Thinking Machines Labs loses an important co-founder to Meta, OpenAI partners up with Broadcom for even more AI compute, and Anduril unveils an AI-powered helmet for soldiers that could reshape the battlefield.
Here are our main key takeaways in more detail:
Thinking Machines Labs co-founder Andrew Tulloch departs to return to Meta
OpenAI and Broadcom partner up to deploy 10 GW worth of OpenAI’s in-house designed AI chips
Anduril debuts EagleEye, an AI-powered mixed reality helmet that gives soldiers a “new teammate”
Join us at AI Tangle as we untangle this week's happenings in AI!
AN AI CONFERENCE FOR EXECUTIVES
On the 27th and 28th October at the UNC Charlotte City Center, join the AI//FORWARD premier AI conference for senior business executives and enterprise leaders.
Over those two days, you will learn first-hand about strategic frameworks, operational playbooks, and P2P exchange for leaders tasked with building out their AI from pilots to full-scale enterprise productivity, with keynotes and workshops held by AI leaders like The AIE Network's very own John Willis and Mark. R. Hinkle.
However, spots to the event are limited - grab your invitation and purchase a ticket now!
6 QUICK HITS
Following up on its August hiring blitz, Meta has hired Thinking Machines Labs co-founder Andrew Tulloch, soon after announcing his departure to fellow employees in a message on Friday, as confirmed by a spokesperson. The Wall Street Journal reports that Meta could have offered Tulloch a compensation package worth as much as $1.5 billion over six years. Meta had previously attempted to acquire Mira Murati’s AI startup for a reported billion-dollar offer earlier in the summer of this year.
OpenAI has announced yet another compute deal, this time with American chip manufacturer Broadcom, to help deploy its own in-house designed AI chips. A blog post from OpenAI claims the contract to be worth 10 gigawatts of compute by the end of 2029, though no financial figures were disclosed, with Broadcom also denying the potential of OpenAI being its coincidentally recent $10 billion customer. The deal is the latest of OpenAI’s spree of AI chip and compute deals, including one with Nvidia in September and another with AMD just last week.
Founded by the creator of Oculus, Palmer Luckey’s AI defense startup, Anduril, has just unveiled EagleEye, an AI-driven mixed-reality (MR) helmet system made to give soldiers on the battlefield a “new teammate.” With help from Meta, EagleEye integrates a heads-up display, spatial audio, drone or robotics control, and much more to give soldiers an informational and tactical advantage, as showcased in an X/Twitter thread. The device expands on Anduril’s previous iterations of military MR software and marks a reunion between Luckey and Zuckerberg, whose Meta acquired Oculus in 2014 before parting ways in 2017.
Nvidia has announced that its DGX Spark “personal AI supercomputer” will be hitting store fronts starting October 15th to sate an emerging market of home-owned AI computers. The $3,999 device, available on Nvidia’s website and partner stores, comes with data-center-level power in a rather tiny package, featuring the company’s GB10 Grace Blackwell Superchip, 128GB of unified memory, and up to 4TB of NVMe SSD storage. One of Spark’s biggest selling points, its one petaflop of AI performance, is the ability to handle models with up to 200 billion parameters by itself, from Qwen3 to FLUX.1.
In a first-of-its-kind penning, California Governor Gavin Newsom has signed off on a landmark bill that demands AI chatbot providers to ensure sufficient safety mechanisms for AI companions. Dubbed SB 243, the new law is designed with vulnerable and under-18 individuals in mind, and can hold chatbot operators legally accountable with fines upwards of $250,000 per offense. Per the bill’s language, SB 243 requires operators to implement age verification, warnings regarding AI companions, and clarity that any interactions are artificial. SB 243 was first introduced in January 2025 and will go into effect on January 1st, 2026.
On Monday, Salesforce revealed that its AI platform, Agentforce 360, will be made available globally across its suite of cloud tools. The platform helps Salesforce clients bring AI-based automation to their businesses, from simple tasks to connecting data, users, and AI agents on a singular platform. As of today, Agentforce 360 already serves a reported 12,000 customers, including Reddit, OpenTable, and Adecco. The company also announced a broader integration between Agentforce 360 and Slack, along with a new AI agent building tool, Agentforce Builder.
4 AI TOOLS
Alta - Alta is an AI-powered writing assistant that understands your style and energy. Alta helps users beat writer's block and turn ideas into content or help brainstorm new ones.
Olvy - Surveys, interviews, reviews, support tickets, and sales calls - bring them together into one workspace for faster insights using state-of-the-art automated AI by Olvy.
AI EXTRA READ
How Easy it is to Poison LLMs (5-min read)
A recent joint study held by Anthropic, together with the UK AI Security Institute and the Alan Turing Institute, found that poisoning LLMs during training is scarily easy - no matter the size of the model. From bad behavior to serious security risks, it only takes a near-constant number of malicious samples.
MORE FROM THE AIE NETWORK
📻 AI Confidential Podcast - Are LLMs Dead?
💼 The Artificially Intelligent Enterprise - OpenAI DevDay 2025 Announces Huge AI Updates
💡 AI CIO - "Failing AI" Narrative Misses the Mark
🎯 The AI Marketing Advantage - 2025 Was The Year Branding Went Synthetic
What did you think of this newsletter? Let us know!
Your AI Sherpa,
Mark R. Hinkle
Publisher, The Artificially Intelligent
Enterprise (TheAIE) Network
Connect with me on LinkedIn
Follow me on X
