Midnight (NIGHT) Plunge

China's large language model (LLM) industry is entering a critical consolidation phase. In a concentrated wave of funding in May 2026, leading players Kimi, StepFun, and DeepSeek reportedly secured over $70 billion combined, signaling a dramatic capital rush towards the few remaining independent contenders. This frenzy masks an impending shakeout. The core dynamic has shifted from a pure technology race to a battle for survival and strategic positioning. LLM capabilities are rapidly commoditized; gaps between top models are narrowing. Consequently, investment logic has pivoted from betting on future potential to prioritizing cash flow, user access, and ecosystem integration. The economic model poses a fundamental challenge: while user growth previously meant profits, in the AI era, it drives soaring inference costs. Startups, lacking the cross-subsidy ability of tech giants like ByteDance or Tencent, face immense pressure to achieve financial sustainability. DeepSeek's open-source, high-performance, low-cost strategy has further compressed industry profit margins. Facing this reality, the top players are scrambling to lock in their status before the window closes. StepFun is accelerating its港股 IPO, embedding itself in hardware supply chains. Kimi is aggressively showcasing revenue growth (ARR doubling to $2 billion in a month) to prove viability. DeepSeek, with new state-backed investment, is solidifying its role as a strategic national asset. The parallel to China's previous AI "Four Dragons" is stark. The industry is witnessing extreme capital concentration at the top, while mid-tier companies face a funding winter. The narrative has evolved from "who can build the best model" to "who can survive." For independent LLM companies, securing a public listing or a definitive strategic identity is no longer about expansion—it's about securing the very right to exist in the impending era of industry clearance.

An 18-year-old American hacker, Dritan Kapllani Jr., has been exposed by on-chain investigator ZachXBT for allegedly masterminding a series of social engineering attacks that stole approximately $19 million from crypto users. The investigation began after a Discord voice call on April 23, 2026, where Dritan shared his screen to boast, revealing an Exodus wallet holding around $3.68 million. Tracing this address, ZachXBT linked it to a major March 14, 2026 theft of 185 BTC (worth ~$13 million at the time), with about $5.3 million of that sum funneled into Dritan's wallet. Further analysis revealed the wallet also contained over $5.85 million from multiple social engineering thefts dating back to 2025. In a May 11, 2026 unsealed criminal complaint against another individual, Trenton Johnson, a key co-conspirator marked "CC-1" is identified, with the on-chain community pointing to Dritan. While not formally charged yet, he is now named in the judicial narrative. Another individual, Meme coin KOL yelotree, faces charges for allegedly assisting in money laundering. Dritan, known for a lavish lifestyle on social media, was previously seen as having a "protagonist aura" within hacking circles, evading consequences as associates were apprehended. However, now that he has turned 18, he is facing legal accountability for his past actions.

NVIDIA is set to report its quarterly earnings after the U.S. market closes on Wednesday, May 20. This event is widely seen as a crucial test for the current AI-driven bull market. The semiconductor sector is exhibiting severe technical overbought conditions, with the Philadelphia Semiconductor Index (SOX) trading approximately 60% above its 200-day moving average—the most extreme deviation since the dot-com bubble peak of 1999/2000. Market sentiment is highly concentrated on a few AI-related stocks, raising concerns about overall market breadth. Analysts highlight a key contradiction: while fundamentals for AI and semiconductors remain strong, significant technical pressures are building. Option market activity reflects this tension. Positions are heavily skewed towards bullish calls, yet there is also notable hedging activity through put options on broad indices and sector ETFs, signaling preparation for potential downside volatility. An unusual pattern of rising stock prices alongside rising implied volatility further underscores the market's expectation for a major move. For NVIDIA specifically, the market's primary focus will be on its forward guidance for the next quarter, which is deemed more critical than the immediate earnings results. Despite a recent seven-day rally adding roughly $1.7 trillion in market cap, historical data shows NVIDIA's stock has often declined the day after its past five earnings reports. The outcome of this report is expected to have a significant ripple effect across the broader technology and semiconductor markets, given NVIDIA's pivotal role.

Global markets regained ground after a brief dip, driven by renewed buying in AI-related stocks. The MSCI World Index rose 0.1%, with Asian equities reversing early losses and the Nasdaq 100 futures narrowing declines. Marvell soared over 14% in after-hours trading. Meanwhile, Iran's suspension of talks with the US heightened Middle East tensions, impacting commodities. Brent crude hovered near $94 a barrel, while gold rose about 1%. Aluminum prices hit a four-year high above $3,775 per ton on supply concerns, and copper also gained. Bitcoin fell 1.4% to just over $70,000. Investors remain focused on AI-driven gains—evident in the Philadelphia Semiconductor Index's surge—but some warn of stretched valuations. Upcoming US economic data, including Friday's jobs report, will be closely watched for clues on Federal Reserve policy.

Anthropic has published an article titled "When AI builds itself," discussing the emerging concept of "recursive self-improvement," where AI begins to actively participate in designing, training, testing, and optimizing its own subsequent versions. The company presents internal data showing that by May 2026, over 80% of code merged into its codebase was written by Claude, its AI model. Claude's capabilities have expanded to handling complex, open-ended engineering tasks, achieving a 76% success rate in such areas, and even contributing to research processes, such as optimizing code performance and conducting AI safety experiments. Anthropic outlines an evolution from human-driven development to AI-assisted workflows, culminating in the current stage where AI agents can autonomously write, run, and delegate code. The company cautions that the path toward a "closed loop," where AI continuously improves itself, is becoming visible. It calls for coordinated global mechanisms to potentially slow or pause frontier AI development to allow safety research and societal structures to catch up. However, the timing of this warning coincides with Anthropic's preparations for an IPO, framing the narrative not just as a safety concern but also as a demonstration of Claude's advanced capabilities and its integral role in accelerating Anthropic's own R&D—creating a potential "flywheel" effect for competitive advantage. This contrasts with OpenAI's recent, more policy-oriented discussion of the same risks, highlighting the competitive dynamics in the AI industry as companies position themselves in both the technological and regulatory landscape.