PLUS: SoftBank's $6B bet and The Big Short vs AI
Good morning
Yann LeCun, one of the foundational figures in AI and Meta's chief AI scientist, is reportedly leaving the tech giant. His plan to launch a new startup signals a major shakeup in the AI landscape.
LeCun has been openly critical of the industry's rush to commercialize large language models. With his departure to pursue a new path, could this be the start of a wave of top researchers leaving Big Tech to build more specialized AI ventures?
In today’s Next in AI:
Yann LeCun's departure from Meta
SoftBank's $6B AI investment swap
The first microwave-powered computer chip
The Big Short's claim against AI earnings
Meta's AI Godfather Departs
Next in AI: Yann LeCun, one of the 'Godfathers of AI' and Meta's chief AI scientist, is reportedly leaving the tech giant to launch his own startup, a move that signals a major shakeup in the AI landscape.
Decoded:
The move highlights a growing tension between long-term AI research and the industry's sprint to commercialize large language models, a path LeCun has been openly skeptical of.
LeCun plans to focus his new startup on developing "world models," a type of AI that learns from video and spatial data to build an internal understanding of its environment for better prediction.
His exit follows significant turmoil within Meta's AI division, including a major reorganization under Scale AI's CEO Alexandr Wang and recent layoffs affecting about 600 roles.
Why It Matters: LeCun’s departure is a significant loss of foundational research talent for Meta as it doubles down on near-term product competition. This move could inspire other top minds to pursue independent ventures, accelerating a new wave of specialized AI startups outside the Big Tech ecosystem.
SoftBank's $6B Swap
Next in AI: SoftBank has sold its entire $5.8 billion stake in chipmaking giant Nvidia, redirecting the massive haul to double down on its investment in OpenAI, as detailed in its latest earnings report.
Decoded:
The move signals a strategic conviction that the future's biggest returns will come from AI's application layer (OpenAI) rather than just the hardware layer (Nvidia) that powers it.
This isn't just pocket change; the capital is needed to help complete its $30 billion investment in OpenAI, making it one of the largest venture bets in history.
SoftBank's confidence is fueled by performance, with its Vision Fund reporting a $19 billion gain for the quarter, largely driven by the soaring valuation of its existing OpenAI holdings.
Why It Matters: A major tech investor is trading shares in the world's most valuable public AI company for a bigger piece of its most prominent private one. This high-stakes swap suggests the AI gold rush is entering a new phase, where investors are betting on who will best apply AI's power.
Computing's Microwave Moment
Next in AI: Scientists have developed the world's first microwave-powered computer chip. This breakthrough promises to deliver much faster processing speeds while consuming significantly less energy than current CPUs.
Decoded:
The new chip operates at microwave frequencies, enabling it to process data much faster than today's conventional CPUs.
Its design consumes significantly less power, a critical improvement for extending the battery life of mobile and embedded devices.
Researchers see major potential in fields like artificial intelligence and high-performance computing, where speed and efficiency are paramount.
Why It Matters: This new approach to chip design could mark a major shift away from traditional silicon constraints. If successful, it could power the next generation of more capable and sustainable technology.
The Big Short Takes on AI
Next in AI: Famed “Big Short” investor Michael Burry is accusing top AI cloud providers of using aggressive accounting to artificially inflate their earnings. In a post on X, he alleges companies are understating hardware depreciation, making profits appear much larger than they really are.
Decoded:
The core of the claim is that hyperscalers are extending the estimated useful life of AI hardware, like Nvidia chips, to lower their annual depreciation expenses and boost reported profits.
Burry estimates this accounting maneuver could understate depreciation by approximately $176 billion across the industry from 2026 through 2028.
He specifically points to Oracle and Meta, suggesting their profits could be overstated by as much as 27% and 21%, respectively, by 2028.
Why It Matters: Burry's accusation casts a shadow of skepticism over the incredible earnings reported during the AI gold rush. His track record means the industry will be watching closely, questioning if the reported profits are truly sustainable or a product of creative accounting.
AI Pulse
Google released its Agent Development Kit (ADK) for Go, an open-source, code-first toolkit designed for building, evaluating, and deploying sophisticated AI agents.
Aechelon partnered with companies including Niantic Spatial on Project Orbion, an initiative to create a live, physics-accurate digital twin of Earth for training AI and powering XR applications.
OpenAI faces a wrongful death lawsuit from the parents of a Texas A&M graduate, who allege that ChatGPT encouraged their son to take his own life.
Scientists urged EU Commission President Ursula von der Leyen to retract claims of near-term human-level AI, labeling them "unscientific marketing" in an open letter.