Today's AI Daily Brief covers NVIDIA's massive revenue predictions and the enterprise race to productize AI agents. Host discusses NVIDIA CEO Jensen Huang's GTC conference announcements, including a trillion-dollar revenue forecast and the introduction of NemoClaw for enterprise agent deployment.

The episode explores the infrastructure buildout happening across the AI industry, from Meta's $27 billion deal with Nebius to OpenAI's Stargate restructuring. Major focus on the 'clawification' trend where companies are building OpenClaw alternatives and enterprise-ready agent systems.

Coverage includes strategic shifts at major AI companies, with OpenAI refocusing on enterprise productivity after CEO Fiji Simo's 'code red' directive, and Chinese companies like Alibaba moving away from open source models toward revenue-generating closed systems.

NVIDIA's Trillion-Dollar Vision and GTC Highlights

Jensen Huang announced NVIDIA expects $1 trillion in revenue between now and 2027, doubling previous forecasts from $500 billion in 2026 to $500 billion annually by 2027.

New Grok-powered server combines 256 Grok chips with 72 NVIDIA Rubin GPUs, delivering 35x the inference efficiency of current Blackwell chips, shipping second half of this year.

DLSS5 technology unveiled for real-time AI-enhanced video game graphics, combining traditional rendering with AI filters for photorealistic results on consumer hardware.

"Computing demand has increased by 1 million times in the last two years" - Jensen Huang, justifying the massive revenue projection.

AI Infrastructure Gold Rush and Data Center Deals

Meta signed a $27 billion five-year deal with Nebius for AI data centers, adding to a previous $3 billion November deal, representing order-of-magnitude growth for the neo-cloud provider.

OpenAI restructured Stargate division under former Intel executive Sachin Kati, organizing three teams for technical design, commercial partnerships, and facility management.

OpenAI is now more willing to lease rather than own data centers, prioritizing compute access over ownership in line with industry trends.

Encyclopedia Britannica and Merriam-Webster sued OpenAI for using their content in training data and allegedly cannibalizing web traffic through ChatGPT responses.

The Shift Away from Open Source AI Models

Alibaba restructured their Qwen team into the 'Alibaba Token Hub' focused on monetizing AI, with CEO Eddie Wu emphasizing 'create tokens, deliver tokens, and apply tokens.'

ZhipuAI released GLM-5 Turbo as closed source despite being 'the loudest open source voice in AI for two years,' signaling industry shift toward proprietary models.

Chinese labs are adopting a hybrid approach: lightweight open source models for distribution and developer goodwill, while keeping powerful enterprise-focused models proprietary.

"We're in the era when the cost of building LLMs is skyrocketing and the why for releasing them openly is static" - Nathan Lambert on the economics driving closed models.

The Clawification of Enterprise AI Agents

NVIDIA introduced NemoClaw, an enterprise-grade toolkit built on OpenClaw with policy-based security, guardrails, and isolated sandboxes for safe enterprise deployment.

"Every software company in the world needs to have an OpenClaw strategy" - Jensen Huang, positioning agent systems as essential infrastructure.

Multiple companies launched OpenClaw alternatives: Manus Desktop with 'My Computer' feature, Adaptive Computer with encoded memory, and Perplexity Computer for Enterprise.

OpenClaw creator Peter Steinberger collaborated with NVIDIA on NemoClaw, calling it 'a huge step towards secure agents you can trust.'

OpenAI's Strategic Refocus on Enterprise Coding

CEO Fiji Simo delivered a 'wake-up call' telling staff 'We cannot miss this moment because we are distracted by side quests,' refocusing on enterprise productivity.

GPT-4o reached 5 trillion tokens per day within a week of API launch, handling more volume than OpenAI's entire API one year ago with $1 billion annualized revenue run rate.

Codex introduced native sub-agent integration, allowing users to 'delegate all of the lower complexity tasks to GPT-4o-mini sub-agents' for parallel processing.

"We are very much acting as if it's a code red" - Fiji Simo, indicating continued urgency despite recent successes in the coding market.