• NATURAL 20
  • Posts
  • Grok Wins Big, Google Strikes Back, and the AI Race Heads to Space

Grok Wins Big, Google Strikes Back, and the AI Race Heads to Space

PLUS: IBM’s Billion-Dollar Bet on "Smart Data", Ads in Your AI? The Confusion Over Gemini’s Future and more.

In partnership with

Get a Senior Growth Team to Review Your Marketing Strategy.

Galactic Fed’s senior growth team has scaled brands like Varo, Edible, and Quiznos. And now they’re opening a limited number of free 1:1 sessions with these growth geniuses to help ambitious founders find their next breakthrough.

They’ll dig into your traffic, funnels, and ad performance to surface what’s working, what’s wasting money, and where your fastest growth levers really are.

Spots are limited, but one session could change your next quarter or your entire fiscal year.

Hey there 👋 

I’ve got a fresh batch of AI-world updates for you. Think of this as a quick coffee catch-up: some of these moves may shape what’s next for AI, how we use it, and even who gets access to what. So grab a seat, and let’s dive in.

Today:

  • Grok Wins Big, Google Strikes Back, and the AI Race Heads to Space

  • US Approves Nvidia Exports to China

  • Claude Finally Joins Your Slack Team

  • ChatGPT Goes Grocery Shopping

  • IBM’s Billion-Dollar Bet on "Smart Data"

  • Ads in Your AI? The Confusion Over Gemini’s Future

"Code Red" Over, OpenAI is about to blow…

Grok 4.20, an experimental version of xAI’s latest model, crushed the Alpha Arena trading competition, earning over 65% profit in just over two weeks making it the only profitable model. Elon Musk confirmed its identity after its victory. 

Meanwhile, Google’s Gemini 3 Pro is dominating AI benchmarks, triggering a “code red” from OpenAI, which is testing new models like Emperor and Rockhopper. Google is also selling physical TPUs to Anthropic and possibly others, challenging Nvidia. Finally, both Musk and Google are pushing for solar-powered AI data centers in space, with Musk claiming we’ll hit orbit-ready costs in under 3 years.

The biggest story right now is coming out of Washington. It looks like the US Department of Commerce is pivoting its strategy on AI hardware. According to reports from Semafor, the US is preparing to greenlight the export of Nvidia’s H200 chips to China.

Why this matters: For a long time, the strategy was "lock it down." The US banned the export of top-tier chips to slow down China's AI progress. But here’s the thing: it didn’t work exactly as planned. The restrictions inadvertently pushed Chinese companies (like Huawei) to innovate faster to fill the void.

Now, the White House is trying a "middle ground" approach. The H200 is powerful, sure, but it’s about 18 months behind Nvidia’s cutting-edge Blackwell architecture. The thinking is that by allowing these exports, the US keeps American tech as the global standard and captures that massive market revenue, while still keeping the very best toys for itself. It’s a high-stakes balancing act between national security and economic dominance.

If you use Slack for work, you’re going to like this one. Anthropic just dropped a deep integration between Claude and Slack, and it sounds like it solves a massive headache: context switching.

The Human Angle: How many times have you tried to catch up on a project by reading 50 disjointed Slack messages? With this update, you can now tag @Claude directly in a thread to get a summary or draft a reply.

But the cooler part is the "connector." When you are using Claude (the app), it can now reach into your Slack workspace (with permission, of course) to find answers. So you can ask, "What did the engineering team decide about the login bug last Tuesday?" and Claude will hunt down the context for you. It’s a big step toward AI acting like a teammate who was actually paying attention in the meeting.

Finally, we have a peek at the future of "agentic" AI—where AI doesn't just talk, it does. OpenAI and Instacart have expanded their partnership, and the integration is now seamless.

What’s new: You can now plan a meal and check out without ever leaving ChatGPT. You can say, "I need a healthy dinner idea for tonight that takes 20 minutes," and once it gives you a recipe, you can have it load the ingredients into an Instacart cart and pay for them right there in the chat.

