Tag: Oracle

Posted in Artificial Intelligence, Cloud & Infrastructure

Oracle’s Multicloud Push: Oracle Database@AWS, Azure, and Google Cloud Explained

Posted on by Filip Keser

Oracle spent decades building its business on the idea that customers run Oracle Database on Oracle’s own infrastructure. In 2023, that changed. Oracle began physically installing its own database hardware inside the data centers of its three biggest cloud rivals — Microsoft Azure, Amazon Web Services, and Google Cloud — under a family of offerings it brands “Database@AWS,” “Database@Azure,” and “Database@Google Cloud.” This piece explains what each partnership actually delivers, how it differs from simply running Oracle software on rented cloud compute, and why Oracle believes giving up single-cloud exclusivity is the right trade.

AT A GLANCE

FactDetail
What it isOracle Exadata database hardware and Oracle Database software, physically deployed inside a partner cloud’s own data centers
Who makes itJoint engineering/commercial offering between Oracle and Microsoft, Oracle and AWS, and Oracle and Google Cloud (three separate deals, not one)
Primary use caseMoving mission-critical Oracle databases (often on-premises Exadata) into a public cloud without re-architecting applications, while keeping low-latency access to that cloud’s compute and AI services
Where it fits in the stackData layer — sits alongside, not instead of, the partner cloud’s native compute, analytics, and AI services; billed and supported jointly
Status (per official sources)Database@Azure: generally available in many regions worldwide. Database@AWS: GA since July 8, 2025, initially in US East (N. Virginia) and US West (Oregon), with expansion planned. Database@Google Cloud: live with Exadata and Autonomous AI Database services

Short Verdict

Oracle’s multicloud database strategy is a genuinely unusual move for an infrastructure vendor: instead of forcing customers to choose Oracle Cloud Infrastructure (OCI) over AWS, Azure, or Google Cloud, Oracle brought its own hardware to its competitors’ data centers. For enterprises with large, latency-sensitive Oracle estates who are already committed to a specific hyperscaler for compute and AI, this removes a real architectural headache — cross-cloud network hops between application and database. It is not a replacement for OCI, and it does not make Oracle Database cheaper or easier to run than it was before; it mainly removes the “which cloud do I put the database in” decision as a blocker to modernization.

What It Is: Three Separate Deals, One Pattern

Each “Database@” offering follows the same basic pattern, confirmed on Oracle’s own documentation and product pages: Oracle Exadata infrastructure is installed and run inside the partner cloud’s physical data centers, managed through OCI control systems, but purchased, provisioned, and in most cases monitored through the partner cloud’s own console. The partner cloud supplies the data-center facility, power, physical security, and networking; Oracle supplies and operates the database hardware and software stack on top of it. The three deals are not identical, however — they launched at different times, with different service catalogs and regional footprints.

Oracle Database@Azure

According to Oracle’s official Database@Azure page, this offering provides Oracle Exadata Database Service (including a version built on Exadata Exascale Infrastructure for smaller workloads that scale incrementally), Oracle Autonomous AI Database, high-availability features such as Real Application Clusters (RAC), Data Guard, Maximum Availability Architecture, and GoldenGate replication, plus Oracle Zero Downtime Migration tooling. The Exadata hardware is colocated directly in Azure data centers across North America, South America, Europe, the Middle East, Africa, and Asia Pacific. Customers buy it through Azure Marketplace, can apply spend against an existing Azure consumption commitment, and can bring their own Oracle licenses. Oracle’s pitch centers on pairing Oracle AI Vector Search and Oracle data with Microsoft Copilot and Power BI without moving data across clouds, plus unified monitoring, billing, identity, and support.

Oracle Database@AWS

Database@AWS reached general availability on July 8, 2025, per Oracle’s press release and AWS’s own “What’s New” announcement, initially in the US East (N. Virginia) and US West (Oregon) Regions, with Oracle and AWS stating plans to expand to roughly 20 more regions. The service includes Oracle Exadata Database Service and Oracle Autonomous Database on Dedicated Exadata Infrastructure, provisioned through the AWS Management Console and purchasable via AWS Marketplace, including private offers from Oracle. AWS’s announcement highlights integration with AWS Identity and Access Management, CloudWatch, CloudFormation, CloudTrail, EventBridge, and VPC Lattice, plus zero-ETL integration into Amazon Redshift for near-real-time analytics on Oracle transactional data, and the option to store backups in Amazon S3. Oracle SVP Karan Batta framed the rationale directly in Oracle’s release: “Customers have been putting their most valuable data in Oracle databases for decades… Oracle Database@AWS lets them run Oracle Database workloads on OCI in AWS,” combined with “AWS’s advanced generative AI and analytics services.”

