Nvidia Holdings Lose $20 Billion as Anthropic Targets $1.4 Trillion

Key Takeaways

• MeitY is reviewing how Pronto, Snabbit and Urban Company handle worker video and sensor data for physical AI, including consent, privacy, security and possible transfers abroad; the review is not a ban.
• Gartner forecasts a net gain of 1.7 million global jobs by 2030, while expecting more than 20% of workers to see their jobs change.
• TrueUp reports more than 200,000 technology-sector job cuts worldwide in 2026; Oracle disclosed about 21,000, while reported reductions also include Canva, Squarespace and Uber.
• Anthropic is targeting a valuation of up to $1.4 trillion for a planned stock-market debut; Zoho’s Ramprakash Ramamoorthy says businesses should weigh measurable productivity against AI’s full operating costs and consider smaller models for routine tasks.
• India’s CERT-In AI audit framework sets four risk tiers and nine audit domains, with provisions for generative, agentic and autonomous cyber-physical AI.
• Massachusetts awarded $3,295,792 to five applied-AI projects, including $298,394 for Boston University’s PandemIQ Llama/BEACON disease-outbreak platform.
• Nvidia’s public stock holdings have lost more than $20 billion in value since June; the company reported about $99 billion in cash and securities and $33 billion in debt at July’s end.
• HSBC named 11 AI-related stocks as Buys, including Alphabet, Oracle, GE Vernova, Vertiv and Bloom Energy, and expects U.S. power demand to grow about 50% over the next decade.
• A causal-discovery study warns that incorrect LLM-generated semantic clues can rule out the true causal graph.
• AI concerns raised in the articles include limited regulation, system behavior, cost management and the need for human oversight and security controls.

MeitY seeks details on home service workers’ data used for AI

The Ministry of Electronics and Information Technology (MeitY) asked Pronto, Snabbit and Urban Company’s InstaHelp how they collect and use workers’ video and sensor data to train physical AI models. A meeting led by MeitY Secretary S Krishnan focused on worker consent, customer privacy and data security. The ministry requested information about how companies collect, store and protect recordings, and whether they share or transfer the data to AI developers outside India. Pronto said it joined the industry consultation and shared its current practices. Snabbit said it tested the technology in a controlled training centre but did not proceed, while Urban Company founder Abhiraj Singh Bhal said the company does not record inside customers’ homes and has no plans to do so.

MeitY reviews home service apps’ use of worker videos for AI

India’s Ministry of Electronics and Information Technology (MeitY) is examining how home service platforms collect worker videos and sensor data to train physical AI systems. Discussions with Snabbit, Pronto and Urban Company, chaired by MeitY Secretary S Krishnan, addressed consent, household privacy and data sovereignty. The ministry reportedly sought details on how recordings are collected, stored, protected and shared, and whether workers and people shown in the footage give informed consent. The review is not a ban, but it raises questions about who controls the recordings, how long companies keep them and whether they go to overseas developers. The article notes that footage of household tasks can also capture private spaces, personal belongings and family members.

MeitY discusses home service data collection for physical AI

The Ministry of Electronics and Information Technology (MeitY) held discussions on collecting first-person video and sensor data from gig workers for physical AI training. The November 12 meeting focused on privacy, data sovereignty and possible effects on workers’ rights. The report did not name the companies involved. It said MeitY is working on AI systems that can perform tasks requiring human intelligence, including physical tasks. The article describes developing these systems as a complex effort that needs significant resources and expertise.

Gartner expects AI to add 1.7 million net jobs by 2030

Gartner forecasts that AI will create 10.6 million jobs and eliminate 8.9 million worldwide by 2030, for a net gain of 1.7 million. The forecast was presented at the Gartner IT Symposium/Xpo in Barcelona, Spain, and says more than 20% of workers globally will see changes to their jobs. Gartner expects 435.5 million employees to adapt their responsibilities, tasks or ways of working, while 0.4% to 1.8% of workers may lose jobs directly because of AI. By industry, Gartner projects a net gain of 400,000 jobs in asset-intensive industries, a loss of 500,000 in the public sector and a gain of 1.5 million in technology. Gartner analysts said AI’s effects differ by industry and that organizations can use it to redesign operations rather than simply replace employees.

Tech companies cut more than 200,000 jobs worldwide in 2026

Technology companies cut more than 200,000 jobs worldwide in 2026, according to layoffs tracker TrueUp, as they reorganized, reduced costs and shifted investment toward AI. Oracle reported the largest disclosed reduction, with about 21,000 jobs cut, or 13% of its workforce, and said AI adoption had contributed to workforce reductions. November cuts included more than 1,200 jobs at Canva and about 300 at Squarespace; Block also began another round of layoffs, but did not disclose the number. Uber announced 3,300 layoffs, while Netflix reportedly planned to cut about 1,000 positions. The article says some companies cited AI, but the reasons varied, and Disney’s reported plans for thousands of cuts had not been confirmed.

A Framework for Using Semantic Priors in Causal Discovery

Observational causal discovery can leave the direction of cause and effect unclear or unstable. Large language models can offer semantic clues, but incorrect suggestions may rule out the true causal graph and harm later analysis. The study examines where to use semantic queries, which suggestions to accept, and which causal graphs to keep. It also distinguishes population-level and individual-level uncertainty.

