AI infrastructure and the companies supplying it feature across several reports. TTM Technologies expects its data-center and networking revenue to more than double in full-year 2026, while Coherent supplies optical components for 800G and 1.6T networks and Fabrinet makes data-center optical transceivers. A separate screen identifies Cadence Design Systems, Accenture and IBM as companies whose governance profiles may reduce dependence on a single leader. Cadence is rolling out AI tools including ChipStack, ViraStack, InnoStack and AuraStack; the screen also found 152 other companies. For businesses selecting development platforms, a review names Google Cloud AI Platform, Microsoft Azure Machine Learning, Amazon SageMaker, IBM Watson and H2O.ai, and highlights expertise, scalability, security and support as selection criteria.
GS Caltex has linked Amazon SageMaker Canvas with SageMaker Pipelines to automate the move from validated models to production. Process engineers build and validate models in Canvas without coding, while machine-learning engineers configure deployment and scheduled inference in Pipelines. The automated handoff cuts deployment from about a week to a few hours. GS Caltex runs batch inference every 30 minutes and says this has meaningfully reduced inference costs.
HUMAN Security says Meta and OpenAI agents generated nearly 90% of agent requests to the sites and apps it protects in October. Requests were about eight times August’s level and 2.6 times September’s, with 85% going to product and search pages. Meta’s Muse accounted for a share 19 percentage points higher than in September and sent 3.7 times as many requests as ChatGPT Agent. Checkout and payment pages received 1.9% of requests, down from 2.4%; the report covers only HUMAN-protected properties, and a checkout request does not establish that a purchase occurred.
Nvidia is facing leadership, governance and market questions after CEO Jensen Huang’s resignation, which the articles connect to reports of an alleged relationship and an unresolved national-security investigation. China denies the spying allegations; Nvidia says it has no reason to believe company secrets or national security were compromised, and its board has begun an independent investigation. One report says Nvidia shares fell 13% last week, wiping out more than $570 billion in market value, while the Philadelphia Semiconductor Index dropped 3.4% on the 13th but finished the week up 8.1%. Morgan Stanley’s Lisa Shalett argues that AI infrastructure orders booked through 2027 and into the first half of 2028 support the broader bull market. Other coverage warns that Huang’s exit could weaken investor and business confidence, disrupt investment-linked collaborations and leave Nvidia exposed to further losses. A GeekWire discussion considers interim CEO Colette Kress and possible successors from Microsoft, Amazon and the chip industry.
Huang’s departure also coincides with growing US debate over data-center construction, including concerns about electricity bills and water use and calls for tighter controls. The article says restrictions may not greatly delay AI progress because companies already have computing resources and AI systems can help improve themselves, while a slowdown in US research could also affect China’s AI industry. Taiwanese financial institutions warn that customers may build second supply chains and turn to ASICs, potentially shifting orders. A separate commenter cautions that state and federal rules for open-source AI could become a costly regulatory patchwork that favors the largest companies and may also affect proprietary models.
AI-assisted security research and agent safeguards raise distinct operational issues. Forescout Research’s Vedere Labs analyzed 10 widely used TCP/IP stacks in 90 hours, finding 99 vulnerabilities—a task estimated to take about nine months without the accelerated approach. Memory corruption accounted for 49% of the bugs. Because TCP/IP software appears across operational technology, IoT, medical equipment and network products, vulnerabilities can have broad reach; Forescout urges organizations to assess reachability, exploitability and impact, limit exposure, isolate unpatchable systems and strengthen procurement requirements. Coinbase, meanwhile, announced server-enforced controls for AI trading agents on November 11: per-trade and 24-hour volume limits, plus a portfolio-loss trigger that pauses new trades until the user re-enables trading in the app. The controls do not promise to cancel open orders or close positions. Agent portfolios support spot crypto, US futures use the default portfolio, and permissions to restrict agents by product category are planned.
Key Takeaways
- GS Caltex uses Amazon SageMaker Canvas and Pipelines to cut model deployment from about a week to a few hours, with batch inference every 30 minutes and lower inference costs.
- HUMAN Security reports Meta and OpenAI agents generated nearly 90% of October requests to protected sites and apps; Meta’s Muse sent 3.7 times as many requests as ChatGPT Agent.
- TTM Technologies expects data-center and networking revenue to more than double in 2026; Coherent supplies 800G and 1.6T optical components, while Fabrinet makes data-center transceivers.
