Wall Street and Asian markets remained turbulent this week as investor anxiety over artificial intelligence spending weighed heavily on tech stocks. The S&P 500 fell 0.5% and the Nasdaq dropped 1.1% in one session, while the Dow rose 0.4%. Oil prices swung sharply, falling 7.6% to $104.77 a barrel before rebounding 2% to $106.43 the next day. South Korea's Kospi plunged 3.59% over AI spending fears, with Samsung and SK Hynix each losing about 4% of their value.
Major tech companies showed mixed results. Nvidia gained 2.3% and Broadcom rose 3.1% in a rally that lifted the S&P 500 by 0.4%. Microsoft, Amazon, and Meta each climbed over 2%. Apple, however, fell 1.9% after warning that profits would suffer from costly memory chips needed for AI data centers. The company's earnings report contributed to the jittery mood across Asian markets as well.
Google introduced its Gemini 4 Carbon AI model, which scored 59 on the Artificial Analysis Intelligence Index, narrowly beating Anthropic's Claude Opus 5.5 at 55. Gemini 4 Carbon also scored 54 on the AA-Omniscience test for accuracy. The model costs roughly $16 per task, far less than Claude at about $51 or GPT-6 Astra at about $54. Google says the model reduces unsupported claims twice as effectively as earlier versions.
Larridin raised $30 million in a Series A funding round led by Pelion Ventures, bringing its total funding to $47 million. The company tracks AI spending against actual productivity so businesses can measure real returns. In one case, a single engineer consumed 65% of an entire company AI budget. Customers include SurveyMonkey, Vertiv, Kalviyo, and Gainsight.
Concerns about AI security deepened after an Accenture contractor connected a personal Claude Code agent to a Defense Department system on NIPRNet for roughly two months. The agent held administrative access and may have transmitted credentials and a network map to an outside site. Defense Secretary Pete Hegseth had already designated Anthropic a supply-chain risk in March. Accenture, which holds over $3.5 billion in Pentagon contracts, called the allegations without merit.
Palantir CEO Alex Karp used the company's third-quarter earnings call to warn that China is excelling at AI development. He argued that American technological leadership is critical and that U.S. dominance in AI could help prevent conflicts. Separately, a Washington Post experiment found that six popular AI chatbots provided unreliable voter guides, inventing nonexistent races and mixing up candidates. Two in five Americans reported using AI for election research.
Anthropic is building an open source verification tool called FlashProve that can confirm which AI model generated a specific output. The system uses published model fingerprints that do not reveal internal weights or numbers, and it is light enough to run on local devices. Meanwhile, ICON Public Limited Company attracted investment from Riverwater Partners, which praised the company for using AI effectively to speed up clinical trial operations for drug developers.
Key Takeaways
- <ul><li>Wall Street closed mixed as tech and AI stocks slipped, with the S&P 500 down 0.5% and the Nasdaq down 1.1% amid profit concerns over massive AI spending.</li><li>Oil prices swung sharply, falling 7.6% to $104.77 then rebounding 2% to $106.43, creating volatility across energy-sensitive sectors.</li><li>Google launched Gemini 4 Carbon, scoring 59 on the Artificial Analysis Intelligence Index and costing about $16 per task versus Claude Opus 5.5 at roughly $51.</li><li>Larridin raised $30 million in Series A funding to help companies track AI spending against actual productivity returns, with total funding now at $47 million.</li><li>An Accenture contractor connected Claude AI to a Pentagon network for two months, raising security concerns after Anthropic was flagged as a supply-chain risk.</li><li>Palantir CEO Alex Karp warned that China is excelling at AI and stressed that U.S. technological leadership is critical for global stability.</li><li>A Washington Post experiment found AI chatbots provided inaccurate voter guides, inventing nonexistent races and confusing candidates across U.S. counties.</li><li>Anthropic is developing FlashProve, an open source tool to verify which AI model produced a given output using lightweight model fingerprints.</li><li>Apple fell 1.9% after warning that expensive memory chips for AI data centers would lower profits, contributing to broader market unease.</li><li>ICON PLC drew investment from Riverwater Partners for its effective use of AI in speeding up clinical trial operations for drug developers.</li></ul>
Wall Street ends mixed as AI and tech stocks keep slipping
Wall Street closed mixed on Monday as tech and AI stocks stayed weak. The S&P 500 fell 0.5% and the Nasdaq dropped 1.1%, while the Dow rose 0.4%. Oil prices fell 7.6% to $104.77 a barrel, helping airlines and cruise companies. Investors remain worried that huge AI spending may not earn enough profit.
Asian shares mixed as AI worries and oil rebound shake markets
Asian shares ended mixed on Tuesday as AI-related stocks stayed shaky. Oil prices rebounded 2% to $106.43 a barrel after falling sharply the day before. Apple reported mixed earnings, and South Korea's Kospi plunged 3.59% over AI spending fears. Australia's S&P/ASX 200 rose 0.34% after the Reserve Bank raised interest rates to 4.85%.
S&P 500 rises as Nvidia and AI stocks rally, Apple falls
The S&P 500 rose 0.4% to 7,568.92 as Nvidia gained 2.3% and Broadcom rose 3.1%. Microsoft, Amazon and Meta also climbed over 2%. Apple fell 1.9% after warning about lower profits due to expensive memory chips for AI data centers. The Dow dropped 197 points while the Nasdaq gained 207 points.