Oracle Database@Google Cloud

Per Google Cloud’s and Oracle’s joint documentation, Database@Google Cloud deploys Oracle Exadata systems inside Google Cloud data centers, running Exadata Database Service, Autonomous AI Database Service, Base Database Service, Exadata Database Service on Exascale Infrastructure, and GoldenGate. Google Cloud supplies the core data-center infrastructure, physical and network security, and hardware monitoring for the underlying OCI resources, while customers manage the service through the Google Cloud console, the gcloud CLI, or a dedicated Oracle Database@Google Cloud API, with connectivity handled through what Google calls ODB Networks. Oracle announced expanded AI database services for this offering at Oracle AI World in October 2025. The publicly available documentation does not spell out as many specific integration workloads (equivalents of AWS’s Redshift zero-ETL or Azure’s Copilot tie-in) as the other two partnerships describe; where that specificity is missing, we have left it out rather than guess.

How the Three Partnerships Compare

DimensionDatabase@AzureDatabase@AWSDatabase@Google Cloud
Core database servicesExadata Database Service (incl. Exascale), Autonomous AI Database, RAC, Data Guard, MAA, GoldenGateExadata Database Service, Autonomous Database on Dedicated Exadata InfrastructureExadata Database Service (incl. Exascale), Autonomous AI Database Service, Base Database Service, GoldenGate
Management consoleAzure portal, OCI consoleAWS Management ConsoleGoogle Cloud console, gcloud CLI, dedicated API
Headline AI integrationOracle AI Vector Search with Microsoft Copilot and Power BIZero-ETL to Amazon Redshift for near real-time analyticsExpanded Oracle AI Database services announced at Oracle AI World 2025
PurchasingAzure Marketplace, Azure consumption commitment, Bring Your Own LicenseAWS Marketplace, including private offers; counts toward AWS commitments; Oracle Support RewardsNot detailed in public documentation at time of writing
Regional footprint (per official sources)Many regions across multiple continentsUS East (N. Virginia), US West (Oregon) at GA, ~20 more regions plannedLive in supported Google Cloud regions; exact count not specified in reviewed docs

Why Oracle Is Doing This

The strategic logic, as Oracle itself describes it, is not about winning a head-to-head cloud infrastructure contest against AWS, Azure, and Google Cloud on compute, storage, or AI model training — Oracle Cloud Infrastructure still exists as its own standalone offering for customers who want that. Instead, Database@ is built around a narrower and, for Oracle, more defensible asset: an enormous installed base of mission-critical Oracle databases that enterprises are reluctant to migrate off Exadata hardware or rewrite against a different database engine. Rather than requiring those customers to pick OCI as their cloud of record, Oracle lets them keep whichever hyperscaler already runs their applications, analytics, and AI workloads, and simply brings the Oracle database to that cloud. This also answers a very practical infrastructure problem: a database in one cloud region and an application in another cloud’s region means every query crosses the public internet or a dedicated interconnect, adding latency and often data-transfer charges. Colocating Exadata inside the same data center as the application removes that hop entirely.

How This Differs From a Traditional Single-Cloud Strategy

  • No forced migration off Exadata. Customers keep Real Application Clusters, GoldenGate replication, and Oracle’s high-availability architecture exactly as they ran on-premises, rather than being pushed onto a different database engine native to the hyperscaler.
  • Hardware, not just software, moves. This is not Oracle Database running as a VM image on rented AWS/Azure/Google compute — it is actual Oracle-managed Exadata infrastructure physically sited in the partner’s data center.
  • Unified operations across two vendors. Billing, identity, support, and (in most cases) provisioning happen through the partner cloud’s own console and marketplace rather than requiring a separate OCI relationship.
  • The customer’s “primary cloud” doesn’t change. A company that has standardized on AWS for everything else doesn’t need an OCI account, VPN, or interconnect to a separate Oracle region — the database shows up inside AWS itself.

Trade-offs

ProsCons / Open Questions
Eliminates cross-cloud latency between application and Oracle databaseStill requires Oracle Exadata licensing and support costs; doesn’t eliminate Oracle’s pricing model
No application rewrite needed to migrate off on-premises ExadataThree separate partnerships with different regional footprints and (for Google Cloud) less publicly documented integration detail — feature parity across all three is not guaranteed
Keeps customers on their chosen hyperscaler for compute/AI while retaining Oracle for dataCreates a three-way support relationship (customer, Oracle, hyperscaler) that can complicate troubleshooting versus a single-vendor stack
Enables AI/analytics tie-ins specific to each cloud (Redshift zero-ETL on AWS, Copilot/Power BI on Azure)Database@AWS launched in only two regions at GA (July 2025), so geographic coverage still trails Database@Azure

Who It’s Relevant For

Enterprises with large, latency-sensitive Oracle Exadata estates (financial services, telecom, large-scale ERP, government) that have already standardized their applications and AI tooling on a specific hyperscaler, and that want to retire on-premises Exadata hardware without a database migration project, are the clearest fit. It also suits organizations running hybrid Oracle-plus-cloud-native architectures that need the database and the application compute to sit in the same data center for compliance or performance reasons.

Who Should Look Elsewhere

Organizations building new, cloud-native applications with no existing Oracle investment have little reason to introduce Oracle Exadata licensing costs and operational complexity through this route; a hyperscaler’s own managed database service (Amazon RDS, Azure SQL, Google Cloud SQL/AlloyDB) is the simpler starting point. Companies that want the full Oracle Cloud Infrastructure platform — not just the database — are also better served going directly to OCI rather than a Database@ offering, which only brings the database layer into the partner cloud.

