OpenAI Wants Control of Its Compute

Welcome back to The Shift. Let’s get straight to what matters in AI today…

Today we have:

🌶️ OpenAI Reveals Its First Custom AI Chip

🚀How to Deploy Your AI Agent to iMessage, WhatsApp, and Telegram With Spectrum

🫁 AI Leaders Back A $500M Health Initiative

🔨Tools and Shifts you cannot miss

🌶️ OpenAI Reveals Its First Custom AI Chip

OpenAI has pulled back the curtain on Jalapeño, its first in-house AI chip developed with Broadcom. The move marks a major step toward building its own AI infrastructure while reducing long-term dependence on third-party hardware.

The Decode:

Built At Speed - OpenAI says Jalapeño progressed from design to factory-ready in just nine months, with its own AI models helping optimize parts of the chip development process.

Built For Inference - Unlike training chips, Jalapeño is an ASIC designed specifically to run finished models such as ChatGPT, Codex, and future AI agents more efficiently.

Better Efficiency - According to OpenAI, internal testing showed Jalapeño delivers substantially better performance per watt than current state-of-the-art inference hardware, improving efficiency as AI usage continues growing.

Long-Term Infrastructure - While Nvidia remains the backbone for model training, OpenAI plans to scale custom-chip infrastructure to 10 gigawatts of compute capacity by 2029.

Custom silicon is becoming a strategic advantage in AI. By designing chips tailored to its own workloads, OpenAI is aiming to lower costs, improve efficiency, and gain greater control over the infrastructure powering future AI products.

Together with Masterworks

The Last Time Stocks Were This Expensive Was December 1999.

"Right now, it's good. But it was in '72, '86, 2000, and 2007." - Jamie Dimon, May 2026.

The Shiller CAPE ratio just hit 42.3. The only time in 140 years it's been higher? December 1999.

Stocks can stay expensive for a long time...

It’s one metric to consider, but when your portfolio is built around the most expensive equities in modern history, what else you diversify with could really matter.

Blue-chip contemporary and post war art has shown near-zero correlation with the S&P since 1995.* Prices are largely driven by private collectors competing for a fixed supply of artwork by artists like Banksy, Basquiat, and Picasso.

Masterworks lets you invest in shares of that market.

• $1.3B deployed across 500+ artworks

• 29 exits to date

• Net annualized returns like 16.5%, 17.6%, and 17.8%, not including those unsold

My subscribers skip the waitlist here.

^{*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.}

🚀How to Deploy Your AI Agent to iMessage, WhatsApp, and Telegram With Spectrum

Nobody downloads a new app just to talk to an agent. Spectrum puts your agent where people already are. Here's how to connect it.

Step 1: Install Spectrum

npm install spectrum-ts

One command. Works with any existing agent stack.

Step 2: Initialize Spectrum with your providers Configure the platforms you want to reach, iMessage, WhatsApp, Telegram, Slack, Discord, Instagram, in a single config object. One API connects to all of them.

Step 3: Handle incoming messages Use spectrum.onMessage to listen for user messages across every connected platform. Route text, attachments, and reactions through your agent logic in one unified handler.

Step 4: Send responses natively Spectrum adapts your output to each platform automatically. The same message renders as a native iMessage bubble, a WhatsApp reply, or a Telegram card without you handling the formatting differences.

Step 5: Set fallback routing Route to iMessage with Telegram as fallback in one line. Spectrum picks the best available channel per user automatically so messages always get through.

Step 6: Monitor and ship to production Audit logs, message histories, and human-in-the-loop controls are built in. Scale from prototype to millions of users without changing your architecture.

Sub-1 second latency on Photon's edge network. SOC 2 ready. Open source.vYou can get started here.

🫁 AI Leaders Back A $500M Health Initiative

A coalition including Stripe, Anthropic, and the OpenAI Foundation has launched Intercept, a new nonprofit focused on reducing the global burden of everyday respiratory illnesses. Rather than funding finished products, the organization aims to bridge the gap between early research and commercial development.

