AI Stocks Fall as Leaders Call for Slower Development

Key Takeaways

• Anthropic CEO Dario Amodei, OpenAI's Sam Altman, and xAI's Elon Musk called for slower frontier-AI development on September 14, triggering a major tech selloff.
• Nvidia fell 3.4% and the Philadelphia Semiconductor Index dropped 5.9% on September 14, while cybersecurity stocks surged.
• The world's five largest tech companies plan to invest over $1 trillion in AI across 2025 and 2026, per the Bank for International Settlements.
• CrowdStrike shares jumped 13.85% after record net-new ARR of $333 million (up 51% YoY) and revenue of $1.47 billion (up 26% YoY).
• Micron reported Q3 revenues of $41.46 billion (up 346% YoY), while AMD posted Q2 revenues of $11.54 billion (up 50.1% YoY) with Data Center growth of 107.3%.
• AMD is expanding partnerships with Anthropic and Microsoft for its Instinct GPUs.
• Google DeepMind launched Gemini 3.8 Live models, claiming better performance at lower cost than GPT-Live 1, Astra, and Grok Voice Think Fast 2.0.
• Evercore analysts warn frontier AI systems will make cyberattacks more scalable through autonomous agents, boosting demand for identity security tools.
• Pennsylvania Senator Art Haywood plans AI safety legislation requiring transparency reports and critical incident disclosures.
• A Canadian AI CEO warned that excessive fearmongering about AI could undermine Canada's opportunity to lead in the field.

AI CEOs Call for Slower Development, Triggering $700 Billion Investment Reassessment

Leaders from top AI companies including Anthropic's Dario Amodei, OpenAI's Sam Altman and xAI's Elon Musk urged the industry to slow advanced AI development. Their warnings caused major tech stocks to fall on September 14, with Nvidia dropping 3.4% and the Philadelphia Semiconductor Index falling 5.9%. The Bank for International Settlements warns the world's five largest tech companies plan to invest over $1 trillion in AI across 2025 and 2026. The debate over AI safety is now influencing corporate strategy and financial markets.

AI Hardware Stocks Fall as Investors Shift Toward Cybersecurity

AI hardware stocks sold off on September 14 while cybersecurity stocks surged, marking the biggest performance gap between software and semiconductor ETFs. The split came after Anthropic CEO Dario Amodei called for slower frontier-AI development, a view supported by Sam Altman and Elon Musk. President Trump rejected the idea of slowing US development during a call with Nvidia CEO Jensen Huang. Despite the selloff, demand signals remain strong, with Broadcom reporting $16.7 billion in AI semiconductor revenue and Microsoft spending $41 billion on capital expenditure.

The AI Trade is Splitting Between Chips and Software

The AI investment trade is dividing as chip stocks lose momentum while software companies recover. Semiconductor stocks like Nvidia and Broadcom have softened from recent highs, while Salesforce, ServiceNow and Workday have rebounded sharply. The split was fueled by calls from Anthropic CEO Dario Amodei and other AI leaders for slower development of advanced AI models. If development cycles slow, chipmakers may feel the impact first, while software companies that were once feared to be replaced by AI could benefit from their ability to monetize AI tools.

AI Stocks Sell Off as Research Firm Names Six Strong Buy Picks

AI stocks faced heavy selling pressure due to AI safety concerns and rising Treasury yields, but Zacks ranked six companies as Strong Buy picks including NVIDIA, TSMC, Arista Networks, Dell, Hewlett Packard Enterprise and Super Micro Computer. These firms showed strong earnings momentum, with NVIDIA's data center revenue up 117% and Dell's AI server backlog reaching a record $95 billion. BofA estimates major cloud providers will spend over $860 billion on capital expenditure this year, up 80% from last year. The Philadelphia Semiconductor Index fell 5.86% on September 14, while software stocks like ServiceNow and Workday rose.

Micron vs AMD: Comparing Two Leading AI Chip Stocks

Micron Technology and AMD are both benefiting from AI infrastructure spending but differ in growth and risk profiles. Micron reported third-quarter revenues of $41.46 billion, up 346% year over year, driven by strong data center and high-bandwidth memory demand. AMD posted second-quarter revenues of $11.54 billion, up 50.1% year over year, with its Data Center segment growing 107.3%. Micron faces risks from memory market cycles and heavy capital spending, while AMD is ramping up its Instinct GPUs and expanding partnerships with companies like Anthropic and Microsoft.

