Nvidia Accelerates AI Infrastructure While Anthropic Faces Lawsuit

The rapid expansion of artificial intelligence is driving a massive infrastructure boom, with an estimated $5.2 trillion investment needed by 2030 for data centers and power grids. Companies like Equinix, which formed a $15 billion joint venture in 2024, and Digital Realty, launching a $10 billion fund in 2025, are building large-scale data facilities. Energy providers such as NextEra Energy and Williams are simultaneously expanding power grids and gas pipelines to meet the surging demand, as the U.S. alone may require 50 gigawatts of power for AI by 2028. This growth, particularly evident in Northern Virginia, the world's data center capital, brings both economic benefits and environmental concerns. While these facilities contribute $9.1 billion annually to Virginia's economy and create 74,000 jobs, they consume vast amounts of water and electricity, raising worries about energy bills and noise for residents. Senator Bernie Sanders has called for a temporary halt on new data center construction, citing concerns about jobs, wealth distribution, and environmental impacts. This contrasts sharply with President Trump's administration, which has supported rapid AI development, eliminating AI chip export limits and accelerating data center construction, even engaging with figures like Nvidia CEO Jensen Huang. Amidst this expansion, major AI companies face legal challenges. Authors, including journalist John Carreyrou, have filed a lawsuit against Anthropic, Google, OpenAI, Meta, xAI, and Perplexity, alleging that these companies trained their AI models using pirated copyrighted books. This legal action highlights ongoing debates about intellectual property and fair use in AI development. Meanwhile, advancements continue, with Liquid AI launching LFM2-2.6B-Exp, an experimental language model that uses pure reinforcement learning to enhance instruction following and math abilities for small AI models, making them efficient for devices like phones and laptops. Despite AI's growing capabilities, the human element remains crucial. Cybersecurity firm Cyber Crucible, led by CEO Dennis Underwood, finds AI surprisingly increases human interaction, allowing sales teams to focus on closing deals while AI handles client finding. However, AI's limitations are also apparent; an AI triathlon coach named Hugo from HumanGO struggled to adapt to real-life training changes, requiring manual adjustments. The CEO of Serve Robotics emphasizes a user-centric approach for AI product development, focusing on human needs, understanding AI's strengths and weaknesses, empowering human-AI collaboration, and adding delightful features to ensure products are truly loved.

Key Takeaways

• The AI revolution is projected to require $5.2 trillion in infrastructure investment by 2030, primarily for data centers and power grids.
• Equinix formed a $15 billion joint venture in 2024, and Digital Realty launched a $10 billion fund in 2025 to build large-scale data centers.
• The U.S. may need 50 gigawatts of power for AI by 2028, leading companies like NextEra Energy and Williams to expand energy infrastructure.
• Northern Virginia's data centers generate $9.1 billion annually and 74,000 jobs but raise concerns about electricity consumption, water use, and noise.
• Authors, including John Carreyrou, are suing Anthropic, Google, OpenAI, Meta, xAI, and Perplexity for allegedly training AI models on pirated copyrighted books.
• President Trump's administration has supported big tech and AI development by eliminating AI chip export limits and speeding up data center construction, including interactions with Nvidia CEO Jensen Huang.
• Senator Bernie Sanders advocates for a temporary halt on new AI data center construction due to concerns about jobs, wealth concentration, and environmental impacts.
• Liquid AI introduced LFM2-2.6B-Exp, an experimental language model using pure reinforcement learning to improve performance on small devices like phones and laptops.
• Cybersecurity firm Cyber Crucible uses AI to automate tasks, allowing human employees to focus on higher-value interactions and sales.
• AI training coaches, like HumanGO's Hugo, can provide structure but currently lack the flexibility to adapt to real-life changes in user schedules.

AI Gold Rush Boosts Data Centers and Power Grids

The AI revolution needs huge investments in infrastructure, estimated at $5.2 trillion by 2030. This includes data centers, which house AI hardware, and massive amounts of electricity. Companies like Equinix, Digital Realty, and Brookfield Infrastructure are building large-scale data centers. Equinix formed a $15 billion joint venture in 2024, and Digital Realty launched a $10 billion fund in 2025. Energy companies such as NextEra Energy and Williams are expanding power grids and gas pipelines to meet the growing demand. The US alone may need 50 gigawatts of power for AI by 2028.

AI Makes Cybersecurity Firm Focus on People More

Dennis Underwood, CEO of Cyber Crucible in Fox Chapel, states that AI has surprisingly increased human interaction at his cybersecurity company. While many companies use AI for initial candidate screening, Cyber Crucible now prioritizes face-to-face and video interviews to ensure candidates truly possess the needed skills. The company also uses AI in its sales department to automate tasks like finding clients and personalizing messages. This allows the sales team to focus on closing deals. Underwood believes AI is a powerful assistant but requires human supervision for quality control.

