AI billion-dollar surge with OpenAI $852B valuation, xAI co-founder exits, Meta Hyperagents and next-gen AI models

OpenAI Hits $852B Valuation, $100M Ads ARR in Record Time, While xAI Loses All 11 Original Co-Founders

OpenAI closed a $122 billion funding round this week at an $852 billion valuation, with SoftBank, Amazon, and Nvidia among the anchoring investors. And ChatGPT Ads pilot exceeded $100 million in annualized recurring revenue (ARR) within 6 weeks.

Not every AI company is scaling up. xAI lost its final two co-founders this week, completing a full exit of all 11 original co-founders. Musk has acknowledged the company needs a ground-up rebuild, which makes the timing of these departures more consequential than routine attrition. Google had a dense week on multiple fronts. Let’s dive in.

OpenAI Raises $122B, Hits $852B Valuation Before IPO

OpenAI closed a $122 billion funding round at an $852 billion valuation. OpenAI investors, SoftBank, Andreessen Horowitz, Amazon, and Nvidia anchored the raise. The company generates $2 billion in monthly revenue and serves 900 million weekly ChatGPT users, but remains unprofitable ahead of a potential IPO.

xAI Loses All Co-Founders as Musk Rebuilds From Scratch

All 11 xAI cofounders exit, as Manuel Kroiss and Ross Nordeen, the last two remaining, departed this week. Kroiss led pretraining, and Nordeen was Musk’s right-hand operator. Musk recently said “xAI was not built right the first time” and is now rebuilding from the ground up.

ChatGPT Ads Hit $100M Run Rate in Just Six Weeks

OpenAI’s Ads pilot crossed $100 million in annualized revenue within 6 weeks of launch, signaling strong early traction. Over 600 advertisers are now on board, with self-serve capabilities launching in April to accelerate OpenAI monetization further.

Understanding The Economics of Event Transport

Most event teams reconcile only direct transport costs like emergency vehicle hire and staff overtime. Indirect costs, including sponsor activation loss, F&B revenue compression, and refund liability, outpace direct costs by a 3:1 to 5:1 ratio and rarely appear in any post-event financial review.

Google Gemini Poaching Rival AI Apps

Google rolled out Gemini app memory and chat history import tools for all consumer accounts. Users can transfer personal context, preferences, and full conversation history from ChatGPT or Claude directly into Gemini. The move targets the Gemini app acquisition as AI assistants compete to reduce switching friction.

Google TurboQuant Just Made AI Models Run on a Laptop

Google Research announced TurboQuant, a new AI model compression algorithm. Known as the KV cache, it shrinks AI working memory by at least 6x without impacting performance. Using vector quantization methods, PolarQuant and QJL, it could make AI significantly cheaper to run during inference.

Google’s ‘Agent Smith’: Matrix-Level AI Agents Just Got Real

Google built Agent Smith, an internal autonomous coding and workflow agent, for its employees. It plans tasks, executes workflows, and operates asynchronously across internal systems. Access got restricted within weeks due to overwhelming internal demand, signaling strong productivity gains in software engineering workflows.

The Hidden Cost of Manual Event Shuttles (And How to Fix It)

Most event transport systems work. Until they suddenly don’t. This framework helps ops leaders assess when manual dispatch breaks down and whether to shift to hybrid or full automation based on complexity, frequency, and risk. This guide shows how to assess when your event operations need to move to hybrid or fully automated transport.

Meta Releases Hyperagents, a Self-Modifying AI System

The latest Meta Hyperagents framework introduces self-improving AI that edits its own optimization procedures. Built on the DGM-H architecture, it scored 0.710 on paper review tasks and transferred learned strategies to Olympiad math grading, outperforming traditional self-improvement baselines by a significant margin.

Meta Builds AI That Maps How Your Brain Processes Input

Meta’s FAIR team built TRIBE v2 (Trimodal Brain Encoder v2), a multimodal brain-response model that predicts how different brain regions react to video, audio, and language simultaneously. Trained on large-scale fMRI data, it generalizes neural predictions across individuals better than earlier single-modality models.

Anthropic Confirms Claude Mythos After Accidental Data Leak

Anthropic confirmed Claude Mythos, a new model tier above Opus, after a CMS misconfiguration exposed nearly 3,000 unpublished assets. The model scores dramatically higher than Opus 4.6 on coding and reasoning, but raises serious AI cybersecurity risk, with Anthropic calling it “far ahead of any other AI model in cyber capabilities.

When Event Shuttles Go Off‑Schedule: How to Handle the Crisis

Four gaps account for most event transport incident failures: no classification system, undocumented protocols, limited real-time visibility, and undefined escalation paths. Each gap is foreseeable and preventable. Without fixing them before event day, dispatchers improvise under pressure with no consistent framework.

Court Sides With Anthropic in Pentagon Supply-Chain Dispute

A federal judge ruled against the Trump administration in the Anthropic Pentagon fight, ordering it to rescind Anthropic’s “supply-chain risk” designation and restore federal agency ties. The dispute started after Anthropic sought to ban its AI models from use in autonomous weapons and mass surveillance systems.

Oracle Lays Off Thousands as AI Debt Drains Its Cash Flow

Oracle is cutting thousands of jobs as its stock drops 25% this year. The company raised $50 billion in debt for Oracle AI infrastructure buildouts while cash flow tightened. Analysts at TD Cowen estimate cuts of 20,000 to 30,000 employees could free $8 to $10 billion in incremental cash flow.

What Else Is Happening?

Subscribe to our tech news newsletter. Receive regular insights that keep you informed. And if you find it valuable, please share it with your network to help spread the word. 
Catch you next time with fresh insights on AI and Tech, right here.