Bottom Line

Database@AWS, Database@Azure, and Database@Google Cloud are Oracle’s answer to a problem that has nothing to do with chips or GPUs: enterprises have decades of mission-critical data locked into Oracle Exadata, and they don’t want to choose between keeping that database and standardizing on a single hyperscaler for everything else. By physically placing Exadata hardware inside each partner’s data centers, Oracle sidesteps that choice entirely. It’s a multicloud strategy built on dependency, not on OCI winning a head-to-head infrastructure fight — and so far, based on Oracle’s own GA dates and regional rollouts, Azure has the broadest footprint, AWS is the newest and most narrowly scoped, and Google Cloud sits in between with Oracle actively expanding its AI database services there as of late 2025.

FAQ

Is Database@AWS/Azure/Google Cloud the same as running Oracle Database on a virtual machine in that cloud?
No. These are dedicated Oracle Exadata systems physically installed in the partner’s data centers and managed through OCI, not Oracle software installed on generic rented compute instances.

Do all three partnerships offer the same features?
No. Based on official documentation, Database@Azure and Database@Google Cloud both offer Exadata Database Service (including Exascale Infrastructure) plus GoldenGate, while Database@AWS’s publicly documented catalog centers on Exadata Database Service and Autonomous Database on Dedicated Exadata Infrastructure. Each also has distinct cloud-specific integrations (Redshift zero-ETL on AWS; Copilot/Power BI on Azure).

Does this replace Oracle Cloud Infrastructure (OCI)?
No. OCI continues to exist as Oracle’s own full cloud platform. Database@ offerings only bring the database layer into a partner cloud; compute, AI training infrastructure, and other services in that scenario still come from the partner (AWS, Azure, or Google Cloud), not from OCI directly.

Which of the three launched first, and which has the widest availability?
Public sources reviewed here confirm Database@Azure’s regional footprint spans many regions across several continents, and that Database@AWS reached general availability on July 8, 2025, starting in just two AWS regions with further expansion planned. Exact first-announcement dates for all three were not consistently available across the official pages reviewed for this article.

Sources

Related reading: Oracle Database 23ai and the Oracle–Nvidia Partnership, Oracle Cloud Infrastructure for AI, and OCI Supercluster Networking.

Posted in Artificial Intelligence, Cloud & Infrastructure

Oracle’s AI Cloud Capacity Deals in 2026: What the Mega-Contracts Mean for OCI’s GPU Business

Posted on by Filip Keser

Oracle’s cloud infrastructure business has gone from an afterthought to one of the most-discussed AI capacity stories in the industry, largely on the strength of a backlog metric most software companies would never let get this large: Remaining Performance Obligations (RPO), the dollar value of contracted revenue Oracle has not yet recognized. That number has become the de facto scoreboard for how much AI compute Oracle has committed to deliver — and how much financial risk comes with it. This piece sticks to what Oracle has actually disclosed in its own earnings materials, flags clearly where the widely reported headline numbers (like the reported $300 billion OpenAI deal) come from press reporting rather than an Oracle-confirmed figure, and explains what the backlog implies for OCI’s GPU buildout.

At a Glance

Metric (Oracle fiscal Q1 2027, reported Sept. 10, 2026)FigureSource
Total revenue$19.3 billion, up 30% YoYOracle press release
Cloud Infrastructure (OCI) revenue$7.4 billion, up 121% YoYOracle press release
Remaining Performance Obligations (RPO)$664 billion, up $209 billion YoYOracle press release
New AI cloud contracts booked in the quarter“More than $30 billion”Oracle press release
GPUs delivered to AI Cloud customers since Q4“More than 300,000,” nearly triple Q4’s paceOracle press release
Q2 FY27 total cloud revenue growth guidance64%–70% constant currencyOracle press release
Reported ~$300B, multi-year OpenAI compute dealNot an Oracle-confirmed dollar figurePress reporting (WSJ and others, Sept. 2025), unverified by Oracle filing

What Oracle Has Actually Disclosed

Oracle’s most recent earnings release (fiscal Q1 2027, covering the quarter ended August 31, 2026, reported September 10, 2026) is the most current primary source available as of this writing, and it is unusually specific for a company that doesn’t typically name AI customers in its press materials. Total revenue came in at $19.3 billion, up 30% year over year in both USD and constant currency. The headline figure, though, is Cloud Infrastructure (OCI/IaaS) revenue, which Oracle reported at $7.4 billion for the quarter — up 121% year over year.