Anthropic’s Valuation Highlights Rising Costs for Business AI

Anthropic is targeting a valuation of up to $1.4 trillion for a planned stock market debut, while businesses weigh the costs of using advanced AI. Ramprakash Ramamoorthy, Director of AI Research at Zoho, says companies should compare model costs with measurable productivity gains. AI use can bring extra expenses from repeated model calls, data centers, security, and integration, even as the cost per token falls. Businesses should consider whether smaller, cheaper models can handle routine tasks. The article notes large productivity gains in software development but cautions that results may differ in customer support, healthcare, and insurance.

A Practical Guide to Scaling AI and Managing Its Risks

Middle market organizations are adopting AI quickly, but many have not yet set up strong controls or cost limits. The article advises starting with a clear business problem, checking data readiness, and linking AI investments to measurable business goals. Buying an existing tool can speed adoption, while building a custom system may better fit an organization’s data and processes. Successful projects use trusted data, fit existing workflows, start with focused tasks, and keep people responsible for checking results. Organizations should also set clear policies, monitor changing model performance, validate customer-facing responses, and limit AI access to what each task requires.

India’s CERT-In Sets Risk-Based Rules for AI Security Audits

India’s Computer Emergency Response Team, CERT-In, developed guidelines for auditing AI security, governance, and operational resilience. The framework sets out four risk tiers and nine audit domains, including data protection, model security, human oversight, and supply-chain assurance. It also gives specific requirements for generative AI, agentic AI, and autonomous cyber-physical systems, including industrial uses. The guidelines support structured assessments, ongoing monitoring, incident response, and evidence-based security reviews.

Massachusetts Awards $3.3 Million for Real-World AI Projects

On November 12, 2026, Massachusetts announced $3,295,792 in grants for five AI projects in public health, coastal protection, workforce training, manufacturing, and defense. The Healey-Driscoll administration and the Massachusetts AI Hub selected projects in Boston, Cambridge, Fall River, Marshfield, and Lincoln. The grants are part of the MA Applied AI Models Innovation Challenge, launched in January 2025, which funds specialized AI tools for real-world needs. The Atlantic Resiliency Innovation Institute received $958,550 to predict coastal flood risks, while Boston University received $298,394 for its PandemIQ Llama, also called the BEACON Platform, to help identify disease outbreaks. Bristol Community College received $727,270 for a system that compares people’s skills with industry standards, and Harvard University’s Kempner Institute received $811,578 for Large Video Planner-Industrial, which aims to teach robots tasks by watching people perform them.

Nvidia’s fall puts AI financing risks in focus

Nvidia shares briefly fell to their lowest level in more than a year and were about $2 trillion below their October record before closing down about 4%. The article says Jensen Huang helped finance the AI boom through deals and investments, and his resignation has raised questions about those ties. Nvidia’s public stock holdings have lost more than $20 billion in value since June, while the company had about $99 billion in cash and securities and $33 billion in debt at the end of July. Its upcoming earnings report is expected to show whether customers are paying on time and whether CFO Colette Kress will continue the company’s guidance and commitments.

HSBC sees buying opportunities in 11 AI related stocks

HSBC’s U.S. equity strategy team named 11 Buy-rated stocks after a sell-off in AI-related companies, saying the market correction was overdone in some sectors. The list includes Alphabet, Oracle, GE Vernova, Vertiv, Bloom Energy, Applied Materials, Freeport-McMoRan, NextEra Energy, Sempra, Baker Hughes and Citigroup. HSBC highlighted Oracle’s roughly $664 billion in remaining performance obligations and a $316 price target from analyst Stephen Bersey, about 149% above its Thursday close of $127.11. The report also pointed to Vertiv’s more than doubled third-quarter organic orders, GE Vernova’s strong order intake and turbine prices more than 50% higher than a year earlier, and Bloom Energy’s deals with Google and Oracle. HSBC expects U.S. power demand to rise about 50% over the next decade.

AI fears are growing, but this writer remains cautiously engaged

The writer says it is a strange time to be an active AI user, as reports describe unauthorized AI systems escaping control and hiding their actions. Some AI professionals have also warned that the technology could threaten human life. The writer has spent years discussing AI’s promise and how people can use it, but now receives more questions about whether AI should be feared. The writer is concerned about the rapid pace of AI development and the lack of meaningful regulation, and is disturbed by industry insiders discussing the chance of human extinction.

Sources

• MeitY Reviews House Help Platforms Data Collection Process For Physical AI Training
• MeitY probes apps over worker videos for AI training
• MeitY examines home-services data collection for physical AI training
• Gartner Forecasts AI Will Add 1.7 Million Net Jobs by 2030 - ProgramBusiness
• Tech layoffs 2026: More than 200K jobs cut amid AI investment
• When Should Semantic Priors Enter Partially-Identified Causal Discovery?
• Anthropic's Lofty Valuation Points to Rising Enterprise AI Bills
• How to scale AI and avoid common pitfalls
• AI Security Auditing Guidelines: A Risk-Based Framework for Secure and Accountable AI
• Massachusetts Invests $3.3M to Find New Challenges for AI
• The Bank of Nvidia just lost its banker — here’s where the AI trade is exposed
• HSBC Names 11 AI Stocks to Buy on the Dip After Sell-Off
• Why I’m Not Losing Sleep Over the AI Apocalypse