- Cadence is rolling out ChipStack, ViraStack, InnoStack and AuraStack; a separate AI-platform review lists Google Cloud AI Platform, Microsoft Azure Machine Learning, Amazon SageMaker, IBM Watson and H2O.ai.
- Nvidia shares fell 13% last week, losing more than $570 billion in market value, amid Jensen Huang’s resignation and an unresolved national-security investigation; Nvidia’s board has launched an independent inquiry.
- Morgan Stanley’s Lisa Shalett says AI infrastructure orders booked through 2027 and into the first half of 2028 support the market, while Taiwanese institutions warn of potential second supply chains and increased ASIC use.
- Forescout’s AI-assisted review found 99 vulnerabilities across 10 TCP/IP stacks in 90 hours; 49% involved memory corruption.
- Coinbase’s AI-agent controls include per-trade and daily-volume limits and a loss-triggered pause, but do not promise to cancel open orders or close existing positions.
- Coverage links Huang’s exit to questions about Nvidia governance and succession, while US data-center policy debate and proposed open-source AI rules add uncertainty for the sector.
Cadence, Accenture and IBM stand out for AI leadership depth
An article examines Cadence Design Systems, Accenture and International Business Machines as large companies with governance profiles that may reduce reliance on a single leader. Cadence makes chip design software and semiconductor IP, and is rolling out AI tools including ChipStack, ViraStack, InnoStack and AuraStack. Accenture works on AI, cloud and digital transformation projects, while IBM offers hybrid cloud, AI software, consulting and infrastructure. The article says the three companies are examples from a wider screen that identified 152 additional businesses.
TTM Technologies and two peers supply AI data center networks
The article looks at TTM Technologies, Coherent and Fabrinet as suppliers to AI networking and server systems. TTM makes printed circuit boards, RF assemblies and thermal solutions, and expects data center and networking revenue to more than double in full year 2026. Coherent supplies lasers and optical components, including products for 800G and 1.6T networks. Fabrinet makes optical transceivers and related components for data centers. The article says a wider screen found 23 other companies in the same theme.
GS Caltex speeds production machine learning with SageMaker
GS Caltex connects Amazon SageMaker Canvas with SageMaker Pipelines to move validated models and preprocessing flows into production. Process engineers build and validate models in Canvas without writing code, while machine learning engineers set up Pipelines for deployment and scheduled inference. The company automated the handoff, which previously required engineers to set data paths, model settings and processing jobs. GS Caltex says engineers can now deploy models in a few hours instead of about a week. It runs batch inference every 30 minutes, which has meaningfully lowered inference costs.
Five AI and machine learning platforms to consider in 2026
The article lists five AI and machine learning development options for businesses in 2026: Google Cloud AI Platform, Microsoft Azure Machine Learning, Amazon SageMaker, IBM Watson and H2O.ai. It says these platforms and services help developers build, deploy and manage machine learning models, with tools for tasks such as data processing and model training. When choosing a partner, businesses should consider expertise, technology, scalability, security and support. The article notes that AI and machine learning can help automate workflows, improve forecasting, personalize customer experiences and analyze data.
Meta’s Muse leads AI agent requests tracked by HUMAN Security
HUMAN Security reported that Meta and OpenAI agents made nearly 90% of agent requests to sites and apps it protects in October. Total requests were about eight times higher than in August and 2.6 times higher than in September, with 85% directed to product and search pages. Muse’s share rose 19 percentage points from September, and it sent 3.7 times as many requests as ChatGPT Agent. Checkout and payment pages received 1.9% of requests, down from 2.4% the month before. The report covers only HUMAN-protected sites and apps, and a request reaching checkout does not mean a purchase was made.
Nvidia CEO exit rattles stocks, but Morgan Stanley sees AI boom continuing
Nvidia shares fell 13% last week, losing more than $570 billion in market value, as CEO Jensen Huang resigned amid an affair scandal and a national security investigation. The Philadelphia Semiconductor Index dropped 3.4% on the 13th but gained 8.1% for the week. Morgan Stanley Wealth Management’s Lisa Shalett said AI infrastructure spending remains the main driver of the bull market, with orders booked through 2027 and into the first half of 2028. She said political actions may not have a real effect until 2028 or 2029, and regulation could lower bubble risk. Taiwanese financial institutions warned that clients may build second supply chains and use ASICs, which could shift supply chain orders.