The Shift:

A $500M Commitment - Intercept has launched with $500 million in funding to accelerate technologies that could reduce the spread of common respiratory illnesses such as colds and seasonal flu.

Prevention First - The nonprofit will support research into vaccines, nasal sprays, antiviral pills, and cleaner indoor air systems designed for schools, workplaces, and other shared environments.

Closing The Funding Gap - Instead of bringing products to market itself, Intercept plans to fund early-stage research until pharmaceutical companies and private investors can take over later development.

A Productivity Problem - Intercept estimates routine respiratory infections cost people 15 to 25 days annually, contributing to roughly $600 billion in lost productivity worldwide.

The initiative reflects growing interest from major technology organizations in funding long-term public health infrastructure. Rather than building AI products, some industry leaders are now investing in scientific research that could reduce widespread illness and its economic impact.

Together with Viktor

A $3K spend spike got caught at 2am.

Viktor watches your ad accounts overnight. When CPA jumps 340% on a broad match campaign, he posts in #growth with a recommendation and pauses pending your approval. Across Google, Meta, TikTok, and LinkedIn at once.

Get Started for Free

🔨AI Tools for the Shift

🚀 JustVibe – Turns simple descriptions into fully functional apps, helping you bring ideas to life without coding.

📱 NiuNiu – Builds Android apps from plain-language prompts and delivers installable APKs ready to use on your phone.

📧 Reply – Automates lead generation, multichannel outreach, follow-ups, and AI-powered prospecting to book more meetings.

🛠️ Whacka – Helps you transform app ideas into real mobile products through an AI-powered development workflow.

🎯 AdControlCenter – Creates high-performing advertising campaigns quickly while helping businesses reduce wasted ad spend.

🚀Quick Shifts

🎥 Meta is reviving Facebook Creator Studio as an AI-powered companion app, offering creators personalized growth insights, performance tracking, AI-assisted comment management, and tailored recommendations to help expand their audience and engagement.  

⚖️ Nearly 400 local newspapers sued OpenAI and Microsoft, alleging the companies used their journalism without permission to train AI models, expanding the growing wave of copyright lawsuits from publishers against AI developers. 

📰 People Inc. CEO Neil Vogel signaled a likely clash with Google over AI crawling, saying the publisher can't block Google's crawler without hurting Search, despite licensing content to other AI companies like OpenAI, Meta, and Microsoft. 

🤖 Google DeepMind integrated computer-use capabilities into Gemini 3.5 Flash, enabling the model to power AI agents that can navigate browsers, desktop apps, and user interfaces to complete multi-step tasks autonomously. 

🧪 Former Anthropic, OpenAI, Google, and xAI researchers launched Mirendil, a new AI lab developing self-improving systems that can autonomously conduct research and accelerate scientific discovery through AI-driven R&D. 

🧩 Prompt of the Day

LinkedIn Connection Requests That Start Real Conversations

Most LinkedIn connection requests are generic or immediately ask for something. The best messages create relevance, show genuine interest, and make it easy for the other person to accept.

Build relationships before making requests.

Paste the prompt: Drop this into ChatGPT, then fill in your networking details.

Prompt to paste

Write 5 LinkedIn connection request messages based on my [Insert role], [Insert target company or person type], and [Insert reason for connecting]. Create one message for each scenario: cold outreach to a recruiter, connecting with a potential mentor, following up after meeting at an event, reaching out to a hiring manager, and a warm follow-up with an existing contact. Keep each note concise, personalized, and natural while avoiding generic networking language or transactional requests. Focus on building genuine professional relationships and increasing connection acceptance rates.

🤳AI Shift of the Day

That’s all for today’s edition. See you tomorrow as we track down and get you all that matters in the daily AI Shift!

If you loved this edition, let us know how much:

Forward it to your pal to give them a daily dose of the shift so they can 👇