Micron Plunges 5.25%, HBM Super Cycle Suddenly Questioned: Will $900 Be Lost?

Опубліковано о 2026-09-15Востаннє оновлено о 2026-09-15

Анотація

AI Safety Controversy Sparks Global Chip Stock Sell-Off; Micron Falls 5.25%, SK Hynix and Samsung Electronics Also Show Significant Weakness. Memory Manufacturers' Profits Heavily Depend on HBM Shortage and Price Hikes. Once AI Capital Expenditure Slows, Profit Sensitivity Is Higher Than Some Equipment and IP Companies.

Slowing AI Development Calls Impact Chip Stocks

AI-related chip stocks fell sharply on Monday as the market digested Anthropic CEO Dario Amodei's suggestion to slow the pace of frontier model development. Following resignations of safety and alignment staff at multiple AI labs, the debate over whether safety measures can keep pace with model capabilities has rapidly intensified.

Similar views have resonated with other tech industry leaders, leading investors to worry that if AI model development is actively slowed, data center expansion and high-end chip orders could be affected. This triggered concentrated profit-taking in what was already a crowded AI trade.

Chip Stocks Bear the Brunt of the Sell-Off

Micron fell as much as 6.83% in early trading to $908.65. Intel opened down about 8%, SK Hynix fell 7.38%, Nvidia dropped 4.18%, AMD declined 5.75%, and SanDisk fell 7.45%, indicating a concentrated reduction in risk exposure across the chip supply chain.

This round of declines was triggered by the debate over AI development pace and quickly spread to GPU, CPU, memory, and storage companies. Companies like Micron and SK Hynix benefit significantly from AI server demand, so investors are primarily concerned that a slowdown in computing power build-out could affect HBM orders.

Slowing the Pace Does Not Mean Halting Development

OpenAI CEO Sam Altman supports establishing a unified federal AI safety framework requiring safety measures to keep pace with improvements in model capabilities. He also emphasized that the industry should not interpret calls to control development pace as stopping AI progress, and companies should proactively implement safety measures before legislation.

The market is now reassessing the real impact of safety constraints on chip demand. If the industry merely unifies and raises safety standards, Micron's long-term demand thesis may not change; however, if frontier model development and data center construction slow substantially, HBM and other high-end memory orders will face more direct pressure.

Пов'язані матеріали

Claude Forms Teams to Hunt Bugs! 66 Bugs Found in 116K Lines of Code, Single Agent Maxed Out at 27

Claude Managed Agents, Anthropic's new multi-agent workflow system, significantly outperformed a single AI agent in finding bugs. In a test with 70 bugs hidden across 116,000 lines of code, a single Claude agent found a maximum of 27 bugs. When Claude was tasked with organizing a team of agents itself via the dynamic workflow, it consistently found 66 bugs across three runs. This capability is part of two major updates: "Claude Managed Agents" for developers to integrate multi-agent coordination into their apps, and an expanded beta for "Claude Code Projects," which lets users manage complex tasks in a single conversation, with Claude breaking down work and coordinating parallel threads. The Managed Agents feature allows a primary Claude agent to receive a task, automatically write a workflow program to divide labor among specialized sub-agents, and coordinate their execution and results aggregation. For example, it can orchestrate checking 300 contracts for specific clauses, using multiple agents for parallel reading and verification. The system supports workflows with up to 1000 cumulative agents per run. While this multi-agent approach improves task coverage and results, Anthropic notes that costs and efficiency gains need careful evaluation. Projects use existing plan credits, while Managed Agents incur token usage fees plus a $0.08 per hour runtime fee per active session. The key challenge is whether offloading coordination to Claude leads to faster, more reliable outcomes with less human oversight.

marsbit2 год тому

Claude Forms Teams to Hunt Bugs! 66 Bugs Found in 116K Lines of Code, Single Agent Maxed Out at 27

marsbit2 год тому

Terence Tao's Blog Explodes on Hot Search! 250,000 Math PhDs on the Brink, Must Slow Down AI

