.5%, reflecting the market's expectation of sustained price swings, but with a clear directional bias. The SOX Semiconductor Index futures climbed +1.8% to 7,345 following the HBM news, outperforming Nasdaq 100 futures which gained a more modest +0.9%, highlighting sector-specific strength.

For SK Hynix, significant net institutional buying was observed on the KOSPI, estimated at $850 million for the day, with foreign investors accumulating shares. Options data for SK Hynix on the Korea Exchange showed a notable increase in call option open interest for September 2026 expiry at ₩280,000 and ₩300,000 strikes, with volumes up 150% compared to its 20-day average. This aggressive positioning suggests institutional traders are leveraging real-time insights, likely derived from agentic AI parsing of supply chain data, competitor announcements, and even macroeconomic indicators like energy prices affecting Constellation Energy (CEG), a key power provider for some hyperscale data centers, whose stock also saw a +1.1% uptick. The implied volatility for the overall memory sector, as tracked by a custom basket of memory ETFs, rose by 85 basis points, but the equity delta remained strongly positive, indicating demand for upside protection and participation.

Tags: Memory Chips, SK Hynix, Semiconductors, Wall Street, AI, HBM3e, Nvidia, TSMC