🎉 OpenAI Builds Hardware Team, xAI Launches Grok 4 Fast, Google Introduces AP2 Framework, Gemini Arrives in Chrome, Stanford Creates AI-Designed Viruses
Apple Alumni, Assemble!, Cheap, Fast, Grok’d, AI Buys Your Stuff, The Browser Got Brains, Evo Plays God Mode
Welcome to this week’s edition of AImpulse, a five point summary of the most significant advancements in the world of Artificial Intelligence.
Here’s the pulse on this week’s top stories:
1. OpenAI Raids Apple for Hardware Talent
OpenAI is going all-in on hardware, poaching Apple veterans and locking in iPhone suppliers for its upcoming device lineup, according to The Information.
What’s happening:
Dozens of Apple engineers have joined, lured by $1M+ comp packages.
Tang Tan, ex-Apple exec, is leading hardware and pitching less bureaucracy + big vision.
Manufacturing deals are in place with Luxshare + Goertek (iPhone makers).
First products could include a smart speaker-like device, glasses, a pin, or a recorder, targeting late 2026 / early 2027.
Why it matters: This is the Jony Ive playbook all over again, design veterans, secrecy, hype. With OAI’s ambitions and Apple DNA, the debut could be one of the most anticipated product launches in years.
Further reading: https://www.theinformation.com/articles/openai-raids-apple-hardware-talent-manufacturing-partners
2. xAI Unveils Grok 4 Fast
Elon Musk’s xAI launched Grok 4 Fast, a hyper-efficient reasoning model that matches frontier performance while cutting costs by nearly 100x.
The highlights:
Uses 40% fewer tokens → 98% cheaper than Grok 4.
Scores: 85.7% on GPQA Diamond (science), 92% on AIME 2025 (math).
Outperforms Claude 4.1 Opus + Gemini 2.5 Pro; tops LMArena’s Search Arena.
Supports 2M context tokens + built-in tools for browsing + code execution.
Why it matters: Cost-efficiency like this makes AI “intelligence too cheap to meter” feel closer than ever. Grok 4 Fast shows that the frontier race isn’t just about accuracy, it’s about delivering that power at scale.
Further reading: https://www.engadget.com/ai/xai-debuts-a-faster-and-more-cost-effective-version-of-grok-4-192815570.html
3. Google Launches Agent Payments Protocol
Google announced Agent Payments Protocol (AP2), an open framework that lets AI agents make secure purchases with big financial + tech names backing it.
The details:
Uses digital contracts called “mandates” for user authorization.
Requires dual approvals: Intent (search) + Cart (payment).
Supports cards, bank transfers, stablecoins (with Coinbase).
Partners include AmEx, Mastercard, PayPal, Salesforce, Intuit; specs on GitHub.
Why it matters: AI agents as autonomous shoppers is here. AP2 offers the accountability layer needed for trust, and with industry adoption already deep, this could become the default rail for AI commerce.
Further reading: https://cloud.google.com/blog/products/ai-machine-learning/announcing-agents-to-payments-ap2-protocol
4. Google Embeds Gemini in Chrome
Google is rolling out Gemini integration into Chrome for all U.S. desktop users, adding AI features directly into the world’s most popular browser.
What’s new:
Dedicated Gemini button to analyze across tabs.
Address bar gets AI Mode later this month for conversational queries.
Previewed agentic task automation, think grocery shopping or booking an appointment inside Chrome.
Why it matters: Gemini in Chrome makes the AI assistant as native as the URL bar. While others built entirely new browsers, Google just infused the one billions already use, setting up the largest mainstream AI browsing rollout yet.
Further reading: https://www.reuters.com/sustainability/boards-policy-regulation/google-adds-gemini-chrome-browser-after-avoiding-antitrust-breakup-2025-09-18/
5. Stanford & Arc Build First AI-Designed Viruses
Stanford and the Arc Institute have achieved a world-first in computational biology: AI-generated viruses that successfully infect and kill bacteria.
The breakthrough:
Researchers trained an AI model called Evo on 2M viruses.
Out of 302 designs, 16 worked in lab tests.
The AI created 392 mutations never seen in nature, including combos scientists had failed to engineer manually.
AI viruses beat bacterial resistance where natural viruses failed, breaking through defenses in just days.
One design incorporated a component from a distantly related virus, a goal researchers had tried and failed to achieve for years.
Why it matters: This marks the start of a new chapter in biology: moving from reading and editing genomes to designing them from scratch. AI is now not just accelerating discovery, but creating entirely new building blocks of life.
Take a look: https://www.nature.com/articles/d41586-025-03055-y