- AI Tangle
- Posts
- ☕️ Is OpenAI About To Become a Fully For-Profit AI Company?
☕️ Is OpenAI About To Become a Fully For-Profit AI Company?
With the release of OpenAI's expensive-to-use o1-preview and o1-mini along with a massive funding round that could put it at a $150 billion valuation, OpenAI could be looking to become a for-profit corporation. Other key takeaways of the week include:
Microsoft launches a Copilot agent builder and expands Copilot integration into more of its production suite apps
AMD beats Intel to secure a contract to supply chips to power Sony's future PlayStation 6 console
Intel pauses chip fabrication plans in Germany and Poland but inks a contract with AWS to co-develop an AI chip
Join us at AI Tangle as we untangle this week's happenings in AI!
THE BIG AI STORY
Though the Silicon Valley AI company has been acting the part for a while, OpenAI's future may include formalizing its transition into a more profit-driven organization. The company is in talks to raise $6.5 billion in funding and finalize it in the coming two weeks, which could value it at $150 billion, though this is contingent on removing its profit cap. This change would allow investors higher returns and further distance OpenAI from its original non-profit mission of creating AI for the greater good.
Where is OpenAI's new direction heading?
OpenAI's potential restructuring could convert it into a for-profit benefit corporation, scrapping the current 100x profit cap on investor returns. Early investors like Microsoft and Nvidia would benefit massively from the structure change, though it ultimately requires approval from the board. As OpenAI's focus on AI monetization accelerates, this change could fuel concerns over governance and mission drift. Despite its origins as a non-profit, OpenAI's rapid commercialization, including over 200 million ChatGPT users, signals a turn toward financial grandeur.
6 QUICK HITS
Microsoft is expanding its AI integration across its Microsoft 365 apps, introducing new Copilot features for Excel, PowerPoint, Outlook, and Teams in its "Copilot Wave 2" rollout. Key updates include Copilot's ability to draft presentations, prioritize emails, and a new collaborative tool called Copilot Pages for real-time data manipulation. Microsoft also debuted a Copilot agent builder to help users create tailored AI assistants. All in all, these developments aim to boost productivity and leverage Microsoft's AI investments, with Copilot usage increasing significantly.
A recent report from Reuters states that, amidst Sony's PlayStation 6 focuses, the final battle happened between AMD and Intel to secure the contract for the chip powering the future console ended with AMD coming out on top. AMD, which already powers the PS5 and upcoming PS5 Pro was chosen in part due to its ability to ensure backward compatibility. Intel was allegedly eliminated from contention due to disagreements over profit-sharing, despite the potential $30 billion revenue value of the deal.
Intel is restructuring its Intel Foundry division into an independent subsidiary, maintaining internal leadership but adding an operating board with independent directors. Amid cost-cutting efforts, Intel put major chip fabrication projects in Germany and Poland but inked an important deal with AWS to co-develop an AI chip using Intel's 18A fabrication process, alongside a custom Xeon 6 processor contract. Along with a $3.5 billion Pentagon contract, these deals helped offset losses, despite a tough fiscal year marked by layoffs and division sell-offs.
Oracle's latest earnings call saw founder Larry Elison reveal that, during dinner with Tesla CEO Elon Musk and Nvidia CEO Jensen Huang, he and Musk had to "beg" Huang for GPUs to power their AI ambitions. The meeting turned out to be productive for Ellison, as Oracle announced plans for a Zettascale AI supercluster with 131,072 Nvidia GPUs, surpassing Musk's xAI cluster Colossus. To meet its energy needs, Oracle is looking into modular nuclear reactors, though mobile generators may be used temporarily.
Meta is restarting its plans to train AI systems using public Facebook and Instagram posts from UK users following a pause due to regulatory concerns and objections from the Information Commissioner's Office (ICO) regarding user consent. UK users will shortly receive notifications about the opt-out process, which Meta states is now simpler, though it hasn't explained exactly how so. Despite privacy concerns, Meta argues its approach complies with legal standards, though critics question its reliance on the "legitimate interest" legal basis for AI training.
AI startup World Labs, founded by the Stanford professor touted as the "godmother of AI," Fei-Fei Li, recently exited stealth with $230 million in funding from backers like Andreessen Horowitz, NEA, and Radical Ventures. The startup aims to develop large world models (LWMs) that interact with 3D environments, looking to transform AI from 2D-pixel understanding to spatially intelligent systems that perceive and create virtual and real worlds. This technology has potential use in gaming, AR/VR, robotics, and more, enabling artists, designers, and engineers to build immersive 3D spaces with physics and control.
4 AI TOOLS
Quist - Quist is an AI-powered research assistant that delivers personalized reports that cater to your unique business needs and frees up researchers and analysts for discovery, innovation, and strategic thinking.
GitChat - GitChat is a real-time chat code assistant that helps developers improve their code, catch bugs faster before hitting production, and make AI-powered PR summaries.
Mapify - Create mind maps out of vast amounts of data in minutes with Mapify, an AI tool designed to turn content into simple and clear mind maps to then build on and refine later.
Coffice - Coffice is designed to help build happier teams through effortless peer-to-peer recognition where users can easily send recognition and celebrate milestones such as birthdays and work anniversaries.
AI READ & WATCH
A Global Contingency Plan for AI Catastrophes (3-min read)
Believing that it has not yet developed the necessary science to control and safeguard such advanced AI, a group of leading AI scientists is emphasizing global oversight to prevent potential AI catastrophes if human control is lost in a call for international cooperation.
Can you Differentiate Between an AI Agent & a Human? (1-min watch)
During the 2024 All-in Summit, Salesforce CEO Marc Benioff said that humanity is already at the level where many people can no longer differentiate a human from an AI agent - and the scope of it at Salesforce is going to scale.