AI finds 99 flaws in widely used TCP/IP software in just 90 hours
Forescout Research’s Vedere Labs used AI assisted analysis to examine 10 widely used TCP/IP stacks in 90 hours, work that would have taken about nine months, and found 99 vulnerabilities. The research found that 49% of the discovered bugs involved memory corruption. TCP/IP software appears in operational technology, Internet of Things devices, medical equipment, and network products, so one flaw can affect products from many manufacturers. Forescout said legacy and unsupported devices are especially hard to secure because they may remain in use for decades and may be impossible to patch. The company urged organizations to assess vulnerabilities by reachability, exploitability, and operational impact, limit device exposure, isolate systems that cannot be patched, and strengthen procurement security requirements.
Nvidia faces major risks despite its cheaper stock after Huang’s exit
The article argues that Nvidia’s stock could fall another 50% after Jensen Huang’s resignation, despite being cheaper than it has been in a long time. Congressional investigations into national security issues, including Nvidia’s dealings with China, may continue after Huang leaves. The article also says his departure could weaken Nvidia’s identity and slow collaborations tied to large investments that some analysts have called “circular.” It warns that the resignation may damage the trust investors and businesses place in the company.
GeekWire podcast weighs Nvidia leadership options after Huang resigns
The GeekWire Podcast discusses Nvidia CEO Jensen Huang’s resignation, announced Thursday after reports of extramarital affairs, including one involving a woman suspected of being a Chinese spy. The hosts examine governance concerns at the world’s most valuable chipmaker and discuss interim CEO Colette Kress, whose Microsoft career has been long but low profile. They also consider possible permanent successors from Microsoft, Amazon, and the chip industry. The episode includes the show’s latest AI tips and tricks.
Huang’s Nvidia exit adds uncertainty as US debate over AI data centers grows
Jensen Huang resigned as Nvidia CEO amid reports of a relationship with Alice Huang, a former executive at Nvidia customer Megaspeed International whom US intelligence officials suspect of gathering information for Beijing. China has denied the spying allegations, and Nvidia said it had no reason to believe national security or company secrets were compromised. Nvidia’s board launched an independent investigation, while the allegations remain unresolved. Huang had defended rapid AI infrastructure growth, but his exit comes as Democrats gain support for tighter controls on data center construction and communities raise concerns about power bills and water use. The article says restrictions may not greatly delay AI progress because companies already have computing resources and AI systems can help improve themselves, while a slowdown in US research could also affect China’s AI industry.
Coinbase adds trading limits for AI agents but leaves open orders
Coinbase announced server enforced limits for AI agents using its trading service on November 11. The controls include a maximum amount per trade, a 24 hour trading volume limit, and a portfolio loss trigger that pauses new trades until the user turns trading back on in the Coinbase app. The announcement does not promise to cancel open orders or close existing positions, so a pause does not guarantee that losses are capped. Dedicated agent portfolios support spot crypto only, while US futures require the default portfolio. Coinbase says permissions for choosing which product categories an agent can trade are coming soon.
Commenter warns of a regulatory maze for open source AI
A commenter praised an article about open source AI and Jensen, while noting that personal conduct could affect how people view the events. They said losing the biggest open source advocate and competitor would be a major blow. Drawing on their career in the Department of Defense and federal work, they predicted blunt policy and regulatory changes at both state and federal levels. They warned that these rules could create a complex and costly patchwork that only the largest companies can manage, giving them a regulatory advantage that could also affect proprietary models.
Sources
- Cadence Stock And 2 AI Software Names With Stronger Governance Profiles
- TTM Technologies And 2 AI Infrastructure Stocks To Own
- Production MLOps at GS Caltex with Amazon SageMaker Canvas, part 1
- Top 5 Artificial Intelligence and Machine Learning Development Companies (2026)
- Meta's Muse Now Sends Most AI Agent Requests One Security Firm Sees
- Jensen Huang's Resignation Shakes AI Stocks; Morgan Stanley Says Bull Market's Main Engine Still Running
- AI tears through legacy TCP/IP stacks, exposing vulnerabilities across critical infrastructure, OT and IoT devices
- Think Nvidia Stock Is a Bargain? 6 Reasons the Face of the AI Build-Out Can Plunge Another 50% Following Jensen Huang's Resignation.
- Nvidia’s CEO meltdown, who could replace Jensen Huang, and our latest AI tips and tricks
- Jensen Huang resisted calls to slow AI. Could his Nvidia exit amid China spy scandal reshape the industry's future?
- Coinbase AI trading tool pauses bots, but leaves orders open
- Open-source AI without Jensen
Comments
Please log in to post a comment.