Google Reclaims AI Crown with Gemini 3
4IR - Daily AI News
Welcome back to 4IR. Here’s the top stories from last week:
Google reclaims AI crown with Gemini 3 - Search giant launches most intelligent model yet, tops every leaderboard with 1501 Elo score, and deploys to 2 billion users instantly while unveiling Antigravity coding platform
Elon cuts deal to make Saudi Arabia “Grok Nation” - xAI announces unprecedented partnership to build hyperscale data centers and deploy Grok 4.1 nationwide across every Saudi government agency and business as first country-wide AI rollout
Nvidia posts monster earnings but bubble fears won’t die - Chipmaker crushes expectations with $57B in sales and $32B profit, Jensen Huang says demand is “off the charts,” yet stocks tanked anyway as circular funding deals spook investors
Google reclaims AI crown with Gemini 3
The story: Google dropped Gemini 3 on November 18th and basically said “we’re back.” The model scored 1501 Elo on LMArena—a record that beat GPT-5.1 and everything else. But the real flex? They pushed it live to 2 billion people through Search and 650 million through the Gemini app on day one. No beta testing, no waitlist, just boom—half the internet gets the upgrade. They also launched this thing called Antigravity where you just tell AI what you want to build and it writes the code across multiple windows while you watch. CEO Sundar Pichai basically admitted Google spent two years playing catch-up, but that chapter’s over now.
What we know:
Hit 1501 Elo on LMArena, beating GPT-5.1 and Claude Sonnet 4.5
Went live immediately to 2 billion Search users and 650 million Gemini users
Antigravity platform lets you describe apps in plain English, AI builds them
Handles 1 million token context with top-tier multimodal understanding
India’s Jio giving away 18 months of free Gemini 3 access to 5G customers
Why it matters: Nobody else can do what Google just did. OpenAI launches to their users, Google launches to billions overnight. That’s not just a better model—that’s distribution at scale. And with Antigravity, they’re not just competing on model quality anymore, they’re building the whole stack so developers never leave Google’s ecosystem. The other labs can have the best benchmarks all they want, but Google’s got the users and the infrastructure to actually shape how people interact with AI.
Elon cuts deal to make Saudi Arabia “Grok Nation”
The story: This one’s wild. xAI announced on November 19th they’re partnering with Saudi Arabia to basically turn the entire country into a Grok testing ground. They’re building massive GPU data centers there and deploying Grok 4.1 across every government office, business, and public service. Not a pilot program—the whole country. It’s plugging into HUMAIN ONE, which is their national AI platform that handles everything from government services to enterprise software. Meanwhile, xAI also dropped Grok 4.1 and Grok 4.1 Fast last week. The new models supposedly hallucinate 3x less, handle 2 million tokens, and can autonomously chain together hundreds of tool calls. Oh, and Elon’s out there saying Grok 5 with 6 trillion parameters is coming Q1 2026.
What we know:
First country to deploy one AI model across their entire nation
Building hyperscale GPU data centers in Saudi Arabia
Grok 4.1 integrates into every government agency and business through HUMAIN ONE
xAI raising $15 billion at a $230 billion valuation
Grok 5 (6 trillion parameters) announced for early 2026
Why it matters: This is AI as literal national infrastructure. Saudi Arabia just picked one horse and went all-in. If this works, every government on earth starts thinking “should we have a national AI model?” If it fails, well, that’s going to be the most expensive and public government tech disaster anyone’s ever seen. Either way, this sets the template for how nation-states might approach AI sovereignty going forward—not just regulation, but actual adoption at the country level.
Nvidia posts monster earnings but bubble fears won’t die
The story: Nvidia dropped earnings on November 19th that should’ve been a slam dunk—$57 billion in revenue (up 62%), $31.9 billion in profit, Jensen Huang saying Blackwell chips are selling so fast they can’t keep them on the shelves. Classic beat-and-raise quarter. Except the stock jumped 5% and then immediately nosedived to close down 3.2%. Why? Because investors are finally asking the awkward questions about circular funding. Nvidia’s investing billions into OpenAI and Anthropic, who then turn around and spend that money buying Nvidia chips. Tech companies are pouring $400 billion a year into AI infrastructure while OpenAI reportedly burned through $12 billion in just Q3 2025. Even Michael Burry—the guy from “The Big Short”—is now betting against Nvidia and calling the whole thing a house of cards.
What we know:
Revenue hit $57 billion, up 62%, with $31.9 billion in profit
Stock popped 5% after earnings, then reversed to close down 3.2%
Michael Burry shorting Nvidia, says it’s dot-com bubble 2.0
AI companies spending $400B yearly while OpenAI lost $12B in one quarter
Even Google’s CEO admitted there’s “irrationality” in AI investments
Why it matters: When earnings this good can’t save the stock, that tells you something’s broken. Nvidia’s numbers prove demand is real, but so was demand for fiber optic cable during the dot-com boom. The problem is the math—AI companies are hemorrhaging cash faster than they’re making it, yet their valuations keep climbing because nobody wants to be the one who missed the future. The circular funding deals mean the same dollars are getting counted multiple times as “revenue” across the industry, which makes it really hard to tell what’s actually sustainable versus what’s just financial engineering.
Note: Commentary sections are editorial interpretation, not factual claims