AI's capital and its models both moved in bulk over the past day: Nvidia rallied more than half a trillion dollars in outside financing for data centers, new specialized models arrived from NVIDIA and OpenAI, and Anthropic began watermarking Claude's output. Here's the rundown →

TOGETHER WITH OUR SPONSOR

NVIDIA releases Nemotron 3.5 Lightning, an open-weight agent model. NVIDIA published Nemotron 3.5 Lightning, a model with 30 billion total parameters that activates roughly 3 billion per token through a mixture-of-experts design and is distilled from its larger Nemotron 3 Ultra. NVIDIA aims it at high-volume, long-running agent workloads and says it runs on a single consumer RTX GPU or a DGX Spark, claiming up to four times faster output and about 30% faster task completion than comparable open models. The weights are free for commercial use on Hugging Face and NVIDIA's build platform, and the company also released NeMo Switchyard, an open-source library that routes agent requests to the most efficient model for each task. Source

OpenAI ships GPT-5.6-Cyber, a model tuned for security work. OpenAI introduced GPT-5.6-Cyber, a variant of its GPT-5.6 Sol model trained for cybersecurity tasks such as vulnerability discovery and exploit development. On an internal evaluation it completed 95% of advanced security requests, against about 1.5% for the general Sol model, and OpenAI split its Daybreak defender program into two tiers—Blue for vetted defenders using general models, and Red, the only route to GPT-5.6-Cyber. OpenAI said it used the model to find two previously unknown vulnerabilities in V8, the JavaScript engine in Chrome, which Google has since patched; access is limited to approved security professionals. Source

Anthropic begins watermarking Claude's text output. Anthropic said it is embedding machine-readable, invisible watermarks into text generated by Claude models launched on or after August 2, and attaching signed C2PA provenance metadata to generated files such as PNG, JPG, and SVG images. It is applying the markings worldwide rather than only in the EU, citing transparency requirements under Article 50 of the EU AI Act. Anthropic cautioned that the signals show content was processed by Claude rather than proving authorship, and noted they can be removed by editing, format conversion, or screenshots. Source

Nvidia lines up more than $500 billion in financing for AI infrastructure. Nvidia said it is working with six large financial firms—Apollo, Blackstone, BlackRock, Brookfield, Goldman Sachs, and KKR—to mobilize over $500 billion in third-party capital for AI infrastructure. The aim is to create dedicated pools of credit at favorable rates so hyperscalers, AI labs, and enterprises can build data centers and buy Nvidia hardware without draining their own cash. CEO Jensen Huang described GPU clusters as long-term, revenue-generating assets and said he approached only those six firms, none of which declined. Source

Unitree opens a heavily oversubscribed humanoid-robot IPO. Chinese robot maker Unitree Robotics opened subscriptions for its initial public offering on Shanghai's STAR Market, seeking about 6.1 billion yuan (roughly $904 million) at a valuation near $9 billion and becoming mainland China's first publicly traded humanoid-robot company. Retail demand was extreme, with the retail tranche reported oversubscribed thousands of times over. DeepSeek is among the offering's strategic investors. Source

River AI raises $1.1 billion to help enterprises build custom models. River AI, a roughly two-month-old startup founded by former xAI, OpenAI, and DeepMind researcher Igor Babuschkin, raised $1.1 billion in a round co-led by General Catalyst and AMP PBC, with participation from NVIDIA, AMD Ventures, Temasek, and Y Combinator. The company is building a platform that lets enterprises fine-tune, train, and serve their own models—based on open-weight systems—without a dedicated infrastructure team, and says such runs can be two to four times more cost-effective than closed alternatives. River AI argues companies will increasingly own and customize models rather than rely solely on general-purpose systems from large labs. Source

TOGETHER WITH OUR SPONSOR

Tax season doesn't have to mean wondering if you have the right forms, second-guessing your deductions, or scrambling to pull everything together before the deadline.

With BELAY’s tax prep support, you can approach tax season with confidence. Stay organized with one centralized place to gather and check off your documents, keep track of valuable deductions like HSA contributions and education expenses while leaning on experienced professionals who make tax preparation accurate, efficient, and completely hands-off.

Download BELAY's free Personal Tax Checklist and start preparing with confidence, today.

Download the Free Checklist

That's today's AI Current. Thanks for reading—reply and tell us what you'd like us to cover next. See you tomorrow.

TOGETHER WITH OUR SPONSOR

Dictate into Slack, email, LinkedIn, or any app and get polished, send-ready text. Wispr Flow strips filler and formats everything. 89% of messages sent with zero edits. Works on Mac, Windows, and iPhone.

Start flowing free