The New Open-Source LLM Everyone’s Talking About...
$500B AI Bet | Google’s Big Move in AI | DeepSeek: China’s Quiet Rival
Welcome to the 108 new Autopreneurs who subscribed since last Friday!
If you haven’t subscribed, join 2033 people reading about the latest in AI & Automation:
Interested in getting your brand or business seen by thousands of readers every month?
Check out these sponsorship opportunities here.
It’s Friday, January 24th—time to catch up on some good AI stories from this week (and maybe your next big opportunity).
This week: DeepSeek’s open-source reasoning LLM is redefining what’s possible with AI—offering tools that tackle math, coding, and workflows without breaking the bank. Meanwhile, a $500 billion AI mega-project kicks off in the U.S., Anthropic secures (another) billion-dollar boost, and OpenAI’s new web-browsing agent promises to handle the tasks you don’t want to.
Here’s what’s in today’s issue:
Stargate: The $500 billion plan to dominate AI.
Anthropic’s automation-heavy roadmap for 2025.
DeepSeek’s reasoning models (and why they matter to you).
Perplexity Sonar Pro API: Real-time accuracy redefined.
OpenAI’s Operator: AI that clicks, types, and scrolls.
Let’s dive in. And don’t keep this to yourself—share it with a friend who needs to be in the know!
As always, reply with your thoughts—I’m at leo@theautopreneur.co.
Trump + OpenAI + Oracle + SoftBank = Stargate Project
Donald Trump announced the project with those happy guys on the back (the big players: OpenAI, Oracle and Softbank).
In short: $500 billion is being poured into AI and infrastructure across the U.S.
What do you think $500 billion can buy? Your GI Joe collection, for sure—but in this case, the plan is to build cutting-edge AI infrastructure through twenty massive data centers across the U.S., creating 100,000 jobs and solidifying American dominance in AI.
The project, already under construction in Texas, promises game-changing capabilities for AI applications, from health tech to AGI development. OpenAI, SoftBank, and Oracle have committed $100 billion for immediate deployment, while the remaining $400 billion will be paid in installments over the next 179 years… Just kidding! But Elon Musk isn’t. He’s casting doubt on whether the remaining cash actually exists.
Sam Altman clapped back, calling the initiative “great for the country” and inviting Musk to visit the construction site. And just like that, the drama begins.
Here’s the bottom line: With 100,000 jobs on the horizon and a focus on data centers, this project signals a massive opportunity for AI tools, apps, and services that will leverage this infrastructure. If you’ve been waiting for a sign to dive into The Automation Lab community and learn about AI, automation, and entrepreneurship, this is it.
Need another reason? Keep reading.
Google invests $1bn in AI startup Anthropic
I know, I know—$500 billion to just $1 billion? But stay with me, because this might be more relevant to you than you think…
Anthropic just landed another $1 billion from Google, bumping the tech giant’s total investment in the AI startup to $3 billion.
While Google is doubling down, Anthropic is also in talks to secure another $2 billion from investors like Lightspeed Venture Partners, pushing its valuation to a staggering $60 billion. This move places Anthropic firmly in the AI arms race, alongside heavyweights like OpenAI and Amazon.
But here’s why you, my fellow autopreneur, should care: Anthropic’s roadmap for 2025 includes the game-changing "Virtual Collaborator" system. This tool promises to execute workflows, write and verify code, and integrate effortlessly into your favorite apps like Slack and Google Docs.
Translation? Automation, automation, and even more automation—which means more opportunities for you to make money.
Talking about competitors…
Have you tried DeepSeek yet?
Build a Perplexity clone in 90 minutes
If you’re in the AI space and haven’t tested DeepSeek yet, here’s your wake-up call. DeepSeek, a Chinese AI lab making waves with its cutting-edge reasoning models, just launched DeepSeek-R1, an open-source powerhouse designed for complex problem-solving like math, coding, and step-by-step reasoning. You wouldn’t mind having another weapon in your arsenal, would you?
If you are skeptical like Elon Musk, don’t take my word for it—check out this post on X of a Perplexity clone built with the DeepSeek reasoner and Cursor, without a single line of code. It’s open-sourced, easy to replicate, and just might be your inspiration for next automation project.
And since we talked about Perplexity…
Perplexity’s Sonar Pro API Is Here
Don’t get left behind, read it here
Perplexity’s Sonar Pro API combines LLMs with real-time web-wide research for unparalleled accuracy. Developers now have access to cutting-edge features like customizable sources, JSON mode, and search domain filters. Sonar Pro delivers in-depth, multi-step queries with double the citations and a larger context window, making it ideal for enterprises needing precise, nuanced answers.
Examples of Sonar in action? Companies like Zoom use it for live Q&A during video calls, while Doximity powers research-backed tools for doctors. Its factuality score leads the industry, ensuring trust and reliability.
Now that I think about it, that Perplexity clone doesn’t seem so cool after all, but…
Have you heard about Operator?
It’s OpenAI’s new AI tool. Think of it as an assistant that doesn’t just generate text—it actually interacts with websites like a human would.
It can click, type, scroll, and handle tasks on its own, like booking trips, filling out forms, or even managing repetitive workflows. All you do is describe what you need, and it handles the rest, handing control back to you only when necessary, like entering passwords or confirming a payment.
What’s cool is that it’s powered by this model called CUA (Computer-Using Agent), which allows it to “see” the web through screenshots and interact with interfaces in real-time. And OpenAI has built a ton of safety features into it, so it won’t just take random actions without your approval. Right now, it’s available for Pro users in the U.S., and they’re refining it based on real-world feedback.
But here’s why this matters: the idea is to actually do things for you on the web. And soon, developers will be able to use its model to create their own agents for all kinds of applications.
That’s all for our edition of the Friday AI Forecast.
We publish articles 3x per week:
…teaching you how to use AI inside your Business.
See you on Sunday, Autopreneur.
~ Leo
Let us know if you liked today’s content, comment below!
You Might Also Like:
Here’s a simple prompt for your client meeting
Learn to build ELITE level prompts…using these 7 easy steps:
I’ve moved all my social audiences to Substack…here’s how you can too:
