google, nvidia and microsoft Updates

Google is embarking on an ambitious expansion of its artificial intelligence capabilities, planning a massive 1,000-fold increase in AI capacity by 2030. This strategy involves doubling its AI serving capacity every six months and a 15-fold boost by the end of 2024, driven by rising demand for services like Gemini and Google Cloud. The company is heavily investing in custom hardware, including its own Tensor Processing Units (TPUs), Ironwood, and Axion chips, alongside Nvidia AI accelerators. DeepMind CEO Demis Hassabis credits Google's control over its entire AI stack, from custom silicon to software, as the secret behind Gemini 3's success, with the model trained entirely on specialized TPUs. This aggressive growth comes as analysts suggest the industry is entering "stage two of AI," where physical limits like power and cooling present the main challenges. Amidst this expansion, Google recently addressed user concerns regarding data privacy, specifically denying claims that it secretly trains its Gemini AI model using private Gmail data. A Google spokesperson clarified that such allegations are false and that the company does not use Gmail content for Gemini training without explicit user permission. Users retain control over data sharing through the "Smart features and personalization" toggle in Gmail settings, with Google emphasizing that explicit consent is required for personalized features. Meanwhile, Microsoft faces hurdles in the widespread adoption of its Copilot AI assistant among businesses. Many IT buyers at the Microsoft Ignite conference have questioned the value proposition of the commercial 365 Copilot, which costs $30 per person per month. Despite Microsoft's significant investment in OpenAI and its strong cloud infrastructure with Azure, customers are not consistently seeing enough return on investment to justify the monthly fee, leading some companies to opt for alternative AI coding tools. Beyond these tech giants, AI is finding diverse applications and sparking broader discussions. Cities like San Jose, California, and states such as Hawaii and Texas are deploying AI-powered cameras to inspect aging roads, identify potholes with high accuracy, and improve overall road safety. In the media sector, AI's impact is evident as traditional radio hosts in Oklahoma transition to online platforms like Facebook and YouTube, reflecting changing listener habits. The sports world also sees AI in action, with SportsLine's Data Science team using a self-learning AI to predict NFL game winners and scores. Additionally, content platform Beehiiv has integrated new AI tools for website building and digital product sales, aiming to become a comprehensive infrastructure for content businesses. On the global stage, Iran plans to host its first International Artificial Intelligence Exhibition, Iran AIX Expo 2026, in Tehran, fostering collaboration among AI companies and industries. As AI progresses, experts like Paulo Carvão emphasize that human agency, rather than fear, must guide its future, with researchers noting that current AI excels at pattern recognition but still lacks true generalized understanding or self-awareness.

Key Takeaways

• Google plans a 1,000-fold increase in its AI capacity by 2030, aiming to double AI serving capacity every six months and achieve a 15-fold boost by the end of 2024.
• Google is heavily investing in custom hardware, including its own Tensor Processing Units (TPUs), Ironwood, and Axion chips, alongside Nvidia AI accelerators, to support its AI growth.
• Google denies using private Gmail data to train its Gemini AI model, stating that explicit user permission is required for such use.
• Google DeepMind CEO Demis Hassabis attributes Gemini 3's success to Google's control over its entire AI stack, from custom silicon to software, with the model trained on specialized TPUs.
• Microsoft is encountering challenges with the widespread adoption of its 365 Copilot, as businesses question the $30 per person per month cost and its return on investment.
• AI-powered cameras are being used by cities and states, including San Jose and Hawaii, to inspect roads, identify hazards, and improve safety with high accuracy.
• AI is impacting jobs, with radio hosts in Oklahoma transitioning from traditional airwaves to online platforms.
• Beehiiv has launched new AI tools for website building and digital product sales, aiming to provide comprehensive infrastructure for content businesses.
• Experts emphasize that human agency should guide AI's future, noting that current AI excels at pattern recognition but lacks generalized understanding or self-awareness.
• Iran will host its first International Artificial Intelligence Exhibition, Iran AIX Expo 2026, in Tehran from January 12 to 15, 2026.

