Good morning, AI enthusiast.

OpenAI is strategically restructuring its leadership to accelerate both current applications and its long-term superintelligence goals. The appointment of Instacart's CEO to head a new Applications division allows Sam Altman to deepen his focus on pioneering research and the formidable challenge of developing "superintelligence."

With Altman concentrating on foundational research and the path to "superintelligence," what new breakthroughs might emerge, and how will the company navigate the complex safety considerations that come with such advanced AI?

In today’s AI recap:

• OpenAI’s leadership shakeup & superintelligence drive

• AI search tools chip away at Google's Safari traffic

• Apple’s custom silicon push for AI hardware

• Google’s AI co-scientist speeds scientific breakthroughs

OpenAI Shakes Things Up: New Applications CEO, Altman Eyes Superintelligence

The Recap: OpenAI is making significant leadership moves: Sam Altman announced that Instacart CEO Fidji Simo joins as its first CEO of Applications. Altman, remaining OpenAI CEO, will sharpen his focus on research, compute, and safety, crucial as the company approaches "superintelligence."

Unpacked:

• Fidji Simo steps into the newly created CEO of Applications role, reporting directly to Sam Altman, a move signaling OpenAI's increased commitment to productizing its advanced AI models.

• Sam Altman, while remaining OpenAI's CEO, will dedicate more energy to foundational research, scaling compute power, and ensuring AI safety as the company ventures towards what he terms "superintelligence."

• Simo, who will transition from her Instacart CEO role over the next few, expressed immense excitement about contributing to OpenAI's ambitious mission.

• OpenAI President Greg Brockman also welcomed Simo, highlighting her unique insights and his excitement to collaborate on delivering AGI that benefits all humanity.

Bottom line: This restructuring at OpenAI underscores a dual strategy: rapidly translating cutting-edge research into real-world applications while simultaneously pushing the frontiers of AI development towards superintelligence. For professionals, this means anticipating even more powerful and accessible AI tools, alongside a heightened focus on the responsible advancement of increasingly capable AI systems.

The Search Wars Heat Up: AI Challenges Google, Apple Explores New Paths

The Recap: Testimony from Apple executive Eddy Cue, revealing that AI search tools like ChatGPT and Perplexity are causing the first-ever dip in Google search volume on Safari, has shaken Google's perceived dominance.

Unpacked:

• Cue revealed Apple is actively exploring AI-driven search options like Perplexity, potentially disrupting Google's lucrative $20 billion annual deal to be Safari's default search.

• The revelation initially sent Alphabet's stock tumbling, wiping over $150 billion in market value before a partial recovery, signaling significant investor concern.

• Google countered swiftly in a blog post, asserting that overall search queries, including those from Apple devices, continue to grow and highlighting ongoing AI-driven enhancements to its search products.

• This entire saga unfolds as Google defends its market position in a high-stakes antitrust trial with the U.S. Department of Justice, adding another layer of complexity to its future.

Bottom line: The search landscape is undeniably shifting, with AI emerging as a formidable new player. While Google is a titan, Apple's moves and the rapid evolution of AI search tools promise more innovation and choice for users. For professionals like you, this signals exciting advancements in how we access information and could usher in a new era of search efficiency and capabilities.

Apple's AI Silicon Push: Custom Chips for Glasses, Servers, and Future Macs

The Recap: Apple is reportedly accelerating development of specialized custom chips for its ambitious AI future, spanning smart glasses, powerful server infrastructure, and next-gen Macs. This significant move aims to bolster Apple's control over its hardware and software integration for upcoming AI-driven products.

Unpacked:

• Its upcoming smart glasses will likely feature a chip that is a derivative of its Apple Watch silicon, optimized for power efficiency and multi-camera control, with mass production potentially starting by late 2026.

• Apple is also engineering advanced AI server chips, reportedly codenamed "Sotra" and "Baltra" (a potential Broadcom collaboration), to run Apple Intelligence and enhance its cloud-based AI capabilities.

• Future Macs are set to receive a boost with the development of next-generation processors, provisionally dubbed the M6 ("Komodo") and M7 ("Borneo").

Bottom line: This deep investment in custom silicon highlights Apple's strategy to vertically integrate and meticulously control the hardware powering its ecosystem. Tailoring chips for specific AI tasks enables Apple to optimize performance and efficiency, paving the way for innovative user experiences across new and existing product lines.

Google's AI Co-Pilot for Science: From Brain Maps to Drug Discovery

The Recap: Google Research deploys advanced AI to accelerate scientific breakthroughs across multiple disciplines. These AI tools help researchers tackle complex problems, from mapping brain connections to discovering new uses for existing drugs.

Unpacked:

• An AI co-scientist from Google helped identify Vorinostat, an existing cancer drug, as a potential treatment for liver fibrosis by synthesizing information and proposing novel research avenues.

• In neuroscience, Google introduced LICONN, a novel method enabling scientists to map neural connections in brain tissue using common light microscopes, making connectomics more accessible.

• The AI co-scientist itself is a multi-agent system, leveraging specialized AI agents to iteratively generate, evaluate, and refine hypotheses based on researchers' natural language inputs.

• Google invites research organizations to further explore these capabilities through its Trusted Tester Program for the AI co-scientist.

Bottom line: Google positions AI not just as an analytical tool, but as a collaborative partner in the scientific process. This approach could significantly shorten discovery cycles and bring innovative solutions to challenging medical and scientific problems much faster. We're excited to see these tools empower the next wave of research.

The Shortlist

Stripe launched a Payments Foundation Model, an AI trained on billions of transactions to boost fraud detection and authorization rates, alongside new stablecoin financial accounts for businesses.

Microsoft embraced Google's Agent2Agent (A2A) open protocol, enabling AI agents built with Azure AI Foundry and Copilot Studio to communicate and collaborate with agents across different platforms and ecosystems.

An Arizona court witnessed a believed U.S. first as the family of Christopher Pelkey, a road rage victim, used AI to generate a video of him delivering his own victim impact statement at his killer's sentencing.

Meta teamed up with NVIDIA to integrate NVIDIA cuVS into Faiss, Meta’s open-source library, significantly boosting performance for similarity search and vector indexing on GPUs.

Signing off,

David, Lucas, Mitchell — The Recap editorial team