AI.com Super Bowl Crash: The $85M AGI Bet That Collapsed Under Its Own Traffic

The only B2B auth solution built for AI-native companies (Sponsor)

Multi-tenant auth has different requirements. Make sure you’re meeting them, whether you’re onboarding humans or AI agents.

With PropelAuth, you get:

• MCP support

• Organizations

• Roles & permissions

• Enterprise SSO (SAML/OIDC)

• Custom security policies

Don’t get bogged down fighting with subpar auth tools - try PropelAuth today.

Learn More

---

On February 8, 2026, in front of roughly 130 million Super Bowl LX viewers, AI.com attempted one of the boldest launches in consumer AI history.

The pitch was minimalist and confident:

“AGI is coming. Get your handle now.”

The execution? An $85 million bet.

• $70 million for the premium AI.com domain (acquired April 2025)

• $15 million for the Super Bowl ad slot

• A mass-market push into agentic AI — autonomous systems capable of booking flights, managing portfolios, and executing real-world workflows

Within minutes of the ad airing, AI.com collapsed under traffic. Users reported:

• 504 errors

• Infinite OAuth loops

• Unreachable landing pages

What was positioned as the “front door to AGI” became, briefly, the internet’s most expensive error page.

CEO Kris Marszalek — fresh off scaling Crypto.com to 150 million users — later posted that the team had prepared “for scale, but not for THIS,” citing Google OAuth rate limits as the primary bottleneck.

But this wasn’t just a traffic spike.

It was a live demonstration of what happens when brand ambition outpaces infrastructure maturity.

And for founders, engineers, and product managers across the U.S. tech ecosystem, the lessons are career-defining.

From Hype to 504 in Minutes

AI.com’s funnel was intentionally frictionless:

Super Bowl Viewers (130M)
↓
AI.com Landing Page
↓
Single “Continue with Google” OAuth
↓
Google API rate limit
↓
Global signup failure

There were no fallback authentication paths in place. Users were limited to a single Google OAuth option, with no alternative such as Apple Sign-In or email-based registration to absorb excess demand. The system also lacked a visible queueing mechanism or any form of progressive degradation that could have managed traffic surges more gracefully. As a result, when Google’s rate limits reached peak thresholds under the massive influx of simultaneous users, onboarding effectively halted.

The irony was difficult to ignore: a company positioning itself at the forefront of autonomous AI agents was unable to scale its own user signups during its most important public debut. Although services were restored within hours, the reputational impact had already taken hold.

According to EDO analytics, the advertisement generated engagement levels nine times higher than the average Super Bowl spot, but sentiment skewed heavily negative. Social media quickly filled with screenshots of error messages, memes, and user frustration.

In a year when AI companies reportedly accounted for 23% of Super Bowl advertisements, with rivals such as Anthropic, Amazon, and Meta delivering seamless executions, AI.com’s technical failure stood out starkly.

The core issue was not the company’s ambition or willingness to make a bold, high-profile bet. Rather, it was the decision to pursue that scale without the operational redundancy required to sustain it.

Why This Moment Matters in the Agentic AI Era

This launch didn’t happen in 2016. It happened in 2026 — at a pivotal shift in AI evolution. The market is transitioning from: Chatbots answering prompts
to agentic AI executing real-world actions.

Agentic systems promise:

• Booking flights

• Executing stock trades

• Managing enterprise workflows

• Integrating across apps autonomously

When AI moves from information retrieval to action execution, reliability expectations change.

Consumers don’t judge autonomous AI against startup beta apps.
They judge it against:

• Stripe

• AWS

• Google

• Apple

This was not just a crash. It was a credibility event.