AI Infrastructure AI News & Updates

Nvidia's Vera CPU Targets $200B Agentic AI Market with $20B Initial Sales

Nvidia CEO Jensen Huang announced that the company's new Vera CPU, designed specifically for agentic AI, has already generated $20 billion in sales and opens a new $200 billion total addressable market. Huang argues that while GPUs handle AI "thinking," agents primarily run on CPUs, and Vera's token-processing optimization makes it ideal for the billions of AI agents he predicts will exist. This positions Nvidia to compete directly with Intel, AMD, and cloud providers' custom CPU offerings in the emerging agentic AI infrastructure market.

Google and SpaceX Explore Orbital Data Centers for AI Computing

Google and SpaceX are reportedly in discussions to launch data centers into orbit, potentially revolutionizing AI compute infrastructure. SpaceX is positioning orbital data centers as a cost-effective solution for AI workloads ahead of its $1.75 trillion IPO, with Google planning to launch prototype satellites by 2027 under Project Suncatcher. However, current analysis suggests terrestrial data centers remain more cost-effective when factoring in construction and launch expenses.

AI Industry Leaders Discuss Infrastructure Bottlenecks, Energy Constraints, and Alternative Architectures at Milken Conference

Leaders from across the AI supply chain convened at the Milken Global Conference to discuss critical challenges facing AI development, including severe chip shortages expected to last 3-5 years, energy constraints prompting exploration of space-based data centers, and physical limitations in training real-world AI systems. The panel also explored alternative AI architectures like energy-based models that could run thousands of times faster than large language models, and discussed geopolitical sovereignty concerns around physical AI deployment.

Stripe Launches Link Digital Wallet with Autonomous AI Agent Payment Capabilities

Stripe has introduced Link, a digital wallet designed for both human users and autonomous AI agents to manage payments securely. The wallet allows users to grant AI agents controlled spending permissions without exposing raw payment credentials, using OAuth authentication and approval workflows. Link supports payment methods including cards, banks, crypto wallets, and buy now/pay later services, with plans to add agentic tokens and stablecoins.

Google Commits Up to $40B to Anthropic Amid Escalating AI Compute Race

Google plans to invest up to $40 billion in Anthropic, with $10 billion committed immediately at a $350 billion valuation and $30 billion contingent on performance targets. The investment includes providing 5 gigawatts of computing capacity over five years, following Anthropic's release of its most powerful model, Mythos, which has significant cybersecurity applications but restricted access due to misuse concerns. This deal is part of an intensifying competition for AI compute resources, with Anthropic securing multiple infrastructure partnerships including additional investments from Amazon totaling up to $100 billion in compute capacity.

Google Cloud Unveils Specialized TPU 8t and TPU 8i Chips for AI Training and Inference

Google Cloud announced its eighth generation tensor processing units (TPUs), splitting into two specialized chips: TPU 8t for model training and TPU 8i for inference. The new chips promise 3x faster training, 80% better performance per dollar, and support for clusters exceeding 1 million TPUs. Despite this advancement, Google continues to offer Nvidia's latest chips alongside its own custom processors, with both companies collaborating on networking optimization.

OpenAI's Acquisition Strategy and Anthropic's Powerful Unreleased Model Highlight Growing AI Industry Divide

OpenAI is aggressively acquiring companies across various sectors including finance apps and media properties, while a shoe company has repositioned itself as an AI infrastructure provider. Anthropic has developed a model deemed too powerful for public release but suitable for demonstration to Federal Reserve Chair Jerome Powell, highlighting a widening gap between AI insiders and the general public.

Google and Intel Expand Multi-Year Partnership for AI Infrastructure and Custom Chip Development

Google and Intel announced an expanded multi-year partnership where Google Cloud will utilize Intel's Xeon 6 processors for AI, cloud, and inference workloads. The companies will also continue co-developing custom infrastructure processing units (IPUs) to accelerate data center tasks, addressing the growing industry demand for CPUs needed to run AI models.

Cognichip Raises $60M to Use AI for Accelerating Semiconductor Chip Design

Cognichip has raised $60 million to develop deep learning models that assist engineers in designing computer chips, aiming to reduce development costs by over 75% and cut timelines by more than half. The company uses proprietary AI models trained on chip design data rather than general-purpose LLMs, though it has not yet delivered a chip designed with its system. Notable investors include Intel CEO Lip-Bu Tan, and the company competes with established players like Synopsys and well-funded startups in the AI chip design space.

SK hynix Plans $10-14 Billion U.S. IPO to Fund AI Memory Chip Expansion Amid 'RAMmageddon' Crisis

SK hynix, a major South Korean memory chip manufacturer, has confidentially filed for a U.S. listing targeting the second half of 2026, potentially raising $10-14 billion. The company, a critical supplier of high-bandwidth memory (HBM) for AI systems, aims to close its valuation gap with global peers and fund massive capital investments totaling $400 billion by 2050 for semiconductor facilities. The move comes amid a severe memory shortage dubbed 'RAMmageddon' that is constraining AI development and other industries.