Google Launches Private AI Compute with AMD and Apple

The artificial intelligence landscape continues to evolve rapidly, with major players like Google, Tesla, and Eli Lilly making significant moves in privacy, hardware, and drug discovery. On November 11, 2025, Google introduced Private AI Compute, a new cloud platform designed to offer advanced AI features while ensuring user data remains private, even from Google itself. This system, which operates similarly to Apple's Private Cloud Compute, leverages Google's custom Tensor Processing Units (TPUs) and an AMD-based Trusted Execution Environment to secure information. It powers features such as Magic Cue on Pixel 10 phones and enhances the Recorder app's summarization capabilities across more languages, with Jay Yagnik emphasizing Google's commitment to responsible AI. Meanwhile, Elon Musk outlined ambitious plans for Tesla at its 2025 shareholder meeting, proposing a "terafab" – a massive chip manufacturing plant. This facility aims to secure Tesla's supply of AI chips, as current suppliers like TSMC and Samsung may not meet future demand. Tesla's in-house AI5 processor, designed for autonomous and robotics programs, is nearing production, with Musk stating it uses less power and costs less than Nvidia's Blackwell. The terafab would initially produce 100,000 wafer starts per month, eventually scaling tenfold to support Tesla's AI ambitions, including Optimus robots and autonomous fleets. However, Truist analyst William Stein questions the timeline for these "physical AI" products to generate significant income, noting that Tesla's shift from hardware sales to intelligence sales, akin to Nvidia's model, remains largely untested. Beyond these developments, AI is making waves across various sectors. Eli Lilly is expanding its partnership with Insilico Medicine for AI-based drug development, a collaboration that could bring Insilico over $100 million in payments and royalties. Lilly, which has used Insilico's Pharma.AI software since 2023, also launched an "AI factory" with Nvidia on October 28, utilizing 1,016 NVIDIA Blackwell Ultra GPUs for accelerated drug discovery. In finance, a WealthCharts panel on November 11, 2025, discussed how users trust AI tools like Anthropic's Claude for smaller investments, but still prefer human advisors for critical retirement accounts, highlighting the importance of trust in financial advising. Fastly's CEO, Kip Compton, is navigating the challenges of AI and internet security, particularly concerning AI bots that scrape content for large language models. Fastly offers AI bot mitigation services and partners with TollBit to help publishers control and monetize access to their content. Additionally, Iconic Digital, a London-based marketing agency, launched an online AI hub on November 11, 2025, to help businesses and government leaders understand and adopt AI trends. Despite the widespread adoption and innovation, challenges persist. An MIT 2025 report indicates that 95% of enterprise AI initiatives fail to deliver a measurable impact, often due to unclear business goals, poor project management, and inadequate data quality. Furthermore, content platforms like Wikipedia face threats from AI models such as Elon Musk's "Grokipedia" and products like Perplexity, which use Wikipedia's data but siphon off traffic by providing direct answers without linking back to original sources. This trend could endanger the site's future and reduce web diversity, underscoring the need for users to verify information sources. Even in fields like architecture, while AI offers an "extra set of hands" to speed up design, experts worry about the potential loss of critical thinking and human reflection in the creative process.

Key Takeaways

• Google launched Private AI Compute on November 11, 2025, offering secure AI experiences with Gemini models, similar to Apple's Private Cloud Compute, utilizing Google's TPUs and an AMD-based Trusted Execution Environment.
• Tesla's Elon Musk plans a "terafab" chip manufacturing plant to produce its AI5 processor, which he claims is more power- and cost-efficient than Nvidia's Blackwell, starting with 100,000 wafer starts per month.
• Truist analyst William Stein expresses skepticism about the near-term revenue generation from Tesla's "physical AI" products, including self-driving cars and Optimus robots.
• Eli Lilly expanded its AI drug discovery partnership with Insilico Medicine, potentially paying over $100 million, and launched an "AI factory" with Nvidia on October 28, using 1,016 Blackwell Ultra GPUs.
• Consumers trust AI tools like Anthropic's Claude for small investments but prefer human advisors for critical retirement accounts like 401(k)s and IRAs.
• Fastly CEO Kip Compton addresses the challenge of AI bots scraping content, offering mitigation services and partnering with TollBit to help publishers control and monetize their data.
• An MIT 2025 report indicates that 95% of enterprise AI initiatives fail to deliver measurable impact, often due to unclear goals, poor project management, and inadequate data quality.
• Wikipedia faces a threat from AI models like Elon Musk's "Grokipedia" and products like Perplexity, which use its data but siphon off traffic by providing answers without links.
• Architects view AI as a tool to speed up design and spark ideas but are concerned about losing critical thinking and human reflection in the creative process.
• Iconic Digital launched an online AI hub on November 11, 2025, to provide businesses and government leaders with guides and insights on rapidly changing AI trends.