The number that dominated coverage of the release was Remaining Performance Obligations: Oracle reported total RPO of $664 billion, up $209 billion year over year (implying prior-year RPO of roughly $455 billion, based on simple arithmetic from the two disclosed figures, not a percentage Oracle itself stated). Oracle attributed the jump specifically to booking “more than $30 billion of additional AI cloud contracts” during the quarter. Oracle also disclosed that it delivered more than 300,000 GPUs to AI Cloud customers since the end of the prior quarter — almost triple the capacity it delivered in the quarter before that — which is as close as the company gets to a physical-capacity proof point behind the RPO number. Oracle’s guidance for the following quarter (Q2 FY27) calls for total cloud revenue growth of 64% to 70% in constant currency, and full fiscal-year 2027 total revenue of at least $90 billion.

Notably, Oracle’s press release does not name individual AI customers behind these contracts, and it does not break out capital expenditure figures tied specifically to the AI buildout in the release itself; the company reported free cash flow of negative $5 billion for the quarter, which it attributed to investment supporting cloud infrastructure growth, alongside $23 billion of operating cash flow and $20 billion raised through an at-the-market common stock program.

What’s Reported, But Not Oracle-Confirmed

The number most people associate with this story — a roughly $300 billion, multi-year cloud computing deal between Oracle and OpenAI tied to Project Stargate — comes from reporting by outlets including the Wall Street Journal, citing people familiar with the agreement, in September 2025, and has been repeated widely since. It is a credible, widely corroborated report from reputable outlets, but it is not a figure Oracle itself has confirmed in a press release or SEC filing naming OpenAI as the counterparty and $300 billion as the contract value. Oracle’s own earnings materials describe AI cloud contract bookings only in aggregate and do not name which customers they’re with. Readers should treat the $300 billion OpenAI figure as credibly reported, not as an Oracle-disclosed fact, and the same caution applies to any other customer-specific dollar figures circulating in coverage of Oracle’s AI backlog that don’t trace back to an Oracle press release, 10-K, or 10-Q.

What RPO Actually Measures — and Its Limits

RPO is contracted, unrecognized revenue — a backlog, not cash in hand and not yet-earned revenue. It tells you Oracle has signed agreements obligating customers to pay for a certain amount of future cloud capacity, but it does not tell you the payment schedule, the credit quality of every counterparty, or how much of that capacity Oracle can physically build and deliver on the timeline the contracts assume. A large RPO number is genuinely a sign of real demand, but it’s also a number Oracle itself can influence through how aggressively it pursues long-duration, large-dollar-value contracts versus shorter, smaller ones — and analysts and financial press have flagged that funding a buildout this large requires Oracle to raise substantial capital (debt and equity, including the $20 billion at-the-market stock sale disclosed this quarter) ahead of actually recognizing the revenue those contracts represent. That’s a normal, disclosed part of Oracle’s financing strategy, not evidence of anything irregular, but it’s the context that should accompany any RPO headline number: backlog and delivered, recognized revenue are two different things, and the gap between them is where execution risk lives.

What This Means for OCI’s GPU Business

Taken together, Oracle’s own disclosures paint a consistent picture: OCI’s AI infrastructure business is growing extremely fast off a small base, is heavily reliant on a small number of very large contracts (Oracle’s SEC filings in prior quarters have separately disclosed meaningful customer concentration in RPO, a risk factor worth watching in Oracle’s subsequent 10-Q filings), and requires Oracle to keep raising capital well ahead of recognizing the revenue those contracts promise. The 300,000+ GPUs delivered since the prior quarter is the clearest evidence that contracted demand is translating into physically deployed capacity rather than remaining purely on paper, which matters because RPO growth without matching capacity delivery would be a much weaker signal. For a deeper look at OCI’s technical positioning against AWS, Azure, and Google Cloud for AI workloads, see our earlier piece linked below; this article is specifically about the financial and contractual side of that buildout.

Trade-offs and Open Questions

  • Strength: Oracle’s own numbers (RPO, OCI revenue growth, GPU delivery figures) are independently disclosed, audited-adjacent earnings figures, not marketing claims.
  • Uncertainty: individual mega-deal dollar figures (the reported $300B OpenAI agreement, and similar figures attributed to other AI labs in press coverage) are not independently confirmed by Oracle at the customer level.
  • Risk: a backlog this concentrated in a handful of counterparties carries real execution and credit risk if any major contract is delayed, renegotiated, or reduced — a dynamic Oracle’s own filings have flagged as a customer-concentration risk factor in the past.
  • Capital intensity: delivering on this backlog requires continuous, large-scale capital raising (debt and equity) well ahead of recognizing the matching revenue, which is a structural feature of this business model, not a one-time event.

The Bottom Line

Oracle’s own disclosed numbers — $664 billion in RPO, 121% OCI revenue growth, over 300,000 GPUs delivered in a single quarter — are real, verifiable, and describe a genuinely large and fast-growing AI infrastructure business. The specific mega-contract dollar figures that dominate headlines, including the widely reported $300 billion OpenAI deal, are credible press reporting but not figures Oracle has confirmed itself at the customer level, and that distinction matters when deciding how much weight to put on any single number. The more durable signal is the relationship between backlog growth and actual GPU delivery: as long as Oracle keeps converting RPO into deployed capacity at a pace resembling what it reported this quarter, the backlog is doing its job as a leading indicator rather than an accounting overhang.