Google plans massive AI growth by 2030 despite bubble worries

Google has internally ordered a huge 1,000-fold increase in its AI capacity by 2030. This plan goes against CEO Sundar Pichai's public concerns about an AI market bubble. Infrastructure VP Amin Vahdat announced this goal, which includes doubling AI serving capacity every six months. The company will use custom chips like Ironwood and Axion to achieve this growth efficiently. This strategy aims to meet the rising demand for AI services, especially for "inference" tasks.

Google boosts AI serving power as physical limits emerge

Google's AI infrastructure head, Amin Vahdat, announced plans to double the company's AI serving capacity within the next year. This expansion aims to support a massive increase in users and complex requests for AI products like Gemini and Google Cloud. Analyst Shay Boloor from Futurum Equities believes this marks "stage two of AI," where physical limits like power and cooling are the main challenges. Google is using its own Ironwood chips to improve computing capacity and meet this high demand.

Google plans huge AI capacity boost by 2024

Google plans to increase its AI serving capacity by 15 times by the end of 2024. This aggressive expansion shows a big rise in AI use and deployment. The company invests heavily in specialized hardware like its own Tensor Processing Units and Nvidia AI accelerators. A Bernstein analyst says we are now in "stage two of AI," where physical limits are the main challenge. This strong demand and investment suggest the AI market is not a bubble.

Google denies using Gmail for Gemini AI training

Google has denied claims that it secretly trains its Gemini AI model using users' private Gmail data. A Google spokesperson stated these claims are false. Concerns spread on social media, especially X, about personal email communications being used. Google clarified it does not use Gmail content for Gemini training without a user's clear permission. This incident highlights ongoing user worries about data privacy with AI development.

Google refutes claims of using Gmail for AI training

Google strongly denies allegations that it uses Gmail emails and attachments to train its artificial intelligence models. The company states that recent viral reports suggesting a policy change are misleading. Google clarifies that it does not funnel individual Gmail content directly into models like Gemini. Users can manage data sharing through the "Smart features and personalization" toggle in Gmail settings. While Google uses anonymized data for AI improvements, it emphasizes that explicit user consent is required for personalized features.

Google DeepMind CEO shares Gemini 3 success secret

Google DeepMind CEO Demis Hassabis revealed the "secret" behind Gemini 3's success is Google's control over the entire AI stack. This includes everything from custom silicon to software. Google has a long history in AI, pioneering the transformer architecture and making other breakthroughs. The 2023 merger of DeepMind and Google Brain created Google DeepMind, combining top research with strong engineering. Google also uses its own Tensor Processing Units, or TPUs, which are custom chips designed for AI workloads. Gemini 3 was fully trained on these specialized TPUs, giving Google a unique hardware advantage.

Iran to host first AI exhibition in January 2026

Iran will host its first International Artificial Intelligence Exhibition, called Iran AIX Expo 2026, in Tehran. The event will take place from January 12 to 15, 2026, at the Tehran International Permanent Fairground. The Iran Trade Promotion Organization officially licensed the expo, with support from the Ministry of Information and Communications Technology. This gathering aims to boost collaboration among AI companies, financial groups, and leading industries. The exhibition will feature expert panels, workshops, and networking opportunities to strengthen partnerships and share knowledge.

AI cameras help fix America's roads and spot dangers

Cities and states are using artificial intelligence to inspect America's aging roads and identify hazards. Hawaii officials are giving away 1,000 Nextbase dashboard cameras for their "Eyes on the Road" campaign. These cameras use AI to automatically check guardrails, road signs, and pavement markings, finding problems quickly. San Jose, California, successfully used cameras on street sweepers to identify potholes with 97% accuracy. Texas also uses AI with cameras and cellphone data to improve road safety. This technology helps officials prioritize repairs and prevent accidents, as seen in Hawaii's efforts to reduce traffic fatalities.