It seems small, but it’s a huge signal. We are moving away from chatbots that just give advice and toward agents that complete errands. My only question is: will it remember that I hate cilantro?

🧠RESEARCH

"PaperDebugger" integrates AI directly into academic writing tools like Overleaf. Instead of copying text elsewhere, the AI lives in your editor, understanding your full document's history. It uses multiple smart assistants to help write, review, and fix your paper in real-time, making the editing process smoother and more accurate.

"MG-Nav" helps robots navigate new spaces better. It combines a "big picture" memory map for planning long routes with a quick-reaction system for dodging nearby obstacles. This two-part approach allows robots to find specific items and move through complex, unseen environments more reliably than before.

SIMA 2 is a new AI agent for 3D video games, built on Google's Gemini technology. It can hold conversations, follow complex visual or spoken commands, and even teach itself new skills. It acts like a smart, flexible partner rather than just a simple tool for executing basic orders.

🛠️TOP TOOLS

Each listing includes a hands-on tutorial so you can get started right away, whether you’re a beginner or a pro.

Aomni – AI Sales Agent - AI agent designed to automate the heavy lifting of B2B sales research and outreach.

AppFlowy – A Versatile and Secure Workspace Solution - privacy-first collaboration platform often described as an open-source alternative to Notion.

ArchitectGPT – AI Home Design - visual AI tool that generates photorealistic design concepts for homes, buildings, and interiors.

📲SOCIAL MEDIA

🗞️MORE NEWS

IBM to Acquire Confluent IBM has agreed to buy data streaming company Confluent for $11 billion to strengthen its AI services for businesses. The deal gives IBM access to technology that moves data in real-time, which is essential for powering advanced "agentic" AI systems that act on their own. This acquisition is designed to help companies better manage the massive flow of information required to run modern generative AI tools.

Google Gemini Ads in 2026 Reports indicate that Google told advertising partners it plans to introduce ads into its Gemini chatbot interface by 2026. However, Google executives publicly disputed this claim, stating they have no current plans to test ads in the main Gemini app. This conflict highlights the ongoing tension as tech giants figure out how to make money from their expensive AI tools without alienating users.

SoftBank and Nvidia Invest in Skild AI SoftBank and Nvidia are reportedly in talks to invest nearly $1 billion in Skild AI, a startup building "brains" for robots. This deal would value the company at $14 billion, nearly tripling its worth from just a few months ago. The massive funding aims to help Skild create a universal AI model that allows robots to see, understand, and navigate the physical world.

BNY Boosts AI with Google’s Gemini 3 Bank of New York (BNY) is upgrading its internal employee assistant, Eliza, with Google’s powerful new Gemini 3 model. This integration will turn Eliza into an autonomous agent capable of researching market trends and analyzing complex financial data on its own. The bank expects this move to automate tedious research tasks and significantly speed up decision-making for its workforce.

Google AI Glasses Coming in 2026 Google and Samsung are teaming up to release a pair of lightweight, AI-powered smart glasses in 2026. These glasses will connect to your smartphone and use Google's Gemini AI to perform tasks like real-time language translation and visual searches. The project aims to offer a comfortable wearable device that overlays digital help onto the real world without the bulk of a VR headset.

Trump’s Executive Order on AI President Trump plans to sign an executive order establishing a single federal approval process for AI, effectively blocking individual states from making their own rules. He argues that having 50 different sets of state regulations would destroy the American AI industry before it can mature. While the move aims to streamline innovation, it is expected to face strong legal opposition from states protective of their local laws.

AI Identifies Brain Tumors vs. Dead Tissue A new AI method can accurately distinguish between active brain tumors and "radiation necrosis," which is dead tissue caused by radiation treatment. This distinction is often impossible for doctors to see on standard MRI scans, leading to misdiagnoses and unnecessary surgeries. By correctly identifying the tissue type, this tool could help doctors choose the correct treatment path and save patients from harmful side effects.

What'd you think of today's edition?

Login or Subscribe to participate in polls.

Reply

or to participate.