A telecommunications outage at FAA air traffic control centers triggered a nationwide ground stop on Friday, with more than 3,100 aircraft bound for U.S. airports already in the air. Federal officials are investigating whether the FAA’s new cloud-based SMART traffic-planning system played any role, but have not established a cause. Homeland Security Secretary Markwayne Mullin said there was no threat to the public and that the outage did not appear connected to Microsoft 365’s service disruption; Microsoft also reported no indication of a link. SMART analyzes about 200 data streams, including weather, airport congestion and flight paths, and its recommendations are reviewed by FAA specialists before local facilities decide whether to use them.
The investigation comes as the FAA expands SMART across U.S. airspace. The $875 million machine-learning system advises controllers on routes, and the agency says it has helped reroute 180 flights and reduce delays since its initial rollout began in September. Airline executives support the expansion, while the air traffic controllers’ union says it was not involved in designing or testing the system. The FAA has not said when normal operations will resume after Friday’s disruption.
Several companies are investing in tools to build and operate AI systems. LangChain raised $300 million to expand its developer products: LangChain connects agents to language models and other tools, LangGraph coordinates agents in workflows, and Deep Agents adds planning and subagent features. Its open-source tools were downloaded about 90 million times a month after its previous funding round; LangSmith, used by Autodesk and Cisco, helps teams monitor and maintain agents. LangChain reportedly plans to nearly double its workforce to 850 by the end of 2027.
Mirendil raised $1 billion in Series A funding led by Kleiner Perkins and Clearlake Capital, with Google among the investors. Founded by former Google DeepMind and Anthropic researchers, it developed Lor, a platform that uses hundreds of AI agents to automate model-training tasks. Mirendil says one researcher built its Mirendil-Base coding model for $10 million in about 3.5 months, with agents writing more than 99% of the code. The model underpins Fern, unveiled last month, which can process 10 million input tokens per prompt—10 times more than Claude Opus 5.5. Mirendil plans to use larger versions of Mirendil-Base to develop more capable language models.
In research, Yale mathematician Alexander Petrov and students Cyrus Ghane and Ray Zhao argue that human understanding, creativity and critical thinking remain essential as AI advances. Their comments follow OpenAI’s September announcement that an internal model had produced a possible solution to the Navier–Stokes problem, one of seven Millennium Prize Problems. Petrov said AI can provide useful ideas but cautioned that companies may prioritize prestige over scientific progress and ethical standards. Ghane also said studying mathematics can prepare people to help develop future AI systems.
The growth of AI tools is also creating practical pressures. App developers say Apple’s review queues have stretched from the usual one or two days to weeks as more iOS apps and AUv3 plugins arrive, with some submissions facing repeated or confusing rejections. Unfiltered Audio’s Sinusoidal Run Rhythm was rejected four times, including once as MIDI spam, so the developer released only its audio version for now. Pittsburgh Modular’s iOS Microvolt 3901 launched November 12, after its VST version on October 27, despite plans for a simultaneous release. For banks, Feedzai has upgraded Digital Trust to identify AI agents and assess risk throughout a customer session. It says shared signals detected 44% more fraud in the riskiest 0.1% of sessions without raising false positives, while reducing unnecessary challenges by 6,000 per million logins and SMS one-time passcodes by 41%.
AI’s expansion is also shaping jobs, infrastructure and public policy. Hood College and Frederick County Workforce Services are offering Maryland residents a free, 12-week AI-enhanced data analytics program for jobs in health, biotechnology, cybersecurity and logistics. The spring hybrid course runs February 13 to May 7; the county is contributing $34,000, and applications are due December 4. Separately, voters in California, Illinois, Maine, New York, Ohio and other states backed limits on AI expansion or rejected proposals supporting data centers in the 2026 midterms. Concerns cited include electricity demand, pollution, hazardous waste, water use and utility bills.
At the corporate level, the articles describe intensifying competition alongside tighter funding for AI startups, with NVIDIA CEO Jensen Huang’s departure from the company’s board presented as a marker of a “post-Jensen” phase. Cisco reported record quarterly revenue of $18.5 billion, up 24% year over year and its fastest growth in more than 15 years; CEO Chuck Robbins credited demand for AI and machine-learning products, alongside software and cloud services. Google Cloud more than doubled revenue to $30.7 billion in Alphabet’s September quarter, while operating income more than tripled to $11.1 billion. Google is offering select customers on-premises TPU 8tx chips, has nearly 100 customers using its chips, and introduced TPU 8tx for confidential computing and TPU 8i Max for demanding AI workloads. Alphabet reported $127.6 billion in total quarterly revenue, expects most revenue from customer TPU agreements in 2027, and forecast capital spending of $195 billion to $205 billion this year. Free cash flow was negative $9.1 billion, including a $5.2 billion European Commission charge.
Key Takeaways
- Federal officials have not established a cause for Friday’s nationwide FAA ground stop; they are investigating a possible connection to SMART, while the DHS secretary and Microsoft said the outage did not appear linked to Microsoft 365.
- The FAA’s $875 million SMART system is expanding nationally; the agency says it has helped reroute 180 flights and reduce delays, but the air traffic controllers’ union says it did not help design or test it.
- LangChain raised $300 million for its AI-agent tools and reportedly plans to grow its staff to 850 by the end of 2027; its open-source tools receive about 90 million downloads per month.
- Mirendil raised $1 billion in Series A funding, with Google among its investors. It says AI agents wrote more than 99% of the code for a coding model built for $10 million in about 3.5 months.
- OpenAI said an internal model produced a possible solution to the Navier–Stokes problem; Yale mathematicians stress that human mathematical understanding, creativity and critical thinking still matter.
- Apple app developers report review delays of weeks as AI makes it easier to produce apps and plugins; Feedzai says its updated Digital Trust found 44% more fraud in the riskiest 0.1% of sessions without increasing false positives.
- Hood College and Frederick County Workforce Services launched a free 12-week AI-enhanced data analytics course for Maryland residents; applications are due December 4.
- Voters in multiple states backed limits on AI expansion or opposed data-center proposals, citing concerns including electricity demand, pollution, water use and utility bills.
- Cisco’s revenue rose 24% to a record $18.5 billion, with CEO Chuck Robbins citing AI and machine-learning demand as well as growth in software and cloud services.
- Google Cloud revenue more than doubled to $30.7 billion in the September quarter. Google is expanding customer access to its TPUs, while Alphabet forecast $195 billion to $205 billion in capital spending this year and reported negative free cash flow of $9.1 billion.
Officials investigate possible AI link to nationwide FAA ground stop
A telecommunications outage at FAA air traffic control centers led to a nationwide ground stop on Friday, leaving more than 3,100 aircraft bound for U.S. airports in the air. Homeland Security Secretary Markwayne Mullin said the outage does not appear connected to Microsoft 365’s service disruption, and Microsoft also said it had no indication of a link. Officials are investigating whether the FAA’s new cloud-based SMART traffic planning system played a role, but they have not established a cause. SMART uses about 200 data streams, and FAA specialists review its recommendations before local facilities decide whether to use them. The FAA has not said when normal operations will resume.
FAA expands its AI traffic advice system across US airspace
The Federal Aviation Administration has expanded SMART, an AI tool that advises air traffic controllers, across the national airspace system. The $875 million system uses machine learning to review traffic patterns and provide route advice. Since its initial use, SMART has helped reroute 180 flights, and the FAA says it has reduced delays. Airline executives support the rollout, while the union representing U.S. air traffic controllers says it was not involved in designing or testing the system.
Federal officials probe whether FAA AI system is tied to ground stop
Federal officials are investigating whether a new FAA artificial intelligence program contributed to a telecommunications problem that prompted a nationwide airport ground stop on Friday. Homeland Security Secretary Markwayne Mullin said the Department of Homeland Security is working with other agencies through CISA, and stressed that there is no threat to the public. The FAA system uses real-time information such as weather, airport congestion and flight paths to help adjust routes and reduce delays. The FAA began an initial rollout in September and plans to expand the system, with specialists overseeing its use. Officials have not established that the AI system caused the disruption, and technicians are investigating the communications failure at air traffic control centers nationwide.
LangChain raises $300 million to grow its AI agent tools
LangChain, a San Francisco startup known for open-source tools that help developers build AI agents, has raised $300 million. Its LangChain framework connects agents to large language models and other tools, while LangGraph links multiple agents into workflows and can require human approval for sensitive actions. Deep Agents adds features such as a virtual file system, work planning and subagent creation. The company earns revenue from LangSmith, which helps teams build, monitor and maintain agents, including saving versions and finding technical problems. The article says LangChain’s open-source tools were downloaded about 90 million times per month after its previous funding round, and LangSmith is used by Autodesk and Cisco. LangChain reportedly plans to nearly double its staff to 850 by the end of 2027.
Mirendil secures $1 billion to automate AI model development
Mirendil, a startup that makes software to speed up AI development, has raised $1 billion in Series A funding led by Kleiner Perkins and Clearlake Capital, with Google among the other investors. Founded last year by former Google DeepMind and Anthropic researchers, the company developed Lor, a platform that uses hundreds of AI agents to automate model training tasks. Mirendil says one researcher used its technology to build the Mirendil-Base coding model for $10 million in about 3.5 months, with agents writing more than 99% of its code. Mirendil-Base also underpins Fern, a coding model unveiled last month that can process 10 million input tokens per prompt, or 10 times more than Claude Opus 5.5. The company says it plans to use larger versions of Mirendil-Base to develop more capable language models.