AI threatens jobs in Oklahoma including radio hosts

Artificial intelligence is putting thousands of jobs at risk in Oklahoma, including those in radio. Oklahoma City radio legends Jack Elliott and Ron Williams, known for their "Jack and Ron" show, are no longer on traditional airwaves. They now host their show online on platforms like Facebook, YouTube, and iTunes. This change reflects a trend where traditional radio listeners are tuning into voices from distant stations, hinting at AI's growing impact on various professions.

AI predicts NFL Week 12 game winners and scores

A self-learning AI from SportsLine's Data Science team is providing NFL picks and score predictions for Week 12 games. This AI uses historical data and continuously updates to evaluate team strengths and predict outcomes. The AI PickBot has successfully hit over 2,000 top prop picks since the 2023 season. For Week 12, the AI predicts the Seattle Seahawks will comfortably cover their -12.5 spread against the Tennessee Titans, forecasting a 30-14 win. It rates the Seahawks as an A+ spread pick.

Beehiiv adds AI tools and digital sales on anniversary

Beehiiv celebrated its fourth anniversary by launching new features, including AI tools for website building, podcast support, and the ability to sell digital products. Co-founder and CEO Tyler Denk explained that these additions respond to user requests for more flexible and customizable platforms. Beehiiv aims to be a comprehensive infrastructure for content businesses, similar to Shopify, rather than just a newsletter platform. The company emphasizes its model of only charging a subscription fee, not a percentage of user revenue. Beehiiv plans to focus on marketing and brand awareness while continuing to add AI-powered tools and diverse content formats.

Human control is key for AI future not fear

Paulo Carvão argues that human agency must guide the future of AI, not existential fear. Some thinkers warn that advanced AI could soon surpass human abilities, leading to unforeseen and dangerous outcomes if goals are misaligned. They point to rapid progress and the potential for AI to become smarter than humanity. However, others argue that current AI excels at pattern recognition, not true generalized understanding. Researchers like Gary Marcus and Yann LeCun suggest that scaling alone will not lead to super-intelligence. They believe AI still lacks self-awareness and motivation, emphasizing that human responsibility remains crucial.

Microsoft struggles to sell Copilot AI widely

Microsoft faces challenges in getting its Copilot AI assistant widely adopted by businesses. Many IT buyers at the Microsoft Ignite conference question the value of the commercial 365 Copilot, which costs $30 per person per month. Consultants note that customers are not seeing enough return on investment to justify the cost. While Microsoft has a strong presence in cloud infrastructure with Azure and a big investment in OpenAI, selling AI agents like Copilot is a different challenge. Some companies are choosing other AI coding tools and assistants over Microsoft's offerings. Despite some discounts, many users do not feel they get $30 worth of value from Copilot each month.

Sources

• Google Mandates 1,000x AI Capacity Growth by 2030 Despite 'AI Bubble' Fears
• As Google eyes exponential surge in serving capacity, analyst says we're entering 'stage two of AI' where bottlenecks are physical constraints
• As Google eyes exponential surge in serving capacity, analyst says we’re entering ‘stage two of AI’ where bottlenecks are physical constraints
• Google: No, We're Not Secretly Training Gemini on Your Gmail Account
• Google Denies Gmail Data Used for AI Training Amid Backlash and Misinformation
• Demis Hassabis Reveals Google’s ‘Secret’ Behind Benchmark-topping Gemini 3
• Iran to Hold Its First International Artificial Intelligence Exhibition -
• From guardrails to potholes, AI is becoming the new eyes on America’s roads
• AI killed the radio star? Thousands in Oklahoma at risk of losing job to AI
• Week 12 NFL picks, top predictions, best bets today from self-learning AI
• Beehiiv expands with AI tools and digital product sales on anniversary | Ukraine news
• Human Agency Must Guide The Future Of AI, Not Existential Fear
• Microsoft faces uphill climb to turn enterprise dominance into widespread AI chatbot adoption