CrowdStrike stock jumps 13.85% on AI security demand

CrowdStrike shares rose 13.85% on September 14 after a sector-wide shift into cybersecurity driven by AI safety concerns. The company reported record net-new ARR of $333 million, up 51% year over year, and strong growth in Falcon Flex adoption. Revenue reached $1.47 billion, up 26% from last year. However, stretched valuations and an overbought RSI of 72 raise potential downside risks for investors.

TCL Electronics showcases AI home products at IFA 2026

TCL Electronics used IFA 2026 to unveil its AI Inspired Life concept, including X11L SQD Mini LED TVs, RayNeo GT Max AR glasses, and the NXTHOME smart home ecosystem. The company also showed AI-powered appliances and new S-Class Dolby Atmos soundbars. The product push aims to connect entertainment, appliances, and energy solutions into one platform. Investors are divided on the value of this hardware-heavy, AI-focused strategy.

Evercore says AI will grow cybersecurity threats

Evercore analysts said frontier AI systems will create bigger security challenges as autonomous agents make cyberattacks more scalable. AI removes the human constraint that has limited the scale of sophisticated attacks. Attackers could deploy thousands of semi-independent agents to find and exploit vulnerabilities. Evercore still sees cybersecurity as a durable spending area, with identity vendors positioned to benefit from governing access across users and AI agents.

Pennsylvania senator plans AI safety legislation

State Senator Art Haywood announced plans for Pennsylvania legislation on AI transparency and safety requirements. The announcement came as AI leaders including Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman warned about risks from advanced AI systems. The proposed legislation could require AI companies to report how they identify and mitigate risks, publish transparency reports, and disclose critical incidents. Senator Haywood is still seeking input from Pennsylvania residents and experts before introducing the bill.

Canadian AI CEO warns fearmongering could hurt progress

The CEO of a leading Canadian AI firm urged the public not to believe fearmongering about AI technology. He said Canada has a rare opportunity to lead in artificial intelligence but could lose that chance if excessive fear takes hold. The CEO called for balanced discussions about AI safety and growth. The full details of his comments were not included in the article excerpt.

Google launches Gemini 3.8 Live models with better performance at lower cost

Google DeepMind released two new conversational AI models called Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking. The models are designed to perform better than rivals like GPT-Live 1, Astra, and Grok Voice Think Fast 2.0. Google says the new models offer more accurate responses at a lower price. They are aimed at uses like customer service and technical support.

SAIL report highlights AI risks driving demand for real-time identity security

A SAIL report identifies AI risks and agentic identities as key drivers for advanced identity security solutions. The growing use of AI systems creates new security challenges that require real-time protection. Agentic identities, or AI agents acting on their own, increase the need for stronger verification. The report stresses the demand for tools that can handle these evolving threats.

Fraglingo AI model improves molecular design with fragment-based generation

Researchers introduced Fraglingo, a new AI model for molecular design. It combines fragment selection and attachment prediction in one continuous process. Unlike older methods, Fraglingo works in a latent space and can add new fragments without retraining. The model supports molecule generation, scaffold decoration, and optimization. It also generalizes to fragment libraries up to 4 times larger than those used during training.

Sources

• [Insight] AI CEOs Hit the Brakes, Shaking a $700 Billion Investment Cycle
• AI Hardware Stocks Sell Off as AI Safety Shifts the Trade Toward Cybersecurity
• The AI trade is splitting — is money moving from chips into software?
• AI Stocks Hit by Panic Sell-Off; Research Firm Names 6 "Strong Buy" Picks Including TSMC and NVIDIA
• Micron vs. AMD: Which AI Chip Stock Is the Better Investment Bet?
• CrowdStrike Stock Breaks $233.88 as AI-Security Revaluation Puts $239.46 in Focus
• How Investors Are Reacting To TCL Electronics Holdings (SEHK:1070) AI Home Product Launches
• Evercore sees AI as growing cybersecurity threat
• As AI Leaders Warn of Alarming Risks, Senator Haywood Announces Plans for Pennsylvania AI Safety Legislation
• 'Fearmongering' could harm Canada’s rare opportunity in AI, says top CEO
• Google Releases Gemini 3.8 Live-Extended Conversational Model, Claims Better Performance Than Rivals At Lower Price
• SAIL: AI risks and agentic identities drive demand for advanced, real-time identity security solutions
• Fraglingo: Molecular Design via Attachment-Aware Autoregressive Fragment Generation