OpenAI Links Up With Broadcom to Produce Its Own AI Chips 🤝

Welcome to another edition of Horizon AI,

OpenAI just forged a multibillion-dollar deal to produce its own AI chips, aiming to address its insatiable demand for computing power and reduce its reliance on chip giant Nvidia.

Let’s get into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• OpenAI Teams With Broadcom to Build Custom AI Chips

• Microsoft Debuts MAI-Image-1, Its First In-House Text-to-Image Model

• AI Tutorial: How to edit images with Nano Banana in Search

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

OPENAI

OpenAI Teams With Broadcom to Build Custom AI Chips

OpenAI and Broadcom are working together to develop and deploy 10 gigawatts of custom AI chips and computing systems over the next four years.

Details:

• Broadcom will begin deploying those racks in the second half of next year, with the rollout expected to be completed by the end of 2029.

• OpenAI claims that designing its own chips will allow it to integrate what it has learned from developing powerful AI models and products directly into the hardware, which it says will “unlock new levels of capability and intelligence.”

• The agreement is reportedly worth several billion dollars, although neither company has disclosed specific financial terms.

The deal is crucial for OpenAI, which requires vast amounts of computing power to run apps like ChatGPT and Sora and to pursue its goal of developing superintelligent AI. Competitors such as Meta, Google, and Microsoft are also developing custom chips to better support their AI ambitions while reducing their reliance on Nvidia’s chips.

Source

TOGETHER WITH TYPELESS

Typing is a thing of the past

Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.

It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.

With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.

Your voice is your strength. Typeless turns it into a superpower.

Download for Mac today

MICROSOFT

Microsoft Debuts MAI-Image-1, Its First In-House Text-to-Image Model

The company says its text-to-image generator delivers faster outputs than larger models and performs especially well at producing photorealistic images, capturing details such as lighting effects and landscapes.

Details:

• Microsoft says it worked with creative professionals who helped shape the model to reduce “repetitive or generically stylized outputs.”

• Early results are promising, with MAI-Image-1 ranking among the top 10 text-to-image models on LMArena, where users can test the model and compare results with others.

• The model will soon be integrated into Copilot and Bing Image Creator, and Microsoft emphasizes that it’s committed to “ensuring safe and responsible outcomes.”

Microsoft is making significant investments in training its own AI models, with this latest addition joining a growing internal lineup that also includes the voice generator MAI-Voice-1 AI and the chatbot MAI-1-preview.

Source

AI Tutorial

How to edit images with Nano Banana in Search

Google is rolling out its Nano Banana model in Search, allowing you to transform photos using Lens and AI Mode.

1. Open Lens in the Google app on Android or iOS.

2. Tap the new Create mode — look for the yellow banana

3. Pick one of Google’s suggested prompts, like “make a photo booth pic of me.”

4. Or, take a photo and describe the edits you want.

5. Use follow-ups to keep editing, or share your creation with family and friends.

*Notes:

• This feature is available in English in the U.S. and India, with more countries and languages coming soon.

• Those who prefer not to take selfies can use the rear-facing camera or select an existing photo from the gallery to visualize ideas, such as a Halloween costume for a pet.

• You can now also create entirely new images from scratch directly in AI Mode. Just select the “Create image” tool and add a text prompt.

AI Tools to check out

📽 Munch: Extract the most engaging and impactful clips from your long-form videos.

👥 Spiritme: AI-video platform for creating personalized videos with digital avatars.

📰 ModelMatch: Compare top open-source models for image understanding side-by-side, no coding needed.

🦾 Sintra: Build, grow, and scale your business with a team of AI employees.

💥 Chatbox AI: An AI client application and smart assistant.

AI Findings/Resources

💥 I tested OpenAI Sora 2 vs Google Veo 3 for AI video, and there’s a very clear winner

🤔 Paraguay – the Silicon Valley of South America?

👀 Learn how to fine-tune Gemma 3 270M and run it on-device

🥋 Unitree’s humanoid robot wows with impressive kung fu moves

⭐ How to make sense of Meta’s growing AI-powered advertising machine

TOGETHER WITH INTERCOM

Startups get Intercom 90% off and Fin AI agent free for 1 year

Join Intercom’s Startup Program to receive a 90% discount, plus Fin free for 1 year.

Get a direct line to your customers with the only complete AI-first customer service solution.

It’s like having a full-time human support agent free for an entire year.

Apply now

The latest in AI and Tech

Mark Cuban uses AI videos on Sora for free advertising

The billionaire investor and “Shark Tank” star is using OpenAI’s social media platform Sora to promote his company, Cost Plus Drugs, by letting users create AI-generated videos featuring him, each with a baked-in plugin ad for his company.

Cuban said he launched the experiment to see what people would create while subtly advertising his business, adding that “so far it’s worked out great.”

OpenAI says GPT-5 shows 30 percent less political bias than earlier models

OpenAI has released a new internal study claiming that GPT-5 produces about 30% less political bias than previous versions, based on evaluations across 500 prompts covering 100 political and cultural topics.

Apple nears deal to acquire computer vision startup Prompt AI

The company is close to acquiring the team and technology from computer vision startup Prompt AI, founded by UC Berkeley researchers Tete Xiao and Trevor Darrell.

The company’s app, Seemour, uses AI to identify people, pets, and objects in home security footage. While the technology worked well, executives said the business model didn’t, leading Apple to step in.

Gemini can now help schedule meetings in Gmail

Google is rolling out a new Gemini-powered “Help me schedule” feature in Gmail that suggests meeting times using your email context and Google Calendar.

When composing an email, Gemini can detect scheduling intent and display available time slots you can insert directly into your message. Recipients can then pick a time, automatically creating a Calendar invite for both parties.

Oracle to deploy 50,000 AMD AI chips

Oracle plans to offer cloud services powered by AMD’s new Instinct M1450 Series GPUs starting next year, with a broader rollout in 2027 and beyond.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