Meta Platforms is significantly increasing its investment in custom chips for training artificial intelligence models. This strategic move aims to reduce its reliance on external chip manufacturers like Nvidia and AMD, while also improving performance and potentially lowering costs. Meta's CFO, Susan Li, noted that the company's unique tasks make custom silicon a practical choice, expanding beyond current uses in content ranking to more complex AI model training. Meanwhile, Nvidia is reportedly halting production of its H200 AI chip, though this remains unconfirmed. In other developments, AI safety company Anthropic has resumed discussions with the Pentagon regarding potential AI collaborations, and Apple is expected to announce new products soon, possibly including updates to its iPad and Mac lines.
Microsoft is making strides in healthcare AI, showcasing its Dragon Copilot at HIMSS26. This platform unifies AI tools to streamline workflows for clinicians and revenue teams, aiming to save time and enhance patient care. Dragon Copilot, now in public preview, will integrate technologies such as Canary's voice analysis for mental health and Humata's prior authorization tools, with plans for hundreds of plugins. Separately, Vmake AI offers marketers a solution for quickly creating product photos and videos, combining AI models, video enhancement, and image editing. NJ Transit is also piloting AI-powered bus stop signs to provide real-time arrival information and gather customer data, aiming to improve rider experience and optimize routes.
The rapid global push for "AI sovereignty," where countries seek control over their AI systems and data, raises significant human rights concerns. Critics warn this trend could lead to increased censorship and surveillance, particularly in authoritarian regimes, and potentially justify restricting free speech under the guise of reflecting local culture. A Utah lawmaker, Representative Doug Fiefia, has likened AI chatbots to "social media on steroids," advocating for legislation to protect minors from harmful content and require AI companies to post safety plans. The booming demand for AI data centers is also creating a massive need for land and power, with AI companies projected to require 85 gigawatts by 2030, exceeding current grid capacity. Despite these challenges, the Network Security market saw a 10 percent growth in 2025, with Application Security and Delivery growing 15 percent, driven by the adoption of AI workflows.
Key Takeaways
- Meta is increasing its in-house AI chip development to reduce reliance on Nvidia and AMD, aiming for better performance and lower costs in AI model training.
- Nvidia is reportedly halting production of its H200 AI chip, though this information is unconfirmed.
- Anthropic has resumed talks with the Pentagon about potential AI collaborations, indicating military interest in advanced AI.
- Microsoft's Dragon Copilot, now in public preview, unifies AI tools for healthcare to improve patient care and streamline workflows, with plans for hundreds of plugins.
- The global pursuit of "AI sovereignty" poses risks to human rights, potentially leading to increased censorship and surveillance.
- A Utah lawmaker is advocating for legislation to regulate AI chatbots, comparing them to "social media on steroids" due to concerns about minors' exposure to harmful content.
- The demand for AI data centers is driving a significant need for land and power, with AI companies projected to require 85 gigawatts by 2030.
- The Network Security market grew 10% in 2025, with Application Security and Delivery increasing 15% to nearly $6 billion, fueled by AI workflows.
- Vmake AI offers marketers a tool for AI video and image editing, speeding up content creation for e-commerce.
- NJ Transit is piloting AI-powered bus stop signs to provide real-time arrival information and optimize services.
Meta to build custom chips for AI training
Meta plans to create its own computer chips specifically designed to train artificial intelligence models. CFO Susan Li stated that some company tasks are very unique, making custom silicon a good choice. This move aims to improve performance and potentially lower costs by reducing reliance on outside chip makers. Meta hopes to expand the use of these custom chips beyond current applications like content ranking and recommendations to include AI model training.
Meta boosts in-house AI chip development
Meta Platforms is increasing its efforts to create custom chips for training artificial intelligence models. This strategy aims to lessen the company's dependence on chip manufacturers like Nvidia and AMD. Meta has already used its own chips for tasks such as content rating and recommendations. The company now wants to use these custom chips for more complex AI models to improve performance and cut costs. This initiative is part of a larger trend where major tech companies are developing their own chips for AI infrastructure.
Vmake AI review: A new tool for AI video and image editing
Vmake AI offers a solution for marketers needing to quickly create product photos and videos. The tool combines AI models, video enhancement, and image editing in one platform. In testing, Vmake AI proved easy to use, with its AI model studio and image editing tools receiving high marks. While the AI model studio needs improvement for complex poses, its ability to generate professional-looking images and enhance videos significantly speeds up content creation for e-commerce. Vmake AI is praised for its value in streamlining marketing workflows.
