Author: Ejaaz
Compiled by: Deep Tide TechFlow
Deep Tide Guide: This article is a direct counterattack against recent pessimistic voices about xAI. The author systematically argues, from four dimensions—computing power, data, distribution channels, and physical AI—why xAI might surpass all competitors by 2026.
The core point is straightforward: While others are still discussing model architecture, Musk is already building his own power grid, airlifting gas turbines, and feeding data with Tesla robots. This is an analysis with a clear stance, worth reading.
Main text follows:
Lately, I've seen too much criticism of xAI. This article aims to set the record straight.
I will systematically break down a judgment: the upcoming Grok 5 from xAI isn't just about catching up with competitors; it's about directly surpassing them.
Don't forget, we're talking about a company that's only two years old. Yet, they built the world's largest supercomputer in 122 days (normally taking four years), achieved 600 million monthly active users, and possess something no other AI lab has—a physical carrier (yes, autonomous robots).
Enough talk, let's get straight to the point.
Musk is Building His Own Power Grid
Entering 2026, xAI's advantage in computing power is overwhelmingly dominant. Their current real-time computing power (approximately 500,000 GPUs) is greater than that of Anthropic and Meta combined.
And it doesn't stop there. Including Colossus I and II, they plan to deploy 900,000 GPUs by the second quarter of 2026. With the recently announced Colossus III (yes, another new data center already under construction) operational, it is expected to reach 1 million GPUs, with a total investment of $35 billion.
How can others catch up with this scale?
But the issue isn't just how much money was spent or how much hardware was stacked; it's how they did it. Look at this tweet:
Elon directly airlifted gas turbines to power the data centers because the grids in Tennessee and Memphis simply couldn't handle it. These turbines alone can support an additional 600,000 GPUs.
He chose to completely bypass the entire state's power grid (conventional expansion would take years) just to speed up model training. In addition, he deployed about 250MW of Tesla Megapack energy storage batteries to handle situations where the grid can't supply enough during peak usage.
This forward-thinking combined with extremely fast execution is giving xAI a huge computing power advantage over competitors.
You have to understand, the regulatory approvals, talent recruitment, and operational logistics involved in this have never been done on this scale before. xAI not only did it but made it look easy.
If the hypothesis that "more computing power = stronger model" holds true (and it currently seems to), then the rumored 7 trillion parameter Grok 5 will be a monster upon release. For comparison: Grok 4 has 3 trillion parameters—this is more than double.
NVIDIA CEO Jensen Huang on Grok 5:
"Elon has mentioned that the next frontier model, the next version of Grok, namely Grok 5, will be a 7 trillion parameter model."
The infrastructure race is no longer in doubt.
There is no infrastructure expansion war right now because xAI has already won. Their strategy is "build first, talk later." Unless other labs catch up, xAI's models will remain ahead.
X's "X-Factor": Unlocking Personal AI
xAI has stronger computing power than anyone, but top-tier models also require massive amounts of data.
And not just any data. AI labs are increasingly realizing that real-time data is the key to unlocking personalized AI—an AI that deeply understands your desires and goals and helps you execute them before you even think of it.
Google's latest "Personal Intelligence" product is the clearest signal that models are ultimately heading in this direction. But xAI has a unique advantage here that Google doesn't:
A social media platform that feeds them over 100 million posts daily.
This means over 100 million pieces of text, images, and videos can be used to train Grok, enabling:
- Real-time trends and breaking news
- Large-scale understanding of virality, trends, and human behavior
- Real-time sensing of the global cultural zeitgeist
Other models can only tell you what's happening; Grok can simultaneously tell you what's happening and how people feel about it—and faster than anyone else.
This capability is valuable.
If we assume the value users get from a tailored AI model is 10 times that of a general-purpose large model, then X's moat is very difficult to breach.
It's not just the data; X's distribution capability is also insane:
- 70 million daily active users
- 600 million monthly active users
- An "Ask Grok" button next to every post
It's not hard to imagine xAI integrating real-time prediction markets, shopping, banking, dating, and more into the same App in the future, all powered by Grok.
Currently, most model lab valuations are based on GPU count, intelligence benchmarks, and reputation. xAI has all of these, plus the opportunity to penetrate multiple internet monopoly areas—don't forget their goal is to become an "Everything App."
Today, X's recommendation algorithm is powered by Grok, which analyzes every post to make recommendations. Tomorrow, it will provide personal intelligence services for each user.
Grok is clearly not just a standard large language model; its valuation should reflect that.
Physical AI Advantage: xAI is the Most Forward-Looking Lab
It's no secret that robots will have a huge impact on the world in the next five years. The technology has finally matured.
From factory manual labor to "last-mile" delivery, from fast-food chains to top surgeons, all will be assisted or entirely replaced by robots.
The viral videos from Boston Dynamics over a decade ago have now snowballed into autonomous vehicle fleets and (surprisingly impressively) humanoid robots. Honestly, when it comes to these two things, only one company comes to mind: Tesla.
A car that drives better than a human is no longer a fantasy. The latest v14.2.2.3 update is technically already a better driver than you. Once regulations pass, you'll see autonomous Teslas transporting people everywhere.
Similarly, a humanoid robot that can carry your shopping bags and carefully wipe your mother's fine china is becoming a reality. Optimus will begin mass shipments into homes and factories by the end of this year.
What does this have to do with xAI?
Two things:
- Machines need a brain to drive them, and Tesla uses Grok.
- Grok needs diverse data sources to understand the world around it, and this data comes from Tesla's robots.
This symbiotic relationship gives xAI an almost unfair advantage over competitors. I think Google is the only company that can compete at this level, but they are still behind.
Today, Grok is already powering Tesla vehicles—the latest update lets you to simply tell Grok to drive you somewhere while playing music and telling you about Roman history.
Similarly, Grok is now receiving video data from Tesla cameras, distance data from Tesla sensors, etc., helping it understand real-world physics, visual perception, and navigation.
All this data now helps it become stronger in other capabilities, such as generating more physically accurate video content.
You have to admit Musk is playing 5D chess. He's not just building a large language model; he's building the entire ecosystem for AI to live and operate in.
Writing this, I admit it all sounds fantastic, but also incredibly ambitious......which leads to the final part of the article:
Yes, There Are Risks Here
There are risks in everything. Maybe managing 5 companies is Elon's limit, and 6 is too many......but I doubt it. If there's anyone in this world who has repeatedly proven able to prove doubters wrong time and again, it's him.
Call me crazy, I don't care—what he has already achieved is itself extremely improbable.
I think there are three key risks:
The King of Controversy — Elon and headlines are old friends. He is currently involved in a $130 billion lawsuit with OpenAI and is under investigation by EU and Indian regulators. Who knows, this guy might do something outrageous that messes up the entire vision.
Execution Risk — xAI burns about $1 billion per month; that's a huge bill. And Elon alone manages 5 companies (not including Starlink).
Scaling Laws — xAI is betting everything on "more computing power = stronger model," but if a new, better training architecture is discovered, this hypothesis could be overturned. Andrej Karpathy has stated multiple times that he doesn't believe large language models are the final form.
That's it! I think people have been unfairly critical of xAI's efforts to push the frontiers of intelligence lately and seem to forget they are still a force to be reckoned with.
I hope this article changes your perspective. Thanks for reading.