Yale mathematicians say human thinking still matters as AI advances
OpenAI said in September that an internal AI model had produced a possible solution to the Navier–Stokes problem, one of the seven Millennium Prize Problems. Yale professor Alexander Petrov said AI can offer useful ideas, but he warned that companies may put prestige ahead of scientific progress and ethical standards. Petrov and Yale students Cyrus Ghane and Ray Zhao said people still need to build mathematical understanding, creativity and critical thinking. Ghane also said studying math is enjoyable and can prepare people to help develop future AI systems.
AI app boom leaves Apple developers waiting weeks for reviews
AI tools have made it easier to create iOS apps and AUv3 plugins, adding to the number of apps Apple must review. Developers say some submissions now wait weeks instead of the usual one or two days, and some receive repeated or confusing rejections. Unfiltered Audio’s Sinusoidal Run Rhythm was rejected four times, including a rejection that called its MIDI version spam, so the developer released only the audio version for now. Pittsburgh Modular’s iOS version of the Microvolt 3901 arrived on November 12, after its VST version launched October 27, despite plans for a simultaneous release.
Feedzai upgrades Digital Trust to help banks detect AI agents
Feedzai has updated Digital Trust to help banks recognize AI agents and assess risk throughout a customer’s banking session. The system combines persistent device recognition, network-wide fraud signals and cryptographic links between a customer, device and bank. It aims to identify suspicious activity while allowing banks to reduce unnecessary security checks and reliance on SMS one-time passcodes. Feedzai said shared signals found 44% more fraud in the riskiest 0.1% of sessions without increasing false positives, and reported 6,000 fewer unnecessary challenges per million logins and 41% fewer SMS OTPs.
Hood College offers free AI analytics training with workforce group
Hood College and Frederick County Workforce Services have launched a free, 12-week AI-enhanced data analytics program for Maryland residents. The training will prepare participants for entry- and mid-level jobs in health, biotechnology, cybersecurity and logistics through lessons in data and AI, analytics, machine learning and applied AI. Frederick County Workforce Services is providing $34,000 and will help recruit participants and connect graduates with local employers. The spring program will run in a hybrid format from February 13 to May 7, and applications are due December 4.
AI competition grows as startup funding dries up in the post-Jensen era
The article describes a new phase in the AI boom, with established industry rivals maturing and startup funding declining. NVIDIA CEO Jensen Huang’s departure from the company’s board of directors is presented as a marker of this post-Jensen era. Competitors are investing in AI research and development, while startups face less funding to innovate and compete. The article says increased competition and shrinking startup cash are likely to shape the industry’s next phase.
Cisco revenue jumps 24% as AI demand boosts networking
Cisco reported record revenue of $18.5 billion, up 24% from a year earlier and its fastest growth in more than 15 years. The results beat analysts’ expectations, and its stock rose 10% in after-hours trading. CEO Chuck Robbins said demand for AI and machine learning products is helping drive networking sales. Growth also came from Cisco’s software business and cloud-based services as customers shift from hardware purchases to software solutions.
US election results raise pressure for tighter AI data center rules
Voters in California, Illinois, Maine, New York, Ohio and other states backed limits on AI expansion or rejected proposals supporting new data centers in the 2026 midterm elections. Some voters also supported gubernatorial candidates who called for pauses on new projects in Georgia, Ohio and Texas. The article says concerns include rising electricity demand, pollution, hazardous waste, water use and utility bills. It urges companies to support responsible development, including clean energy requirements, strong environmental safeguards and rules that protect community rights and ensure local benefits.
Google Cloud sales more than double as Google offers AI chips to customers
Alphabet said Google Cloud revenue more than doubled to $30.7 billion in the September quarter, while operating income more than tripled to $11.1 billion. The company is expanding its Tensor Processing Units beyond its own data centers, allowing select customers to install the TPU 8tx on their premises; nearly 100 customers now use Google’s chips. Google also introduced the TPU 8tx for confidential computing and the TPU 8i Max for demanding AI workloads. Alphabet reported $127.6 billion in total quarterly revenue and expects most revenue from customer TPU agreements to be recognized in 2027. Capital spending of $195 billion to $205 billion this year contributed to negative free cash flow of $9.1 billion, including a $5.2 billion European Commission charge.
Sources
- DHS chief sees no Microsoft link in FAA ground stop, says possible AI role being investigated
- FAA expands AI advice to manage air traffic congestion across entire US airspace
- FAA investigating whether new AI technology contributed to nationwide ground stop
- AI agent tooling provider LangChain secures $300M investment
- Mirendil raises $1B in funding to speed up AI model development
- Despite AI breakthroughs, math majors say human thinking is important
- AI is turning Apple’s app review process into a nightmare for developers
- Feedzai expands Digital Trust to detect AI agents at banks
- Hood College Partners With Workforce Group for AI Analytics Training
- The AI boom enters its post-Jensen era
- Cisco posts strongest revenue growth in 15 years as AI demand fuels networking rebound
- America voted against AI data centers. How business leaders should respond
- Alphabet's Cloud Sales More Than Double as Google Pushes Custom AI Chips Into Customer Data Centers
Comments
Please log in to post a comment.