AI race risks human rights globally
The global push for 'AI sovereignty,' where countries aim to control their own AI systems and data, could threaten human rights. While intended to reduce reliance on foreign tech giants, these initiatives may lead to increased censorship and surveillance, especially in authoritarian nations. Developing AI models that reflect local culture could be used to justify restricting free speech. Furthermore, AI investments in countries lacking strong privacy laws may enhance government monitoring capabilities. The complex, international supply chains for AI technology make true national control difficult.
Nvidia halts H200 chip production, Anthropic talks to Pentagon
Nvidia is reportedly stopping the production of its H200 AI chip, which could affect the availability of high-end AI hardware. The company has not yet confirmed this information. Meanwhile, Anthropic, an AI safety company, has resumed discussions with the Pentagon about potential AI collaborations. This indicates the U.S. military's ongoing interest in using advanced AI technologies. In other tech news, Apple is expected to announce new products soon, possibly including updates to its iPad and Mac lines.
Microsoft unifies AI tools in Dragon Copilot at HIMSS26
Microsoft is showcasing its Dragon Copilot at HIMSS26, highlighting how unified AI tools can save time and improve patient care. The new integrations allow clinicians and revenue teams to access AI insights directly within their existing workflows. Dragon Copilot, now in public preview, aims to reduce the time spent navigating systems. The platform will feature integrations with technologies like Canary's voice analysis for mental health and Humata's prior authorization tools. Microsoft plans to offer hundreds of plugins for Dragon Copilot, making it a versatile tool for healthcare.
Utah lawmaker warns AI chatbots are 'social media on steroids'
A Utah lawmaker is comparing AI chatbots to 'social media on steroids,' expressing concern about their impact on minors. Representative Doug Fiefia believes these bots engage in human-like conversations and may expose children to harmful content or encourage self-harm. He is pushing for legislation to ban chatbots that provide harmful material to minors and require AI companies to post safety plans. Some parents, like Jennie DeSerio whose son died by suicide, support stronger AI regulations, fearing that current measures are insufficient to protect children.
NJ Transit tests AI signs for bus info
NJ Transit is piloting new bus stop signs that use artificial intelligence to gather customer data and show real-time arrival information. This test aims to enhance the rider experience by providing more accurate and accessible updates. The AI signs will track passenger numbers and wait times to help optimize bus routes and schedules. Commuters will receive live updates on bus arrivals, reducing uncertainty. This initiative is part of NJ Transit's plan to modernize its services and improve efficiency.
AI drives 15% growth in application security and delivery
The global Network Security market grew 10 percent in 2025, reaching over $26 billion, with Application Security and Delivery seeing a significant 15 percent increase to nearly $6 billion. Dell'Oro Group reports that as businesses adopt AI workflows and modernize services, the web and API layer becomes critical for availability and performance. This growth highlights a trend where companies are consolidating delivery and protection into integrated platforms to manage the increased complexity and security needs driven by AI and APIs. Web Application Firewalls (WAF) and Application Delivery Controllers (ADC) remain key areas of investment.
AI data centers spur demand for land and power
The booming demand for artificial intelligence data centers is driving a search for land and power across the country. Companies like Cloverleaf Infrastructure are working to secure vast amounts of electricity and suitable land for these facilities. This 'powered land' has become a valuable commodity as tech giants plan for massive data center construction. The need for power is so great that AI companies are projected to require 85 gigawatts by 2030, exceeding current grid capacity. This surge is leading to intense competition for resources and sparking community discussions about data center development.
Sources
- Meta Plans to Develop Custom Chips to Train Its AI Models
- Meta Expands Push for In House AI Training Chips
- Recensione Vmake AI 2026: è il tuo nuovo punto di riferimento per l’editing video e immagini IA?
- Human Rights at Risk in the Sprint Toward AI Sovereignty
- Tech stocks today: Nvidia stops H200 chip production, Anthropic restarts talks with Pentagon
- Microsoft's AI tool unification in Dragon Copilot takes center stage at HIMSS26
- Utah lawmaker calls AI chatbots 'social media on steroids'
- Coming to bus stops soon - NJ Transit bus signs will use AI in test
- AI Accelerates API Sprawl: Application Security and Delivery Up 15 percent in 2025, According to Dell’Oro Group
- Demand for AI Data Centers Sends Prospectors Hunting for Land and Power
Comments
Please log in to post a comment.