Virginia Data Centers Show AI Future Challenges

Northern Virginia, known as the world's data center capital, processes over half of the internet's traffic. These hundreds of facilities in Loudoun and Fairfax counties create 74,000 jobs and add $9.1 billion to Virginia's economy each year. However, the rapid growth of data centers consumes vast amounts of water and electricity, leading to concerns about rising energy bills and noise for residents. States across the country are watching Virginia's efforts to regulate this development. While some experts like Ali Fenn of Lancium highlight the huge power consumption, others like Dan Dorio of the Data Center Coalition suggest a minimal impact on energy costs.

Authors Sue Major AI Companies Over Copyrighted Books

A group of authors, led by journalist John Carreyrou, has filed a new lawsuit against several major AI companies. The lawsuit targets Anthropic, Google, OpenAI, Meta, xAI, and Perplexity. Authors claim these companies trained their AI models using pirated copies of copyrighted books. This action follows an earlier case against Anthropic, where a court found that while AI training might be legal, book piracy is not. The plaintiffs argue that past settlements did not fully address the widespread use of their works. They seek more accountability for AI companies profiting from content without proper authorization.

President Trump Supports Big Tech on AI and Chips

Since returning to the White House in January, President Trump has surprisingly aligned with major tech companies. Despite earlier vows to fight tech giants, his administration has granted many of their wishes. Since the summer, Trump eliminated several limits on AI chip exports and sped up the construction of data centers that power AI development. This alliance benefits both sides, though its long-term impact is uncertain. For example, President Trump was seen on a phone call with Nvidia CEO Jensen Huang this month. This shift could affect fast-growing AI technologies and influence next year's midterm elections.

Liquid AI Boosts Small Model Performance with New Training

Liquid AI has launched LFM2-2.6B-Exp, an experimental version of its LFM2-2.6B language model. This new model uses pure reinforcement learning to significantly improve instruction following, knowledge tasks, and math abilities for small AI models. It is designed for efficient use on devices like phones and laptops. The LFM2-2.6B-Exp maintains the original architecture but enhances its behavior through a special training stage. On the IFBench instruction following benchmark, it outperforms much larger models. This makes Liquid AI's LFM2 models strong competitors in the 3 billion parameter segment.

My Experience With an AI Triathlon Training Coach

Charlie Allenby tested an AI coach named Hugo from the HumanGO platform for an eight-week triathlon training plan. The AI coach created a personalized workout schedule based on Charlie's fitness levels and race goals. Initially, the plan seemed effective, mixing low and high-intensity sessions and forecasting improved readiness. However, during a holiday in week four, the AI coach could not adapt to a request to swap bike workouts for runs. This lack of flexibility meant Charlie had to adjust his training manually, which reduced his projected potential. The experience showed that while AI coaches offer structure, they currently lack the human understanding needed for real-life changes.

Four Steps to Build AI Products People Will Love

The CEO of Serve Robotics suggests four ways to create AI products that people will truly love. First, always start by focusing on what users need, not just what the technology can do. Second, understand AI's strengths and weaknesses, like its ability to minimize false positives or false negatives, and design around these limits. Third, empower people to work with AI, allowing humans to handle tasks that AI finds difficult while AI takes on mundane or complex data analysis. Finally, go beyond basic functionality to add surprising and delightful features that exceed user expectations. These steps help ensure AI products are useful, beneficial, and charming.

Bernie Sanders Calls for Halt on New AI Data Centers

Senator Bernie Sanders has called for a temporary stop on building new data centers, which are vital for AI development. He wants to slow down the process to better understand and regulate AI's impact on jobs and its benefits for the wealthy. Sanders also expressed concerns about the environmental effects and rising electricity prices caused by these centers. In contrast, the Trump administration supports rapid AI development, viewing it as a race against China. Trump even issued an executive order to prevent states from regulating AI. Sanders believes the Trump administration is influenced by powerful tech companies, citing Elon Musk's large contributions to Trump's campaign.

Sources

• From Power Grids to Data Centers: The Overlooked Winners in the AI Gold Rush
• AI prompts increased human interaction at this cyber security company
• What ‘data center alley’ portends for America’s AI-powered future
• Prominent Authors File Copyright Lawsuit Against Leading AI Developers Over Training Data
• From A.I. to Chips, Big Tech Is Getting What It Wants From Trump
• Liquid AI’s LFM2-2.6B-Exp Uses Pure Reinforcement Learning RL And Dynamic Hybrid Reasoning To Tighten Small Model Behavior
• Can an AI training coach get you race-ready in 8 weeks?
• Four ways to make AI products that people will love
• Bernie Sanders demands moratorium on new data centers to slow down...