Last updated: October 23, 2026
TL;DR Palantir SAP integration runs through an SAP-certified ABAP add-on, the Palantir Foundry Connector 2.0 for SAP Applications, which reads ECC, S/4HANA, and BW directly. Extract through Business Content extractors and CDS views where you can, use raw tables for gaps, and pick delta methods deliberately. Plan for licensing, authorizations, and Basis team time before the first sync. Our data integration engineering team builds these connections.
Palantir connects to SAP through a certified ABAP add-on installed on ECC, S/4HANA, or BW, which lets Foundry extract tables, CDS views, Business Content extractors, BW InfoProviders, and function modules, in batch or incrementally.
Almost every Foundry program we see ends up needing SAP data. Purchase orders, inventory, plant maintenance, sales orders, the general ledger. It’s where the operational truth lives.
And the clock is loud. SAP’s own timeline ends mainstream maintenance for Business Suite 7, ECC included, at the end of 2027, with optional extended maintenance to the end of 2030. Gartner data reported by The Register found only about 39% of 35,000 ECC customers had licensed S/4HANA by the end of 2024. So most Foundry teams will connect to ECC now and S/4HANA later. Sometimes both at once.
What’s online about Palantir SAP integration is mostly partnership news. This guide is the practical side. How the connection works, what your Basis team needs to do, what to extract from ECC versus S/4HANA, how deltas behave, and the licensing and SAP Note questions that stall projects. For moving a whole legacy warehouse, see our Foundry migration guide instead.
How Does Palantir Foundry Connect to SAP?
Through the Palantir Foundry Connector 2.0 for SAP Applications, an SAP-certified add-on installed on the SAP system. Foundry calls it directly, through an SLT server for replication, or through a gateway system when the source is too old for the add-on.
Palantir’s SAP overview says the add-on was built with Diskover Limited, is installed through SAINT like any other add-on, and covers S/4HANA, ECC, BW, and SLT. It reads tables, functions, BW InfoProviders, and BEx queries.
The architecture documentation lays out three patterns.
| Pattern | How it works | Use it when |
|---|---|---|
| Direct | Add-on on the ERP or BW system; Foundry connects to it | The source meets the minimum SAP_BASIS level, which is the usual case |
| Through SLT | SLT captures table changes with database triggers into ODQ; Foundry reads from SLT | You need near real-time table replication or don’t want load on ERP |
| Through a gateway | Add-on on a newer gateway system that reaches the old source remotely | The source is below the minimum release, such as very old R/3 systems |
There’s a newer fourth path worth knowing. SAP Data Accelerator is an SAP-managed connectivity service through SAP Cloud Connector. SAP’s reference architecture calls it the SAP-endorsed, preferred path for ingesting SAP data into Foundry. Palantir’s docs still label it beta and note it only changes the network path. Same connector underneath. Worth asking about, especially on RISE.
What Does Your Basis Team Need to Do First?
Install the add-on in client 000 through SAINT, activate the SICF services, create a system-type technical user with Palantir’s roles, and open network access. The source needs SAP_BASIS 7.5, or 7.4 at SP5 or later.
Palantir’s installation guide is specific. Two packages, PALANTIR and PALCONN. SICF services under /sap/palantir and /sap/opu/odata/palantir. A system user, type B, assigned the /PALANTIR/ authorization roles. SAP_BASIS 8.x is supported too.
Sounds quick on paper. In practice it’s the longest step of the whole integration. Here’s why.
- Add-on installs go through your SAP change process, with transports through development, quality, and production
- Security will want to review every authorization object in the Palantir roles, and they should
- Firewall rules between Foundry or its agent and SAP take their own ticket queue
- Someone has to decide which client and which company codes Foundry may read
Our rule of thumb? Book the Basis team before you book the data engineers. We’ve watched pipelines get built against a sandbox while the production install waited weeks in a change board queue.
Also check the connector version. Palantir’s SAP FAQ notes CDS view support starts at support pack 21 and HANA information views at support pack 22. Older installs need an upgrade before they can do what this guide describes.
What Should You Extract From ECC Versus S/4HANA?
Prefer Business Content extractors on ECC and released CDS views on S/4HANA, because both carry business logic and delta support. Fall back to raw tables when neither exists. Use BW InfoProviders when the logic you need already lives in BW.
The SAP ERP connector page lists what Foundry can read. ERP tables and views, including cluster, pool, and custom Z tables. ABAP CDS views, including ones with parameters. Business Content extractors. Function modules and BAPIs. And from BW, the FAQ adds DSOs, InfoCubes, InfoObjects, and BEx queries.