Google unveils Private AI Compute for secure AI experiences

Google launched Private AI Compute on November 11 2025, combining powerful Gemini AI models from the cloud with strong privacy features. This new platform ensures personal data stays private, even from Google, similar to on-device processing. Private AI Compute uses Google's custom Tensor Processing Units and encryption to protect user information. It allows for faster, more helpful AI responses while keeping data secure. Jay Yagnik, Vice President of AI Innovation and Research, highlighted Google's ongoing commitment to responsible AI.

Google launches Private AI Compute like Apple's

On November 11 2025, Google introduced Private AI Compute, a new cloud platform. This system allows users to access advanced AI features while keeping their data private. It works similarly to Apple's Private Cloud Compute, sending complex AI tasks to a secure cloud space. Google states that sensitive data processed by Private AI Compute is only accessible to the user, not even Google, aiming to provide more personal and tailored AI suggestions.

Google's Private AI Compute matches local security

Google introduced Private AI Compute, a new cloud system designed to offer secure AI experiences. Google claims this platform is as safe as on-device processing, similar to Apple's Private Cloud Compute. It runs on Google's custom Tensor Processing Units, which use an AMD-based Trusted Execution Environment to keep data private. This technology helps power features like Magic Cue on Pixel 10 phones and allows the Recorder app to summarize in more languages. While local AI offers speed and offline use, Google sees this hybrid cloud approach as the future for complex AI tasks.

Elon Musk plans Tesla Terafab for AI chips

At Tesla's 2025 shareholder meeting, Elon Musk discussed building a "terafab," a huge chip manufacturing plant. This plant would secure Tesla's supply of AI chips, as current suppliers like TSMC and Samsung might not meet future demand. Tesla's in-house AI5 processor is nearing production, designed for autonomous and robotics programs. Musk stated the AI5 uses less power and costs less than Nvidia's Blackwell. The proposed terafab would start with 100,000 wafer starts per month, eventually multiplying tenfold to power Tesla's AI ambitions, including Optimus robots and autonomous fleets.

Analyst questions Tesla's physical AI product timeline

Tesla's future growth depends on its "physical AI" products, such as self-driving cars and humanoid robots. Truist analyst William Stein believes these products are still far from generating significant income. Tesla aims to shift from selling hardware to selling intelligence, similar to Nvidia's success with AI chips. However, Tesla's physical AI ventures remain largely untested. Wall Street analysts are divided on Tesla's long-term stock outlook, with 14 recommending a Buy, 10 a Hold, and 10 a Sell.

People trust AI for small trades not retirement

On November 11 2025, a panel at WealthCharts discussed how people use AI in trading. Chris Josephs, co-founder of Autopilot, noted that users might let AI manage small investments like "Robinhood money." However, they prefer human advisors for important retirement accounts like 401(k)s and IRAs. The panel, including Rob Hoffman and James Putra, agreed that trust is crucial in financial advising, even as AI tools like Anthropic's Claude become more common. This suggests a future where human expertise and AI work together in finance.

Eli Lilly partners with Insilico for AI drug discovery

Eli Lilly is expanding its partnership with Insilico Medicine for AI-based drug development. This collaboration could bring Insilico over $100 million through upfront and milestone payments, plus royalties. Lilly has used Insilico's Pharma.AI software since 2023, and now they will jointly discover new therapies. Insilico's Pharma.AI platforms, including Generative Chemistry and Chemistry42, will combine with Lilly's expertise. Separately, Lilly also launched an "AI factory" with Nvidia on October 28, using 1,016 NVIDIA Blackwell Ultra GPUs for accelerated drug discovery.

Architects discuss AI innovation and human intent

An HOK Up Next panel of experts discussed the role of artificial intelligence in architecture. Architects are excited about AI's ability to speed up design and spark new ideas, acting like an "extra set of hands." However, they also worry about losing critical thinking and human reflection in the design process. Panelists like Supp, Song, and Harrison emphasized that AI should improve, not replace, a designer's creative work. They also explored how AI will change architectural education and what skills future architects will need.

Fastly CEO navigates AI and internet security

Fastly's CEO, Kip Compton, is guiding the company through the challenges of AI and internet security. Fastly, founded by Artur Bergman, aims to make the internet fast, safe, and engaging. The company offers a unified platform for content delivery, edge compute, and security services like WAF and DDoS protection. A major concern for customers is managing AI bots that scrape content for training large language models. Fastly's AI bot mitigation service and partnership with TollBit help publishers control and monetize access to their valuable content.

Iconic Digital launches AI hub for business growth

On November 11 2025, Iconic Digital, a digital marketing agency in London, launched a new online hub. CEO Steve Pailthorpe created stevepailthorpe.com to help businesses and government leaders understand AI's rapid changes. The hub offers guides, articles, case studies, and expert insights on AI trends and how to adopt them. Iconic Digital aims to make AI knowledge accessible for businesses of all sizes to innovate and gain a competitive edge.

Common mistakes cause AI projects to fail

Most AI projects fail to deliver real value despite huge investments, according to Kathleen Walch from the Project Management Institute. MIT's 2025 report shows 95% of enterprise AI initiatives do not make a measurable impact. Common mistakes include unclear business goals, poor project management, and overpromising what AI can do. Companies also underestimate needed resources, ignore real-world challenges, and use poor quality data. To succeed, organizations must define clear goals, use skilled project managers, be realistic about AI's limits, and ensure data quality.

Marubozu candle reveals strong market signals

Traders use chart patterns like the "marubozu candle" to understand market signals. This candle, a solid block without wicks, shows strong conviction in the market. A bullish marubozu means buyers were in full control, while a bearish one shows sellers dominated. Like AI models, traders combine this signal with other data like volume and trend for better predictions. Algorithms can now spot these patterns quickly, but human judgment remains key for interpreting their significance in context. The marubozu helps traders confirm market shifts and make decisions.

How to save Wikipedia from AI threats

Wikipedia is facing a threat from artificial intelligence, according to Nadav Ziv and Sam Wineburg. AI models, including Elon Musk's "Grokipedia," use Wikipedia's data but siphon off its traffic, endangering the site's future. AI products like Perplexity offer quick answers without links, reducing user clicks to original sources. This trend, similar to early search engines, could harm content creators and lead to a less diverse web. To fight back, users need to learn to check citations and trace information to its source.

Sources

• Private AI Compute: our next step in building private and helpful AI
• Google is introducing its own version of Apple’s private AI cloud compute
• Google says new cloud-based “Private AI Compute” is just as secure as local processing
• Elon Musk Floats ‘Terafab’ as Tesla’s Next Big AI Chip Bet
• Tesla Stock Gets Humbled as Analyst Says “Physical AI Products Are Still a Long Way Off”
• People Will Let AI Trade Their 'Robinhood Money'— But Not Their Retirement
• Lilly Expands AI Ties to Insilico, from Customer to Drug Discovery Partner
• AI in Architecture: Balancing Innovation with Intentionality
• Fastly CEO plots course through AI and security
• Iconic Digital Launches New AI Hub to Help Businesses Future-Proof Their Growth
• Why your AI projects keep failing
• From AI Signals to Market Signals: Decoding the ‘Marubozu Candle’ in Trading Charts
• Commentary: How to save Wikipedia from AI