Oracle announced an expanded collaboration with OpenAI at its AI World conference in Las Vegas this week.
It brings OpenAI's models, ChatGPT and Codex, OpenAI's coding tool, into Oracle's applications, data and cloud infrastructure (OCI).
The package includes a ChatGPT-based workspace across Oracle's Fusion and NetSuite business software and OCI, plus a managed security service built on OpenAI's cyber models.
Chips and data
Nvidia and Oracle plan to expand OCI Superclusters, Oracle's large clusters of linked chips, with Vera Rubin NVL72 systems to multi-gigawatt scale.
Vera Rubin NVL72 and Vera CPUs are now generally available on OCI.
Google's tensor processing units (TPUs), its custom AI chips, will arrive on OCI in calendar 2027, making it the first hyperscale cloud outside Google to offer them.
Snowflake, the data platform company, will run its AI Data Cloud as a native, first-party service on OCI, sold and billed by Oracle.
Broker view
Yorkville Ives, the US broker led by Dan Ives, attended the event and says the announcements make OCI a credible home for the largest frontier AI workloads.
It also says Oracle is positioning the cloud as neutral on chips and data platforms, to widen its customer base beyond its own software.
At the event, OpenAI chief executive Sam Altman said GPT-6 adoption has been the strongest of any OpenAI model, according to Yorkville.
Oracle management said roughly 70% of the software at Oracle is now written by AI, Yorkville says.
The risks are plain. The OpenAI expansion deepens Oracle's dependence on a single AI lab that already accounts for a significant part of its $664 billion of remaining performance obligations, the value of contracted work not yet recognised as revenue.
The Vera Rubin build-out adds capacity that Oracle must deliver on schedule and fund without further cost to shareholders, which Yorkville says it continues to watch.
What customers said
Yorkville spoke to customers across Oracle's database, applications and cloud base, and says they are still working out what Oracle's shift from database company to AI company means for them.
Three years ago the keynotes centred on the database. This year it took a back seat to Fusion AI, which several customers said has more functionality than they need.
Customers also worry that Oracle is expanding into adjacent categories just as cheaper, more focused cloud-native specialists are winning workloads.
The database itself drew the most pointed comments. Pricing cut both ways, with some customers saying cost is a reason they choose Oracle and others saying specialists undercut it.
What has changed is the competition. Oracle's position was built when there was no credible open-source option for mission-critical relational workloads, and open-source databases now handle a growing share of them.
The offset is stickiness. Oracle's database is embedded in systems built over decades, ERP replacement cycles are long and switching costs are high.
The risk customers flagged is losing the base at the margin, as specialists take new workloads and AI changes how data is used.
One customer said AI is replacing dashboards, because users ask a question and get an answer. Another said Oracle will be a completely different company in five years.
Yorkville keeps its In Line rating, broadly equivalent to a hold, and its $155 price target on the shares, which stood at $130.53 in its note.
It says the AI pivot is the right direction but it will stay on the sidelines until execution catches up.