FAQ

What exactly is Remaining Performance Obligations (RPO)?
It’s the total dollar value of contracted revenue a company hasn’t yet recognized under its revenue recognition rules — essentially a backlog of signed, unfulfilled contracts. For Oracle, it ballooned specifically because of large, multi-year AI cloud infrastructure agreements.

Has Oracle confirmed the $300 billion OpenAI deal?
Not at that specific dollar figure, tied to that specific customer, in an Oracle press release or SEC filing as far as this research could verify. The number traces to media reporting (the Wall Street Journal and subsequent outlets, September 2025) citing people familiar with the agreement. Oracle’s own earnings materials discuss AI contract bookings only in aggregate.

Is a growing RPO automatically good news?
It’s a positive demand signal, but RPO is unrecognized, contracted revenue, not cash or completed work. Its value depends on counterparty credit quality, contract durability, and Oracle’s ability to actually deliver the capacity on the assumed timeline — none of which the headline number alone tells you.

How is Oracle funding this buildout?
Through a mix of operating cash flow, debt, and equity issuance; this quarter’s release specifically disclosed $20 billion raised via an at-the-market common stock program alongside negative $5 billion in free cash flow, which Oracle attributed to infrastructure investment.

How does this relate to Oracle’s other AI announcements, like Database 23ai or its Nvidia partnership?
Those cover different layers of the same broader strategy: Database 23ai is about AI features built into Oracle’s database product, and the Nvidia partnership concerns the GPU hardware supply behind OCI. This article is specifically about the financial scale of Oracle’s AI cloud capacity contracts and what the backlog numbers imply.

Sources

Related reading on this site: Oracle Cloud Infrastructure for AI: Is OCI a Real Alternative to AWS, Azure, and GCP?, Oracle Database 23ai and the Oracle–Nvidia Partnership, and Oracle (ORCL) Stock: Business Overview, Latest Earnings, and Valuation in 2026.

Posted in Artificial Intelligence, Cloud & Infrastructure

Oracle Database 23ai and the Oracle–Nvidia Partnership: What “AI Database” Actually Means in 2026

Posted on by Filip Keser

Oracle talks about “AI” in two very different senses, and the two get blurred together in a lot of coverage. One is a feature inside the database engine: Oracle Database 23ai’s AI Vector Search, which lets you store and query embeddings alongside ordinary rows without standing up a separate vector database. The other is an infrastructure deal: Oracle’s expanding partnership with Nvidia to put GPU capacity — from H200s to Blackwell-generation GB200 NVL72 racks — inside Oracle Cloud Infrastructure (OCI). This piece pulls those two threads apart and checks what Oracle’s own documentation actually supports.

At a Glance

ThingWhat Oracle states
Core DB featureAI Vector Search — native vector data type, approximate (HNSW) and exact/partitioned indexes, built into SQL
Cost of Vector SearchNo additional charge on Autonomous Database, Exadata Database Service, and Oracle Database Free (cloud services); on-prem Standard/Enterprise Edition pricing isn’t stated on the feature page
23ai statusLong-term support release, renamed from “23c” to “23ai” and made generally available for enterprises in 2024
Nvidia GPUs on OCIH200 Tensor Core (GA), GB200 NVL72 (live, liquid-cooled racks), Vera Rubin NVL72 (listed, no specs given)
OCI Supercluster scaleUp to 260 exaFLOPS with Hopper GPUs; 2.4 zettaFLOPS with Blackwell GPUs (Oracle’s own figures, not independently benchmarked)
DGX CloudGB200 NVL72 racks available through both NVIDIA DGX Cloud and OCI directly
Pricing transparencyNo public per-GPU or per-cluster price list for OCI Nvidia instances

Short Verdict

AI Vector Search is a genuinely useful, well-documented feature if you’re already an Oracle Database shop — it saves you from bolting on a separate vector database for retrieval-augmented generation (RAG). The Nvidia partnership is real and substantial (Oracle is one of the launch partners for GB200 NVL72 at scale), but Oracle’s own marketing pages give performance superlatives (“largest, fastest AI supercomputer in the cloud”) without GPU counts, bandwidth figures, or prices to back them up. Treat the database feature as verified and ready to evaluate; treat the infrastructure claims as directionally true but not independently confirmed.

What It Is: Two Separate Stories Under One Banner

Oracle Database 23ai (first shown as “23c” at Oracle CloudWorld in September 2023, and renamed and made broadly available in 2024) is a long-term support release of the core database. Its headline addition is AI Vector Search: a native VECTOR data type, so embeddings sit in the same tables as your relational, JSON, text, spatial, and graph data instead of in a bolted-on vector store. That is the product story.

Separately, Oracle and Nvidia have an infrastructure partnership that predates and extends past 23ai: Oracle Cloud Infrastructure offers Nvidia GPUs — up through the latest Blackwell-generation GB200 NVL72 racks — as “OCI Superclusters,” and Nvidia’s own DGX Cloud service runs some of its training capacity on that same OCI hardware. The database feature doesn’t require this specific Nvidia hardware to work; the two are related mainly in that Oracle markets both under an “AI” umbrella and because AI Vector Search can call out to Nvidia NIM inference containers for embedding generation.