Which to pick? Here’s the decision table we use.
| Source object | Best for | Delta support | Watch out for |
|---|---|---|---|
| Business Content extractors, mainly ECC | Standard areas like sales, billing, inventory movements, FI line items | Native extractor delta | Must be activated in RSA5; some need setup tables filled first |
| CDS views, mainly S/4HANA | Modern, semantically clean business objects | Depends on the view; often timestamp based | Prefer released views; custom views need an owner |
| Raw tables, like VBAK, EKPO, MSEG, or ACDOCA | Gaps no extractor or view covers | Timestamp comparison, if a change date exists | You rebuild SAP’s joins and business rules yourself |
| BW InfoProviders and BEx queries | Logic already modeled and trusted in BW | Varies by object | You inherit BW’s latency and design choices |
| Function modules and BAPIs | Small lookups or calculated values | Snapshot only | Not for bulk history |
ECC and S/4HANA differ in one way that matters a lot. S/4HANA collapsed many finance and inventory tables. ACDOCA, the universal journal, replaces much of what lived across BKPF, BSEG, and other ECC tables, and MATDOC replaces MSEG for material documents. Pipelines written against ECC tables won’t simply repoint to S/4HANA.
That’s the strongest argument for extractors and CDS views over raw tables. They put a stable interface between Foundry and SAP’s internals. If your S/4HANA move is coming, model the Foundry Ontology on business objects like Purchase Order, Material, and Plant, not on table names. When SAP changes underneath, you remap a pipeline. Your apps don’t notice.
How Do Incremental and Real-Time Syncs Work With SAP?
Three ways. Extractors with native SAP delta, timestamp-based incremental reads on tables and views, and SLT replication for near real-time table changes. Transaction codes and function modules are always full snapshots.
Palantir’s connector documentation describes each approach and its trade-offs.
- Extractor delta uses SAP’s own change tracking, so it’s the cleanest option when an extractor exists
- Timestamp incremental reads rows where a change date is greater than or equal to the last value read. That greater-or-equal comparison can bring back the same rows twice, so build deduplication into the pipeline from day one
- SLT replication captures inserts, updates, and deletes through database triggers, which is the route when operations teams need near real-time data. From connector support pack 34, SLT through the add-on also needs the PALODATA component
One gotcha we see constantly. Tables without a reliable change date. Many SAP tables don’t have one, or only the header table does while item tables don’t. Your options are SLT, an extractor, or periodic full loads of a filtered range. Decide per table, and write the decision down.
Deletes are the other gotcha. Timestamp reads don’t see deleted rows. If a deleted sales order line matters to your numbers, you need SLT or an extractor that reports deletions.
Our Pipeline Builder guide covers how to build the deduplication and incremental logic on the Foundry side.
Does SAP Note 3255746 Affect Palantir’s Connector?
It affects one specific route, not SAP integration in general. The note restricts third-party use of ODP through RFC. SAPinsider reports tables, CDS views, BAPIs, function modules, and DeltaQ remain usable. Check your connector’s extraction method with Palantir and SAP.
The background. SAPinsider’s 2026 explainer says the note limits the ODP RFC interface to data transfer between SAP applications. Version 11 was released in April 2026, with a June 2026 patch that blocks unauthorized calls, and an opt-out available to the end of 2026.
We haven’t found Palantir documentation that addresses the note directly. So we won’t claim it does or doesn’t affect your setup. What we’d do is ask Palantir which extraction methods your configuration uses, and ask SAP or your account team whether that path is covered by your licensing. Get both answers in writing before your 2027 planning, not after.
This is exactly why SAP Data Accelerator exists, we suspect. An SAP-endorsed path removes a lot of these questions. Watch its move from beta.
What About Licensing, Authorizations, and System Load?
Licensing is your responsibility. Palantir’s FAQ says scheduled extraction may count as indirect static read under SAP licensing, and writeback may need additional licenses. Authorizations need security review, and the connector checks system load before each page it reads.
The Palantir SAP FAQ says plainly that SAP license compliance is the customer’s responsibility. Read that sentence twice. Indirect access has been a costly topic for SAP customers, and moving data into another platform on a schedule is exactly what auditors look at. Talk to your SAP account team early. It’s a dull meeting. It’s also cheaper than a surprise true-up.
On system load, the same FAQ says the connector checks CPU, memory, and available work processes before reading each page of data. That helps. It doesn’t replace scheduling big historical loads outside the month-end close.
Authorizations are the other half. Foundry’s technical user can read what its roles allow, so restrict it to the company codes, plants, and tables the project needs. Once the data lands, apply Foundry markings to sensitive fields like pay data from HR or bank details from vendor masters. Our data governance guide covers how.
Where Do Business Data Cloud and the 2026 SAP Palantir Tooling Fit?