Stocks mostly fall as AI concerns outweigh sharp oil price drop
Stocks fell mostly on Tuesday as AI worries overshadowed a 7% drop in oil prices. Asian markets struggled after a tech selloff on Wall Street. South Korea's Kospi fell over 2% as Samsung and SK Hynix lost about 4% each. US and Iran officials expressed cautious optimism about peace talks, but positions have not changed much.
Google launches Gemini 4 Carbon AI model with strong benchmarks
Google revealed its Gemini 4 Carbon AI model, scoring 59 on the Artificial Analysis Intelligence Index, one point above Claude Opus 5.5. It scored 54 on the AA-Omniscience test for accuracy and low hallucinations. The model costs $7.76 per task and uses fewer tokens than competitors. Google claims it cuts unsupported claims twice as much as earlier Gemini models.
Gemini 4 Carbon costs less and nearly tops AI benchmark
Google's Gemini 4 Carbon scored just behind Anthropic's Claude Opus 5.5 in a 206-task benchmark covering fields like cybersecurity, finance, and biology. Claude cost about $51 per task, while Gemini 4 Carbon cost just over $16 per task. Gemini 4 Carbon also beat GPT-6 Astra, which scored lower and cost about $54 per task. Google's older model, Gemini 4 Argon, ranked eighth in the benchmark.
Larridin raises $30M to measure AI value for companies
Larridin raised $30 million in a Series A round led by Pelion Ventures, bringing its total funding to $47 million. The company built a platform that tracks AI spending alongside productivity and output quality so businesses can see real returns on AI investments. One customer found a single engineer used 65% of the company's entire AI budget. Larridin's customers include SurveyMonkey, Vertiv, Kalviyo, and Gainsight.
Accenture contractor ran banned Claude AI on Pentagon network
An Accenture contractor connected a personal Claude Code agent to a Defense Department system on NIPRNet for about two months. The agent held administrative access and may have sent credentials and a network map to an outside site flagged by NIST. Defense Secretary Pete Hegseth had already designated Anthropic a supply-chain risk in March. Accenture said it investigated the allegations and found them without merit. The Pentagon holds over $3.5 billion in federal contracts with Accenture.
Palantir CEO warns China is excelling at AI
Palantir CEO Alex Karp spoke during the company's third-quarter earnings call about the growing global AI race. He said China is doing exceedingly well at AI and argued that American technological leadership is critical. Karp also discussed Palantir's self-description as a pioneer of a viable anti-war model. He stressed that U.S. dominance in AI could help prevent wars.
AI chatbots can mislead voters doing last-minute research
A Washington Post experiment tested six popular AI chatbots on voter guides for every county in the United States. The chatbots invented nonexistent races, mixed up candidates, and cited incorrect websites. Two in five Americans said they used AI to look up election information. Experts recommend checking trusted sources instead of relying on chatbots. Several companies said they made changes after The Post contacted them.
ICON PLC uses AI to speed up clinical trials
Riverwater Partners started a position in ICON Public Limited Company (NASDAQ:ICLR) in its Q3 2026 investor letter. ICON is a global contract research organization that runs human clinical trials for drug developers. The investment firm praised management for using AI effectively in trial operations. The shares were trading below their ten-year average valuation.
Democratic midterm win may slow AI data centers
A Democratic sweep in the midterm elections could worsen recent sell-offs in AI-related stocks. More Democratic control at the state level could slow the buildout of AI data centers. This may create headwinds for hyperscalers, neoclouds, and suppliers that need more compute capacity. Increased regulation could also impact Big Tech, cryptocurrencies, and tokenized-stock trading.
Anthropic builds tool to verify AI model outputs
Anthropic is developing an open source system called FlashProve to verify which AI model produced a given output. The system works by having AI labs publish a fingerprint of each model that can be matched to specific outputs. These fingerprints do not reveal a model's weights or internal numbers. The system is lightweight enough for verification on a local device.
Economists answer questions about AI and jobs
Two economists answered audience questions about AI job displacement and development. Daron Acemoglu, a Nobel Prize-winning MIT professor, and Erik Brynjolfsson, director of the Stanford Digital Economy Lab, discussed the topic. The event was a live Q&A shared as a bonus episode of an NPR show. Questions covered AI's impact on employment, pro-human AI development, and competition with AI-created material.
Sources
- Wall Street rally fades as tech and AI stocks slip again
- Asian shares mixed as AI profit concerns weigh
- S&P 500 rises as Nvidia, AI stocks rally despite Apple decline
- Stocks mostly down as AI concerns top plunge in oil prices
- Google reveals new Gemini 4 Carbon AI model, narrowly beating Claude Opus 5.5 — frontier AI model is also the least likely to hallucinate, benchmarks claim
- Gemini 4 Carbon lags just behind Claude in benchmark, but wins in value
- VCs bet $30M on Larridin's ability to help enterprises demonstrate AI's value for money
- Accenture Contractor Ran Banned Claude AI Agent on Pentagon Network for Months
- 'It's Us or Them': Palantir CEO Alex Karp Says China Is 'Doing Exceedingly Well at AI,' Warns America Must Lead
- Doing last-minute research for voting? Be careful with AI chatbots.
- ICON PLC (ICLR) Is Leveraging AI to Accelerate Clinical Trials
- Blue wave in midterms could slow data centers, prompting further AI-related sell-offs
- Anthropic is working on a system to prove AI models' identity
- Will AI take my job? Two economists answer your questions (NPR+)
Comments
Please log in to post a comment.