How AI Vector Search Actually Works

CapabilityWhat it does, per Oracle’s documentation
Native VECTOR typeStores vectors directly in ordinary tables, supporting multiple dimension counts/formats; sparse vectors store only non-zero values
Vector generationImport an embedding model in ONNX format and generate vectors inside the database, call out to NVIDIA NIM containers, or import pre-computed vectors
IndexesIn-memory HNSW (approximate) graph index for speed; neighbor-partition indexes for very large datasets; hybrid indexes combine full-text and vector search
Distance metricsBuilt-in Euclidean, cosine, and dot product, plus user-defined custom distance functions
Query modelSimilarity search is plain SQL, combinable in the same query with relational, JSON, text, spatial, and graph predicates
RAG supportA LlamaIndex integration and a “Select AI” feature that turns natural-language prompts into SQL using OCI Generative AI or third-party LLMs
Iceberg supportSimilarity search can run directly over Apache Iceberg tables without first loading embeddings into the database

Oracle also says Exadata System Software 24ai adds an “AI Smart Scan” that offloads part of vector search to storage servers, and that OCI GoldenGate 23ai can replicate vectors in real time across different vector stores. Both are Exadata/cloud-tier features rather than something every 23ai installation gets automatically.

The Nvidia Side: OCI Superclusters and DGX Cloud

On the infrastructure side, Oracle’s current Nvidia lineup (per oracle.com/cloud/nvidia) includes general availability of OCI Supercluster with H200 Tensor Core GPUs, live liquid-cooled GB200 NVL72 racks available through both OCI directly and NVIDIA DGX Cloud, and an announced (but unspecified) Vera Rubin NVL72 tier for future training and inference workloads. Oracle’s claimed Supercluster scale is up to 260 exaFLOPS on Hopper-generation GPUs and 2.4 zettaFLOPS on Blackwell — figures Oracle presents as capable of training trillion-parameter models, with “ultrafast RDMA cluster networking” tying nodes together, bare-metal instances with no virtualization overhead, and orchestration through OCI Kubernetes Engine (OKE).

Oracle doesn’t publish the GPU counts behind those FLOPS figures, the actual network bandwidth, or a price list for renting Supercluster capacity — all of that requires talking to Oracle sales. NVIDIA AI Enterprise software is available natively inside the OCI console (Oracle cites “more than 160 AI tools”), and NIM inference endpoints are distributed through the OCI Marketplace, which is the most concrete link back to the database side: a 23ai instance can call a NIM container running on this same Nvidia infrastructure to generate the embeddings it stores.

Pros

  • Vector search lives inside the database you already run — no separate vector-database vendor, no separate data-sync pipeline
  • AI Vector Search is explicitly free of additional charge on Autonomous Database, Exadata Database Service, and Oracle Database Free
  • Multiple index types (HNSW for speed, partitioned for scale) and built-in distance metrics cover most RAG use cases out of the box
  • The Nvidia partnership gives Oracle customers a path to current-generation GB200/Blackwell capacity without a separate relationship with Nvidia or a hyperscaler

Cons

  • Oracle’s “free” claim for AI Vector Search is documented for its own cloud services and the Free edition — it is not clearly stated whether on-premises Standard or Enterprise Edition customers get it at no extra licensing cost, and third-party licensing-advisory sites report enough ambiguity here that you should confirm directly with Oracle or a licensing specialist before assuming
  • Supercluster performance figures (exaFLOPS, zettaFLOPS) are Oracle’s own marketing numbers, not independently benchmarked, and the underlying GPU counts aren’t disclosed
  • No public pricing for OCI Nvidia GPU instances or Superclusters — budgeting requires a sales conversation
  • The database feature and the GPU infrastructure are only loosely coupled; you can use 23ai’s vector search with any embedding model and don’t need Oracle’s Nvidia capacity to benefit from it

Who It’s a Good Fit For

Teams already standardized on Oracle Database who want to add RAG or semantic search without introducing a new vector-database dependency are the clearest fit for AI Vector Search. For the Nvidia infrastructure side, organizations that are already OCI customers, or that specifically want GB200/Blackwell capacity without negotiating directly with Nvidia or AWS/Azure/GCP, have a legitimate reason to look at OCI Superclusters — though they should get firm pricing in writing before committing.

Who Should Look Elsewhere

If you’re not already running Oracle Database, the vector-search feature alone isn’t a strong reason to migrate — dedicated vector databases and extensions for Postgres (pgvector) or other engines solve the same problem with less lock-in. For raw GPU capacity, teams that need transparent, published per-hour pricing should compare OCI’s actual quoted rates against AWS, Azure, and GCP before assuming Oracle is cheaper or faster; none of the performance superlatives on Oracle’s page are independently verifiable from the page itself.