They’re complementary, not replacements. SAP announced Palantir AIP for data migration as an SAP Endorsed App in May 2026, and Palantir connects with SAP Business Data Cloud. Many companies will use SAP’s data platform for SAP analytics and Foundry for cross-system operations.
The May 2026 SAP News announcement said Palantir AIP for data migration became available as an SAP Endorsed App on the SAP Store, with a further integration planned for general availability in the third quarter of 2026. Check its status before relying on it. Separately, SAP NS2 and Palantir announced Business Data Cloud connectivity for defense customers in 2025.
So when do you use which? Here’s how we’d split it.
- SAP reporting and planning that stays inside SAP fits SAP’s own data platform well
- Decisions that join SAP with MES, telematics, CRM, external suppliers, or IoT data fit Foundry better
- Migration from ECC to S/4HANA is where the two meet, and where AIP for data migration is aimed
We’ll say the self-serving part out loud. We build on Foundry. But we’d tell a client with a pure SAP finance reporting problem to look at SAP’s tools first.
Which First Extraction Scope Should an ECC Customer Pick?
One business process, end to end. Pick something like procure-to-pay or plant maintenance, extract only the objects that process needs, model them in the Ontology, and ship one app. Design it so the same Ontology survives the move to S/4HANA.
Here’s the sequence we use.
- Choose one process with a clear owner, such as late purchase orders or open maintenance orders
- List the SAP objects that process needs, starting with extractors or CDS views, then tables for gaps
- Install and authorize the connector in development, quality, and production
- Decide the delta method per object, and document tables with no change date
- Land the data, deduplicate, and reconcile counts and totals against SAP reports
- Model business objects in the Ontology, like Supplier, Purchase Order, and Material, not table names
- Build one app the process owner uses weekly
- Map each Foundry object to its S/4HANA equivalent before your migration starts
Step 8 is the one teams skip. Do it, and your S/4HANA cutover becomes a pipeline change instead of an app rebuild. Our Ontology explainer and supply chain guide show the kind of object model we mean.
Connect a Process, Not the Whole ERP
Three takeaways. Palantir SAP integration runs through an SAP-certified add-on with direct, SLT, and gateway patterns, and SAP Data Accelerator is a newer endorsed path. Extractors and CDS views beat raw tables because they survive the ECC to S/4HANA move. And licensing, authorizations, and Basis team time decide your timeline more than any pipeline does.
PAL4C Labs is an independent, Palantir-exclusive consultancy that connects ERP systems like SAP to Foundry and builds Ontologies and AIP workflows on top. We’re not affiliated with Palantir or SAP.
Connecting SAP to Foundry this year? Send us your SAP release, whether you’re on ECC or S/4HANA, and the first process you want to improve. We’ll draft the extraction plan with you, object by object, with delta methods and the questions to settle with SAP before go-live.
SAP and Foundry Questions Data Teams Ask
Can Foundry Connect to SAP ECC, or Only S/4HANA?
Both. Palantir’s SAP ERP connector supports on-premise ECC and S/4HANA, including S/4HANA Cloud Private Edition, plus BW and SLT. Systems too old for the add-on can connect through a gateway system.
Do We Have to Install Anything on the SAP System?
You do. The Palantir Foundry Connector 2.0 is an SAP-certified ABAP add-on installed in client 000 through SAINT, with SICF services activated and a system-type technical user assigned Palantir’s roles.
What SAP Objects Can Foundry Extract?
Tables and views, including cluster, pool, and custom Z tables. ABAP CDS views. Business Content extractors. Function modules and BAPIs. From BW, DSOs, InfoCubes, InfoObjects, and BEx queries.
Does Foundry Support Delta Loads From SAP?
Through extractor delta, timestamp-based incremental reads, or SLT replication. Timestamp reads can return duplicate rows and miss deletes, so plan deduplication and use SLT or extractors where deletes matter. Function modules and transaction codes are full snapshots every time.
Does Pulling SAP Data Into Foundry Trigger Indirect Access Fees?
Possibly, and that’s for you and SAP to settle. Palantir’s own FAQ says license compliance is the customer’s responsibility, that scheduled extraction may count as indirect static read, and that writeback may need additional licenses. Raise it with your SAP account team before production.
What Is SAP Data Accelerator?
An SAP-managed connectivity service, running through SAP Cloud Connector, that SAP describes as its preferred path for getting SAP data into Foundry. Palantir labels it beta, and it changes only the network route, not what the connector can extract.
Should We Wait Until After the S/4HANA Migration to Connect Foundry?
Usually not. Connecting ECC now gives you value during a migration that may take years, as long as you model the Ontology on business objects. Then moving to S/4HANA means remapping pipelines, not rebuilding apps.