Databricks Says AGI Already Arrived. So Why Doesn’t Your Business Feel Any Smarter?

Databricks closed a $5 billion funding round this week at a $190 billion valuation, and the number that made headlines wasn’t the valuation. It was cofounder and CEO Ali Ghodsi’s claim, made in the same breath as the announcement, that artificial general intelligence has already arrived, at least by the definition most of the industry used before 2022.

That’s a bold claim to attach to a funding round, and it’s worth taking seriously precisely because of who’s making it. Databricks isn’t a model lab trying to hype its own frontier research. It’s an infrastructure company that makes money when enterprises actually put AI to work on their own data. So when its CEO says the intelligence is already here, and then describes, in the same interview, why almost nothing inside a typical company looks meaningfully more autonomous as a result, that gap is the real story. It’s also exactly the gap we spend our days closing for clients.

What Actually Happened, and Why It’s Not Just a Funding Story

The round itself is straightforward enough: Databricks raised the new capital at a valuation up from the $188 billion term sheet it signed in July, after crossing a $7 billion revenue run rate with growth north of 80% year over year. The company says the money is going toward three specific products: Unity AI Gateway, which routes AI workloads across different models and lets enterprises set spending budgets; Lakebase, a serverless database purpose-built for AI agents that has already crossed a $100 million revenue run rate; and Genie, which is designed to connect an organization’s scattered internal information, emails, meeting notes, records, and systems, so AI tools can actually act on it.

None of that is unusual for a growth-stage infrastructure company. What’s more interesting is the argument Ghodsi built around it. He’s using a narrower, older definition of AGI, roughly: a system that can perform the intellectual tasks a person can, and outperform most people most of the time. That’s a meaningfully different bar than what most people mean by AGI today, which has drifted toward describing something closer to superintelligence: systems capable of independently accomplishing what entire research fields do collectively. Ghodsi is explicit that he’s not making that second, more extreme claim. Whether his narrower definition is the right one to call “AGI” at all is a genuinely contested question, and reasonable people in the industry land on different sides of it. That debate isn’t really the point for a business audience, though. The more useful part of what he said is what comes next.

Why “AGI Is Here” and “Nothing Feels Different” Are Both True at the Same Time

If the underlying models really are as capable as Ghodsi argues, the obvious question is why so little inside the average company looks autonomous. His answer is context. A model, no matter how capable, can’t reason usefully about a specific business problem without access to the records, internal policies, permissions, and operational systems that problem actually lives inside. Without that, you get rising token spend and not much else changing.

This isn’t a new idea from Ghodsi. He made essentially the same argument in his keynote at Databricks’ own Data + AI Summit back in June, framing it as a shift from an intelligence problem to a context problem. The consistency matters here. This isn’t a talking point invented for a funding announcement. It’s the thesis Databricks has been building its entire current product roadmap around for months, and the $5 billion raise is that thesis getting a very large vote of investor confidence.

It’s also, frankly, the exact pattern we see in client environments constantly, well before any of this made headlines. Organizations bring in a capable model, point it at a task, and get underwhelming results, not because the model is weak, but because it’s reasoning with a fraction of the actual context a human employee would have: the email thread that explains why an exception was made last quarter, the internal policy that isn’t written down anywhere a model can read it, the three systems that all hold a different piece of the same customer record. Closing that gap is a data and integration problem before it’s an AI problem.

The “Token Maxing to Value Maxing” Shift Is a Cost Story, Not Just a Product Pitch

The other thread running through Databricks’ announcement is cost. Ghodsi points to a real pattern showing up across enterprise AI deployments: companies routing routine tasks through their most expensive, most capable models by default, because nobody built the discipline to route intelligently, and watching token spend climb faster than the productivity gains justify it. Unity AI Gateway exists specifically to let organizations set budgets, compare providers, and route each task to the model that actually fits it, rather than defaulting to the most powerful (and most expensive) option every time.

This should sound familiar if you’ve been in any recent AI budget conversation. It’s the same dynamic driving why CFOs are asking harder questions before approving AI spending, and it reinforces a point worth repeating from our recent piece on the AI business case framework CFOs actually approve in 2026: the organizations getting real ROI aren’t the ones applying AI everywhere, they’re the ones sequencing deliberately and measuring what each use case actually returns relative to what it costs to run. A model routing layer helps with the mechanics of that discipline. It doesn’t replace the discipline itself.

What This Confirms for Organizations Already Running (or Considering) Databricks

Ready to Close Your Own Enterprise Context Gap?

If the model isn’t the constraint anymore, the constraint is whether your data, governance, and internal systems are actually organized well enough for AI to act on them, which is exactly what an AI and data readiness assessment is built to uncover.

As a Databricks partner, Optimum helps organizations get the underlying data foundation right before layering on agents, automation, or tools like Genie and Unity AI Gateway. Reach out for a Databricks and AI readiness consultation and we’ll help you figure out where your own context gap actually is.

About Optimum