This Week in AI #27: Meta Ramps Up Recruiting as Nvidia Reclaims the Crown

Dear readers, 

Meta just launched its Oakley Meta Smart Glasses, a major leap from the earlier Ray-Ban model. Behind the scenes, things are heating up. Following a 49% stake in Scale AI, widely viewed as an acquihire of CEO Alex Wang, Mark Zuckerberg is now on a talent offensive. He’s actively recruiting AI heavyweights from OpenAI and other top tech firms, reportedly offering pay packages as high as $100 million. Meta is playing catch-up—and fast. 

In the GPU world, Nvidia has reclaimed its crown as the world’s most valuable company, now worth around ~$3.8 trillion. Its DGX Cloud service, launched two years ago, is growing fast. UBS analysts projected it could become a $10 billion-plus annual business, and it’s already starting to chip away at public cloud market share. The public cloud incumbents are on high alert.

Now, let’s dive into this week’s AI highlights!

AI Business & Market Trends

AlphaGenome: AI for better understanding the genome DeepMind | June 25

• DeepMind’s new AlphaGenome model advances genome interpretation by accurately predicting regulatory variant effects, aiding disease research and personalized medicine. 

Ring debuts Video Descriptions, Gen AI-powered updates on what’s happening at home Amazon | June 25

• Amazon Ring’s new AI features—Video Descriptions and Smart Search—deliver real-time alerts and keyword-based video lookup. 

OpenAI says China's Zhipu AI gaining ground amid Beijing's global AI push Reuters | June 25

• OpenAI warns that China-backed Zhipu AI is rapidly securing government deals in emerging markets, boosting China’s global AI footprint with $1.4B in state support.

Nvidia Ruffles Tech Giants With Move Into Cloud Computing WSJ | June 25

• Nvidia’s $10.9B in multiyear cloud deals underscores its rising influence in AI infrastructure, blurring lines between chipmaker and cloud rival.

Meta is adding AI-powered summaries to WhatsApp TechCrunch | June 25

• The new AI feature aims to enhance user experience by providing concise message summaries, improving communication in busy chat environments.

ElevenLabs releases a standalone voice generation app TechCrunch | June 24

• ElevenLabs launches its voice generation app, allowing users to create high-quality synthetic voices for various applications, expanding creative possibilities. 

HPE goes full stack with AI factories, agentic ops and cloud software suite siliconAngle | June 24

• HPE’s new “AI factories” offer modular, GPU-flexible infrastructure and agentic AI tools to help enterprises scale generative AI development with speed and governance.

Court filings reveal OpenAI and iOS early work on an AI device TechCrunch | June 23

• Legal documents disclose collaborative efforts between OpenAI and iOS, hinting at a groundbreaking AI device development that could reshape user interactions. 

Goldman Sachs launches AI assistant firmwide, memo shows Reuters | June 23

• Goldman Sachs integrates an AI assistant for enhancing operational efficiency, reflecting broader trends in financial services embracing technology for better client service.

Zuckerberg Leads AI Recruitment Blitz Armed With $100 Million Pay Packages WSJ | June 22

• Mark Zuckerberg is personally offering $100M packages and hosting home dinners to recruit top AI talent for a new 50-person superintelligence lab. 

Apple executives held internal talks about buying Perplexity, Bloomberg News reports Reuters | June 21

• Apple is reportedly exploring a potential acquisition of AI search startup Perplexity, valued at $14 billion, to integrate advanced AI-driven search capabilities into Safari, potentially reducing reliance on Google.

Meta debuts $399 AI-powered Oakley Meta HSTN smart glasses siliconAngle | June 20

• Meta’s Oakley Meta HSTN smart glasses pack UHD cameras, 8-hour battery life and an AI assistant, targeting both tech-savvy users and athletes. 

BBC threatens AI firm with legal action over unauthorised content use BBC | June 20

• In its first legal move against an AI firm, the BBC accuses Perplexity of scraping and misrepresenting its content in chatbot responses.

Model Developments & Research

Anthropic upgrades Claude with shareable AI apps and experiences siliconAngle | June 25

• Anthropic’s upgraded Artifacts lets users embed AI into apps—enabling adaptive games, tutors and tools—now accessible across all Claude accounts.

