Good morning. It's Tuesday, January 20, and we're covering AI in medicine, rogue agents, Meta’s metaverse shutdown, and more. First time reading? Join a community of 600k+ people keeping up with the future of tech. Sign up here.

Don’t keep us a secret: Share this email with a friend. And as always, if you have feedback, we want to hear it: [email protected]

🪦 Meta Scales Back the Metaverse
Meta is dramatically winding down its metaverse ambitions after Reality Labs racked up nearly $73 billion in losses since 2021. The company is shuttering its Horizon Workrooms VR app by Feb 16 (with user data deleted) and has cut ~10% of Reality Labs staff (≈1,000 jobs). CEO Mark Zuckerberg now plans to redirect investment to AI wearables, AR glasses, and mobile features, signaling a strategic retreat from fully immersive VR.

👷‍♂️ AI Brings Labor Shortage, Not Job Loss
Jonathan Ross, founder and CEO of Groq, argues AI will create a labor shortage rather than mass unemployment. He says automation will drive deflation, lowering costs for goods like food and housing. As living gets cheaper, people may work fewer hours and retire earlier, reducing labor supply. At the same time, AI will enable entirely new jobs and industries that don’t yet exist.

🛡️ Privacy-First AI Challenges ChatGPT
Signal co-founder Moxie Marlinspike launched Confer, a chatbot designed to protect user privacy by preventing data collection entirely. Conversations are encrypted end-to-end and processed inside trusted execution environments, so even the host can’t access them. Confer avoids ad targeting and model training on user data, positioning itself as an alternative to ChatGPT or Claude.

💰 OpenAI Revenue Tops $20B Run Rate
OpenAI CFO Sarah Friar revealed the company’s annualized revenue run rate exceeded $20 billion in 2025 (up from ~$6 billion in 2024). This surge mirrors a jump in capacity – OpenAI’s compute power tripled to 1.9 GW last year. Friar noted ChatGPT usage hit record highs and that OpenAI has started testing ads to help fund escalating costs.

Enterprises sit on a gold mine of data locked inside their content, but manual processes have made it nearly impossible to unlock. Introducing Box Extract: an agentic solution for enterprises looking to securely and accurately pull out massive amounts of valuable data from their content.

Powered by leading AI models from Google, Anthropic and OpenAI, Box Extract accelerates workflow automation, powers faster content discovery, and improves decision-making.

Learn more about how Box Extract can transform enterprise content into actionable data, now available.

MIT introduces recursive language models that unlock near-infinite context windows, letting AI search massive prompts without loss.

Dr. Robert Wachter, chair of medicine at the University of California, San Francisco, argues in a New York Times opinion essay that health care should adopt AI now, despite its flaws, because it already outperforms many parts of today’s system. Drawing on his own practice, he describes using tools like ChatGPT and OpenEvidence as modern “curbside consults” that deliver fast, generally reliable guidance when used with proper review. Wachter’s core point is that AI should be judged against the reality of an inefficient, costly, and error-prone health care system, not against an unrealistic standard of perfection.

He points to practical benefits already in use, such as AI scribes and record-summarization tools that reduce paperwork and ease clinician burnout. While supporting strict oversight for autonomous, high-risk medical AI, Wachter warns that waiting for flawless technology will only lock in existing inefficiencies and costs. Framed this way, the debate shifts from perfection to progress—and raises a key question for regulators and patients alike: how quickly will they accept “better than today” as good enough? → Read the full article here. (Paywall)

Venture capitalists are ramping up investment in AI security as enterprises deploy autonomous agents that can behave unpredictably. In a recent example cited by Ballistic Ventures partner Barmak Meftah, an enterprise AI agent allegedly scanned an employee’s inbox and threatened blackmail after its objectives were blocked.

Startups like Witness AI aim to address these risks by monitoring AI use, detecting shadow AI tools, and enforcing governance across organizations. The push reflects growing concern that misaligned, non-deterministic agents—and AI-driven attacks at machine speed—are outpacing traditional security controls. → Read the full article here.

🚘 Tesla Revives Dojo Supercomputer: Elon Musk says work resumes on Dojo3 after a pause to focus on in-car AI chips. Goal: train Full Self-Driving, though space-based plans remain speculative.

📈 Moonshot AI Valuation Jumps: Alibaba-backed startup hits $4.8B, up $500M in weeks, as rival Chinese AI firms surge after Hong Kong IPOs, boosting investor appetite.

⚡ EPA Closes xAI Power Loophole: Federal regulators say xAI’s off-grid gas turbines at its Memphis Colossus data center must meet Clean Air Act rules, tightening oversight of AI energy infrastructure.

🚀 Gemini Demand Surges (Paywall): Google says developer requests for its Gemini AI more than doubled in five months, signaling rising enterprise interest beyond OpenAI models.

🌌 Musk Bets on Space AI: SpaceX plans orbiting, solar-powered data centers using Starship, betting off-planet computing can bypass Earth’s grid limits and political friction.

❤️ Love Forward Future? Share your unique referral link with friends and colleagues to unlock exclusive Forward Future perks! 👉 Get your link here.
📢 Want to advertise with Forward Future? Reach 600K+ AI enthusiasts, tech leaders, and decision-makers. Let’s talk—just reply to this email.

Thanks for reading today’s newsletter—see you next time!
Matthew Berman & The Forward Future Team

🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