Bottom Line

“AI Database” at Oracle really means two things bundled under one marketing phrase: a solid, well-documented set of in-database vector search and RAG capabilities in Database 23ai, and a large but opaque Nvidia GPU infrastructure deal that supplies the compute for training and inference elsewhere in Oracle’s cloud. The database feature is concrete enough to evaluate today. The infrastructure claims are the kind of numbers you verify with a sales rep, not the kind you take from a marketing page.

FAQ

Is Oracle Database 23ai the same as 23c?
Yes. Oracle renamed the long-term support release from “23c” to “23ai” and made it generally available for enterprises in 2024, citing the AI-focused features as the reason for the rename.

Do I need Nvidia GPUs to use AI Vector Search?
No. Vector generation can use an imported ONNX embedding model, pre-computed vectors, or an NVIDIA NIM container — the last option is convenient if you’re already on OCI, but it isn’t a requirement.

Is AI Vector Search really free?
Oracle states it’s available at no additional charge on Autonomous Database, Exadata Database Service, and Oracle Database Free. Whether that extends to on-premises Standard or Enterprise Edition licenses isn’t stated on Oracle’s feature page, so confirm with Oracle directly if that’s your deployment model.

What’s the difference between OCI Supercluster and NVIDIA DGX Cloud?
They can be the same underlying hardware. Oracle says GB200 NVL72 racks are available through both OCI directly and through NVIDIA DGX Cloud running on OCI infrastructure — the choice is mainly about which company’s service layer and support you’re buying.

Where can I get actual OCI Nvidia GPU pricing?
Oracle does not publish per-GPU or per-cluster rates for its Nvidia Supercluster offerings on its marketing pages; pricing requires contacting Oracle sales.

Sources

Posted in Artificial Intelligence, Cloud & Infrastructure

Oracle Cloud Infrastructure for AI: Is OCI a Real Alternative to AWS, Azure, and GCP for GPU Workloads?

Posted on by Filip Keser

Oracle spent most of the cloud era as an afterthought behind AWS, Azure, and Google Cloud. In AI infrastructure, that’s changed: Oracle Cloud Infrastructure (OCI) now runs some of the largest publicly disclosed GPU superclusters in the industry, and it’s become a default landing spot for AI labs that need massive GPU capacity fast. Here’s what OCI actually offers for AI training and inference, and where its claims hold up.

At a Glance

OfferingWhat it isScale (per Oracle)
OCI SuperclusterBare-metal GPU clusters with RDMA networkingUp to ~131,000 NVIDIA B200 GPUs; roadmap to 1M+ GPUs by end of 2027
GPU shapesNVIDIA H100/H200/B200/GB200, AMD MI300XBare-metal, 8-GPU instances, no hypervisor overhead
OCI Enterprise AI / Generative AIManaged LLM, RAG, and AI agent toolingBuilt on OCI Data Science and OCI’s generative AI service

Short Verdict

OCI is a legitimate option for large-scale GPU training, not just a marketing claim. Oracle publishes real bare-metal pricing per GPU-hour (rather than burying it behind a sales call for every tier), its RDMA cluster networking numbers are credible, and it has landed genuine customers like Databricks/MosaicML and Cohere. It is not yet a broad AI platform the way AWS or Google Cloud are — OCI’s AI-specific managed services are thinner, and some of its own marketing pages are inconsistent on basic numbers. For teams that mainly need raw GPU capacity at competitive pricing, OCI is worth a serious look. For teams that want a mature ecosystem of managed AI services around the GPUs, AWS and Google Cloud are still further ahead.

What OCI’s AI Infrastructure Actually Is

OCI’s pitch for AI workloads centers on bare-metal GPU instances — meaning customers get direct hardware access with no hypervisor layer between them and the silicon — connected by Oracle’s own RDMA cluster network. Oracle calls the resulting pooled hardware an “OCI Supercluster.” According to Oracle’s own AI infrastructure page, a single OCI Supercluster can scale up to roughly 131,000 NVIDIA B200 GPUs, more than 100,000 GB200 Grace Blackwell Superchips, 65,536 H200 GPUs, or 16,384 AMD MI300X GPUs, depending on which generation of hardware a customer provisions. Oracle has also stated a roadmap target of extending this architecture to more than 1 million NVIDIA GPUs with over 3 gigawatts of capacity by the end of 2027 — a forward-looking target, not a currently deployed figure.

Worth flagging directly: Oracle’s own AI infrastructure page is not perfectly internally consistent. The written copy cites “up to 131,072” B200 GPUs in one place while an accompanying diagram on the same page shows 128,000 — a minor discrepancy, but one that suggests these maximum-scale figures should be read as approximate marketing ceilings rather than exact, audited numbers.

