Wednesday, July 22, 2026
SuMarket
Market Intelligence, Daily
Saturday, July 11, 2026

Market Overview

bullishSnapshot

AI chip makers and crypto companies are raising historic sums while the market shifts toward cheaper, smarter AI systems—but oil demand is slowing and geopolitical risk in the Middle East is spiking.

SK Hynix raises $26.5B in biggest foreign IPO ever

South Korea's SK Hynix priced its U.S. listing at $149 per share on Friday, raising $26.5 billion—the largest IPO by a foreign company in history and second only to SpaceX among all U.S. IPOs (CNBC, Nasdaq). The stock soared 13% on its first day, closing at $168.01, as investors scrambled for exposure to the booming AI memory market. SK Hynix is the leader in high-bandwidth memory (HBM)—the specialized chips that feed data to AI processors—holding 58% of HBM revenue. The company's chairman told CNBC demand is 'enormous' and 'exponential,' with customers saying even a planned doubling of capacity over five years won't be enough. Proceeds will fund new fabrication plants in South Korea and advanced packaging in Indiana, cementing the race to build AI infrastructure before supply runs out.

CNBC, Nasdaq
Circle gets federal bank charter for stablecoin reserves

Stablecoin issuer Circle received approval Friday from the U.S. Office of the Comptroller of the Currency to operate as a trust bank, sending its shares up 5% (CNBC). The new entity, Circle National Trust, will let the company directly manage the cash and Treasury assets backing its USDC stablecoin, which has over $73 billion in circulation—previously, Circle relied on third-party custodians. The charter doesn't allow Circle to take deposits or make loans, but it gives the company a single federal rulebook instead of juggling 50 different state regulations. This is a major signal: crypto infrastructure is shifting from experimental fintech to regulated financial plumbing, joining recent approvals for Coinbase, BitGo, and others. The move also reflects broader competition: banks including Citi and HSBC just launched a blockchain consortium on the same day, and over 140 companies joined the Open USD stablecoin effort to capture payment flows.

CNBC
AI race pivots from biggest models to task-specific, cheaper systems

The artificial intelligence competition is fundamentally reshaping: instead of chasing the biggest, smartest models, companies are now choosing AI by task, cost, and control (CNBC). As enterprises move from experimentation to production, the product is no longer the model alone but the 'harness'—a routing system that picks the right AI tool for the job. A routine customer service task might use a cheap open-weight model; a complex coding problem gets escalated to a premium frontier model. Perplexity CEO Aravind Srinivas previewed a system using GLM 5.2, a cheaper open model from China's Z.ai, for lighter work. Benchmark GP Peter Fenton argued that over 90% of AI tokens will come from open-weight models within 18 months as companies discover smaller, tuned models often outperform larger general-purpose ones. Ollama, which lets firms run open models locally, has been adopted by 85% of the Fortune 500. The shift threatens OpenAI and Anthropic's pricing power while creating a strategic challenge: many competitive open models come from China, raising U.S. competitiveness concerns.

CNBC
Key takeaway: AI infrastructure is real and hungry, but the competition is shifting from raw model size to efficiency and cost—meaning the incumbents' pricing power is under siege while oil geopolitics threatens to destabilize everything.
Sign in for the full snapshot briefing — every story, every day.
Read free on SuMarket →