Mathematician Terence Tao's blog post ignited a fierce online debate after a video of him warning "we must slow down AI" went viral. He described AI's progress as an "extremely nonlinear dynamic" pursued with unclear consequences. This warning came just days after he promoted a project for open-source AI mathematical models, highlighting a profound tension. The core crisis, explored in a guest post by 3Blue1Brown creator Grant Sanderson, is AI's threat to traditional mathematics. Automated theorem provers combined with large language models can generate proofs at superhuman speeds, potentially rendering the human role of "proving theorems" obsolete. Sanderson argues that the true goal of mathematics is not proof-generation but the advancement of *human understanding*. As AI masters the former, the academic value of proof alone is diminished. Sanderson proposes a paradigm shift: elevating "motivational explanation" over mere proof. Unlike a formal proof, a motivational explanation guides readers through the intuitive reasoning, historical context, and conceptual connections behind a result—the "why" and "how" of discovery. To institutionalize this, he suggests creating "open explanation problems," reforming PhD defenses to prioritize clear exposition, and valuing pedagogical work as highly as journal publications in tenure reviews. The debate drew reactions across academia. Some physicists welcomed mathematicians to a world long prioritizing understanding over rigorous proof. Critics worried this shift could disadvantage brilliant but inarticulate problem-solvers. Others questioned if AI will eventually master explanation too. Supporters countered that the goal is fostering *human* comprehension, which is inherently subjective and interactive. The article concludes that AI's rise presents both a crisis and an opportunity. Without adapting academic culture and reward systems, human mathematicians risk irrelevance. However, by embracing their unique role as architects of understanding and interpreters of AI-generated knowledge, they can steer mathematics toward a new renaissance focused on human insight and clarity.

marsbit5 год тому

Terence Tao's Blog Explodes on Hot Search! 250,000 Math PhDs on the Brink, Must Slow Down AI

marsbit5 год тому

The "Number Game" Behind Trillion-Dollar Valuations: OpenAI and Anthropic's Revenue Can't Be Compared Directly

"Valuation Game in Trillions: The 'Digital Metrics Play' Behind OpenAI and Anthropic's Incomparable Revenues" Investors face significant uncertainty as core financial metrics from the two AI giants, OpenAI and Anthropic, cannot be compared directly due to differing accounting methods. OpenAI projects its annualized revenue run rate (ARR) to reach or exceed $70 billion by year-end, while Anthropic announced in July that its ARR had hit $65 billion. However, a key discrepancy renders these figures incomparable: Anthropic records the total sales value generated through cloud partners like Amazon as revenue, whereas OpenAI only counts its net share from partners like Microsoft. This reporting difference has tangible market impact. A recent tech stock downturn was partly attributed to investor attempts to adjust OpenAI's numbers for a direct comparison, leading to lower estimates and concerns about its growth pace. Currently, OpenAI's ARR is reportedly around $50 billion, with expectations to approach $70 billion by December. The issue highlights broader concerns about ARR as a non-standardized, often manipulated metric used to showcase growth and justify high private valuations. It is calculated by extrapolating short-term performance or summing contract values, offering companies considerable discretion. A significant gap exists between ARR and actual recognized annual revenue: OpenAI expects 2026 revenue of roughly $35 billion (half its year-end ARR projection), and internal documents show Anthropic's 2025 revenue was about $4.6 billion against a published ARR exceeding $9 billion. Experts warn that the opacity surrounding these key players creates a "black box" effect, complicating investment decisions and amplifying market volatility. As their performance is a bellwether for the AI investment narrative, any signs of slowing sales growth could ripple through the broader market, affecting chipmakers and other beneficiaries of the AI boom. The reliance on ARR sidesteps a fundamental question: the scale of real sales needed to justify trillion-dollar valuations and cover massive expenditures. The situation may improve with greater transparency, particularly if Anthropic proceeds with its anticipated Wall Street debut, which would subject the company to standardized GAAP financial reporting.

marsbit7 год тому

The "Number Game" Behind Trillion-Dollar Valuations: OpenAI and Anthropic's Revenue Can't Be Compared Directly

marsbit7 год тому

Торгівля

Спот
活动图片