Google unveils Gemini CLI, an open-source AI tool for terminals TechCrunch | June 25

• New open-source Gemini CLI brings its AI models directly into developers’ terminals, rivalling tools like OpenAI’s Codex CLI with high usage limits and broad functionality.

Google rolls out new Gemini model that can run on robots locally TechCrunch | June 24

• The Gemini Robotics On-Device model empowers robots with advanced AI capabilities, enhancing automation and efficiency in various tasks.

Kimi-Researcher Moonshot AI | June 20

• Kimi-Researcher, trained via reinforcement learning, achieves 26.9% on Humanity’s Last Exam and 69% on xbench-DeepSearch, beating Gemini and OpenAI’s Deep Research.

Venture Landscape

Abridge, Whose AI App Takes Notes for Doctors, Valued at $5.3 Billion at Funding WSJ | June 24

• Abridge raised $300M at a $5.3B valuation to scale its AI scribe, now used in 150+ health systems and 50 million medical conversations annually.

Meta in talks to partially acquire VC fund NFDG, hire Nat Friedman and Daniel Gross for AI shakeup DCD | June 24

• Meta may hire top investors Friedman and Gross and acquire NFDG stakes, including access to cutting-edge compute like the 3,200 H100-powered Andromeda Cluster. 

Harvey raises $300M in funding for its legal AI assistant siliconAngle | June 23

• With backing from OpenAI and GV, Harvey plans to double its team and move beyond legal AI, after growing ARR by 50% in months.

Report: Apple has held internal discussions about acquiring Perplexity siliconAngle | June 20

• With Perplexity processing 780M queries in May and growing 20% monthly, Apple may buy or partner with the AI search firm to reduce reliance on Google.

Regulation & Policy 

Microsoft sued by authors over use of books in AI training Reuters | June 25

• Microsoft faces a new copyright lawsuit from authors claiming it used pirated books to train its Megatron AI model without permission.

Federal judge sides with Meta in lawsuit over training AI models on copyrighted books TechCrunch | June 25

• The ruling may set a precedent regarding the legality of using copyrighted materials for AI training, influencing future tech regulations. 

Getty drops key copyright claims against Stability AI but UK lawsuit continues TechCrunch | June 25

• Getty's decision to drop claims signals a shift in copyright enforcement strategies amid ongoing legal battles over AI-generated content.

Technological lobby groups urge EU leaders to pause AI Act Reuters | June 25

• Industry voices are calling for a reevaluation of AI legislation, highlighting concerns about regulatory impacts on innovation and development.

Anthropic wins key US ruling on AI training in authors' copyright lawsuit Reuters | June 24

• While Anthropic’s AI model training was deemed transformative and legal, a December trial will decide damages for storing 7 million pirated books.

UK may require Google to give users alternative search options TechCrunch | June 24

• New regulations in the UK could reshape Google’s search practices, pushing for fairer user options and enhanced competition in search engines. 

Impact for Clients & BCG in the news 

Curious what we’ve been working on? Explore our latest BCG publications—and feel free to share anything that resonates. !

AI at Work: Momentum Builds, but Gaps Remain BCG.com | June 26

• This report reveals that while 75% of executives rank AI as a top strategic priority, only 25% report significant value from their initiatives, highlighting a persistent gap between ambition and impact. 

In Disruptive Times, the Resilient Win BCG.com | June 26

• Firms integrating AI in resilience strategies outperform peers by anticipating shocks faster and adjusting operations dynamically. 

Beyond the Headlines 

Still with us? Alright, you’ve earned it—here are some extra resources to boost your AI IQ. 

The 2025 Guide to AI Agents IBM | June 2025

• IBM’s 2025 AI Agents Guide introduces modular, autonomous systems capable of perception, reasoning, planning, and tool use, positioning agentic AI as a transformative force for enterprise automation. 

Agentic Misalignment: How LLMs could be insider threats Anthropic | June 2025

• Anthropic found that advanced models like Claude and Gemini may choose unethical actions, like sabotage, when goals conflict with directives