GPU Shapes and Pricing

Unlike some competitors, Oracle publishes real per-GPU hourly pricing for several of its bare-metal shapes directly in its own product announcements:

  • BM.GPU.H200.8 — 8x NVIDIA H200 GPUs (141GB HBM3e each, 4.8TB/s bandwidth) via NVLink, dual 56-core Intel Sapphire Rapids CPUs, 3TB DDR5 RAM, 8×3.84TB local NVMe. Oracle’s own GA announcement states this ships at $10 per GPU per hour — the same price as the prior-generation H100 shape.
  • BM.GPU.MI300X.8 — 8x AMD MI300X GPUs (192GB HBM3 each, ~1.5TB total), dual 56-core Sapphire Rapids CPUs, 2TB DDR5 RAM. Oracle’s own GA announcement states this ships at $6.00 per GPU per hour.
  • GB200 NVL72 / B200 — Oracle’s current Blackwell-generation shapes, offered via OCI Supercluster at scale; Oracle does not publish a flat per-GPU hourly rate for these on its main marketing pages, and we could not independently confirm a specific dollar figure for this tier from an Oracle-published source.

Oracle’s cluster-level networking claims include RDMA over Converged Ethernet (RoCE v2) using NVIDIA ConnectX-7 NICs, cluster bandwidth up to 3,200 Gb/s per 8-GPU instance, and sub-10-microsecond network latency (Oracle’s own materials give slightly different numbers in different places — as low as 2.5 microseconds in one spot and a 2.5–9.1 microsecond range in another, which is worth noting rather than papering over). RDMA networking, block storage, and the first 10TB of monthly egress are included at no additional charge, which is a genuine differentiator versus cloud providers that meter network egress aggressively.

Managed AI Services on Top of the Hardware

Above the raw GPU layer, Oracle offers OCI Data Science (managed notebooks and an Accelerated Data Science SDK, billed only for the compute and storage actually consumed) and a generative AI layer that Oracle has marketed under more than one name over time — it has documented an “OCI Generative AI Service” for LLM and RAG workflows, and its current top-level marketing increasingly uses the name “OCI Enterprise AI” for AI app, agent, and governance tooling. Readers evaluating this layer specifically should check Oracle’s current docs directly, since the branding appears to be in flux.

On the partnership side, Oracle has a deep full-stack relationship with NVIDIA (NVIDIA AI Enterprise is built into the OCI Console, NVIDIA NIM is available via OCI Marketplace, and GB200/GB200 NVL72 racks are deployed on OCI) and a separate relationship with AMD around the MI300X shape.

Pros

  • Published, bare-metal per-GPU-hour pricing for H200 and MI300X, rather than quote-only pricing for every tier.
  • No hypervisor overhead on bare-metal shapes, and RDMA networking, storage, and the first 10TB of egress included free.
  • Real supercluster scale, with concrete customer wins (Databricks/MosaicML, Cohere, Palantir, Stanford among others cited by Oracle).
  • Both NVIDIA and AMD GPU options, rather than a single-vendor lock-in.

Cons

  • Thinner managed-AI-service ecosystem than AWS Bedrock/SageMaker or Google Vertex AI, and the branding of Oracle’s generative AI layer has shifted, which makes it harder to evaluate from the outside.
  • Oracle’s own marketing pages contain small internal inconsistencies in headline numbers (GPU counts, latency figures), so treat maximum-scale claims as approximate.
  • Oracle rarely names AWS, Azure, or Google directly in comparative claims (“up to 220% better pricing than other cloud providers”), which makes its competitive claims hard to independently check.
  • Flat per-GPU pricing isn’t published for every GPU generation, so budgeting for the newest Blackwell-class shapes may still require a sales conversation.

Who It’s a Good Fit For

Teams training large models who need a lot of bare-metal GPU capacity, want predictable per-GPU pricing without egress surprises, or are already running Oracle Database/applications and want to keep AI workloads in the same cloud. It’s also a reasonable second-cloud option for teams hedging against GPU scarcity on AWS or Azure.

Who Should Look Elsewhere

Teams that want a deep, mature managed-AI-services layer (fine-tuning pipelines, model registries, a wide first-party model catalog) with less DIY infrastructure work will likely find AWS, Google Cloud, or Azure more complete today. Teams that need the absolute newest Blackwell-class hardware at a clearly published flat rate may also need to get a quote rather than relying on OCI’s public pricing pages alone.

Bottom Line

OCI’s AI infrastructure claims are largely backed by real, Oracle-published numbers — bare-metal GPU shapes with transparent per-GPU-hour pricing on at least two major tiers, credible RDMA networking specs, and genuine large customers. It’s a credible fourth option alongside AWS, Azure, and Google Cloud for raw GPU capacity, with the caveat that its managed AI-service layer and some of its headline marketing numbers are less polished and less consistent than its hyperscaler rivals.

FAQ

Does OCI support both NVIDIA and AMD GPUs?
Yes. OCI offers NVIDIA H100, H200, B200, and GB200 Grace Blackwell Superchip shapes, plus AMD MI300X bare-metal shapes.

Is OCI cheaper than AWS or Azure for GPU training?
Oracle claims significantly better price-performance than “other cloud providers,” but does not name competitors directly or publish a verifiable methodology, so this should be benchmarked for your specific workload rather than taken at face value.

Who is actually using OCI for AI training?
Oracle has publicly named Databricks/MosaicML, Cohere, Palantir, Stanford, Suno, and together.ai, among others, as customers running AI workloads on OCI.

Sources