Can xAI Keep Up Its Momentum In the AI Race?

Last week, Elon Musk posted another bold claim on X:

It’s easy to dismiss this as classic Musk. After all, he’s been known to make sweeping statements about his own tech that often contain more hype than substance.

But here’s the thing…

This time, he might be right.

Speed Like No Other

In the past six months, xAI has released three major models: Grok 3, Grok 4 and Grok 4 Heavy.

To put that in perspective, OpenAI took nearly a year to move from GPT-4 to GPT-4o. Google’s Gemini 1 debuted in late 2023, but Gemini 2.0 didn’t arrive until early 2025. And Anthropic’s Claude series has had steady releases, but they’ve been nowhere near this aggressive.

In other words, Musk’s team is iterating at a pace that the rest of the industry isn’t matching.

And that’s largely due to Colossus, xAI’s GPU supercomputer.

Nvidia’s H100 chips are the gold standard for training AI models, but global demand has created a shortage. Musk anticipated this crunch and secured massive H100 orders early through Tesla and SpaceX.

By funneling those chips into xAI, he gave Grok access to more raw compute than most of his rivals can dare to dream of.

When Grok 3 was launched in February, it was trained using 10 times more compute than Grok 2.

By July, just five months later, Grok 4 Heavy was live and positioned as a frontier-class model that could compete with anything else on the market.

And on paper, Grok 4 Heavy is a major step forward.

It scored 44.4% on Humanity’s Last Exam, a notoriously tough benchmark designed to measure reasoning under stress.

Source: https://x.com/Prashant_1722

That placed it ahead of Google’s Gemini 2.5 Pro and OpenAI’s o3.

For subscribers willing to pay $300 a month, xAI even offers “SuperGrok Heavy.” To me, this premium offering proves that Musk puts a high value on raw performance.

But speed and performance aren’t the only differentiators here.

You see, Musk wired Grok directly into X. That means millions of users stress-test the system in full view of the world every day.

That’s a high-risk, high-reward strategy.

On the plus side, it gives xAI a feedback loop that is more public and unfiltered than anything its rivals allow. And because it pulls directly from the firehose of X data, Grok is able to stay culturally relevant in ways its rivals can’t.

But the downside is painfully obvious.

When you build in public, you can also fail in public. And Grok has taken some big “L”s.

Earlier this year, Grok generated antisemitic content, including one infamous moment where it described itself as “MechaHitler.”

Musk blamed it on a regression bug.

And just this month, Grok’s main account was suspended after it posted hate speech related to genocide.

Musk called it a “dumb error.”

But for critics, these incidents are proof that Grok’s safety systems lag behind its speed.

And its problems don’t stop there.

Recently, reports show that Grok’s “share” feature left over 300,000 user conversations exposed to search engines, including many that were never meant to be public.

Some of them contained dangerous content, like bomb-making instructions, advice on self-harm and even assassination plots against Musk.

This is a clear failure of privacy and content control. And it highlights one of the biggest challenges in the AI race…

Benchmarks matter, but trust matters more.

Benchmarks can’t erase reputational damage. And they don’t satisfy regulators either.

That’s why OpenAI, Google, and Anthropic all invest heavily in alignment teams and safety tests designed to catch bad behavior early. They’ve also worked closely with U.S. and European regulators on AI safety standards.

That’s just good business.

Because businesses and governments only want to work with AI models that they believe are safe, predictable and aligned with regulatory frameworks.

But Musk tends to treat regulations as roadblocks.

That might win him fans on X. It might even be what’s required to beat China to artificial superintelligence (ASI).

But it could also shut xAI out of exactly the kinds of business and government deals that decide who wins the AI race.

Because of this, I believe OpenAI, Google and Anthropic still hold an advantage.

So where does that leave Musk’s claim?

Right now, xAI is sprinting ahead on iteration speed, using sheer compute power to close the gap. And that’s something his competitors should be worried about.

But the AI race is a marathon, not a sprint.

OpenAI already has deep enterprise adoption through Microsoft, putting its models inside everything from Azure to Word.

Google bundles Gemini across Gmail, Docs and Search, and now it’s integrating into the Android operating system as the replacement for Google Assistant.

Meanwhile, Anthropic has Amazon behind it, with Claude integrated into AWS.

These are ecosystems. And ecosystems can create moats around businesses.

But for now, Musk only has X.

As we’ve talked about before, for xAI to “outpace” its rivals, Grok will need to scale beyond a single social platform.

Because speed alone won’t win the AI race.

Here’s My Take

Elon claims Grok is evolving faster than any other AI.

And I agree with him.

xAI has proven it can move faster than anyone else. And if Grok continues to post benchmark gains at this pace, the rest of the industry could end up chasing its lead.

But to win the AI race, Musk will need to prove that Grok has staying power. And that will come from the patient work of proving that his AI is safe and enterprise-ready.

If he can do that, his recent post won’t look like hyperbole at all.

But if he can’t, then Grok could end up being remembered as the fastest runner in a race of its own.

For investors, though, it doesn’t necessarily matter who wins this race….

As long as you are invested in the infrastructure and power needed to keep it running.

Regards,

Ian KingChief Strategist, Banyan Hill Publishing

Editor’s Note: We’d love to hear from you!

If you want to share your thoughts or suggestions about the Daily Disruptor, or if there are any specific topics you’d like us to cover, just send an email to [email protected].

Don’t worry, we won’t reveal your full name in the event we publish a response. So feel free to comment away!