Analytics Legends logoAI & Analytics Legends Sign inSign up
Guide

SAP Datasphere pricing explained: how to read the capacity-unit rate card

As of 2026-11-07

SAP does not publish a currency price for Datasphere; its pricing page lists it as price on request. What SAP does publish is the consumption mechanism: a rate card in the Business Data Cloud service description guide that converts every hour of compute, every gigabyte of storage and every tenant into capacity units. The currency figure comes from your contract; the capacity-unit arithmetic is public, and it is the part you can check before you negotiate.

Where Datasphere sits in SAP's 2026 commercial model

Datasphere is no longer priced as a free-standing product. In a presentation dated 16 April 2026, SAP's own commercial lead for Business Data Cloud describes a single subscription, Business Data Cloud core capacity, that covers Datasphere, SAP Analytics Cloud, BW Cloud, SAP HANA Cloud, BDC Connect for partners and SAP Databricks under one commercial model, with the capacity unit as the metric. Capacity units can be applied to any of those services in varying quantities.

The consequence for a buyer is that a quote headed 'Datasphere' is really a pool of capacity units that Datasphere draws on, alongside whatever else you run in the same subscription. When you compare a Datasphere offer with a standalone product, you are comparing a pool with a product, and the first question is what else is drawing from the pool.

SAP states that this core capacity is always sold together with at least one intelligent package, the line-of-business bundles built on top of it. That matters in negotiation: part of what you pay is an attach quantity SAP has sized for the package, not a figure you derived from your own workload.

How the bill is computed, step by step

The rule in SAP's service description guide (version 9-2026) is short. For each capacity service there is a unit of measure and a capacity-unit value. You multiply the amount of the unit of measure used in a month by the value, and the result is the capacity units consumed by that service. You may consume up to the total stated in your order form each month. Unused units are not carried over, and units are not prorated for partial months.

Two details change real budgets. First, for the services counted in capacity units per month, usage is measured at its highest point in the month, so a one-day spike sets the month's consumption for that service. Second, memory and storage are counted by the hour: SAP's own calculation example is 256 GB of disk storage (16 blocks of 16 GB) held for 730 hours at 0.004 per block-hour, which gives 46.72 capacity units.

That 730-hour convention is the one to carry into every estimate: a month is treated as 730 hours, so any hourly value in the rate card multiplies by 730 for an always-on service.

The rate card: what each service costs in capacity units

These are the capacity-unit values SAP prints for Datasphere (service description guide 9-2026, Table 2), quoted with the unit of measure they apply to.

Datasphere SaaS Foundation: 0.926 per tenant per hour. Memory-optimised compute: 0.570 per 16 GB of memory per hour. High memory-optimised compute: 0.485. CPU-optimised compute: 0.715. High CPU-optimised compute: 1.065. Elastic variants: 0.798 (memory-optimised), 1.028 (CPU-optimised), 1.488 (high CPU-optimised). Spark compute: 0.149 per 4 GB per hour. Object store: 0.103 per terabyte per hour. Disk storage: 0.004 per 16 GB per hour. Data integration: 0.784 per 5 GB of memory per hour, with 200 hours included each month. BW Bridge: 1.447 per 128 GB of storage per hour. Catalog storage beyond the 512 MB included each month: 1.096 per gigabyte per hour.

Premium outbound integration is tiered in blocks of 20 GB: the first 10 blocks cost 500 capacity units each, the next 15 cost 350, the next 25 cost 200, the next 50 cost 100, and further blocks 35. Outbound data to a third-party tool is therefore the line most likely to surprise a team that modelled only compute.

Worked example: what a tenant costs before you load a row

Our arithmetic from SAP's published values, not a quote. An always-on tenant carries the Foundation charge alone: 0.926 x 730 hours is about 676 capacity units a month. That is the floor of a running tenant, before any compute, storage or integration.

Add compute. Four 16 GB blocks (64 GB) of memory-optimised compute held for the whole month: 4 x 0.570 x 730 is about 1,664 capacity units. The same 64 GB on the CPU-optimised profile: 4 x 0.715 x 730 is about 2,088. Profile choice alone moves the compute line by roughly a quarter, which is why it belongs in the design review and not in the order form.

Add storage. SAP's own example gives 46.72 units for 256 GB of disk. One terabyte of object store held all month is 0.103 x 730, about 75 units. Cold history in the object store is cheap against the same volume held as in-memory data, which is the economic argument for tiering, and the same logic underlies the HANA Cloud tiering discussion in the platform's research.

Note what dominates: compute memory, by an order of magnitude over storage. A sizing conversation about terabytes of history is mostly the wrong conversation; the one that sets the bill is how much memory runs, for how many hours, at which profile.

Elastic compute: paying a premium to avoid a bigger baseline

SAP lists elastic compute at higher values than the fixed profiles. The ratio is consistent: 0.798 against 0.570 for memory-optimised, 1.028 against 0.715 for CPU-optimised, 1.488 against 1.065 for high CPU-optimised. That is a premium of about 40 to 44 per cent per memory-hour (our arithmetic).

The trade is straightforward. If a workload needs the extra memory for a few hours around month-end close or a planning cycle, paying the elastic rate for those hours is cheaper than holding the larger fixed profile for 730. If it needs the memory most of the month, the fixed profile wins. The break-even is roughly 70 per cent of the hours (1 divided by 1.4), so measure how many hours the peak actually lasts before you choose.

Because the metric counts the highest point in the month, the elastic hours still register. The platform's own sizing guidance is to provision to measured peak plus a 20 per cent buffer for any tenant with live business-hours users, and to size average-only just for batch tenants with nobody watching a screen. Its roadmap note on elastic capacity bursting at a 1.3 to 1.5 times rate is a preview item that SAP's guide does not confirm, so do not contract against it.

The decision table: which line to attack first

Overnight batch tenant, no interactive users — Lever: size on average load · Watch: the data integration hours beyond the 200 included.

Live business-hours users on analytic models — Lever: peak plus 20 per cent, with an elastic profile for the spike · Watch: the highest-point-in-month rule.

Heavy export to external tools — Lever: redesign before you buy, because premium outbound blocks start at 500 units each · Watch: partner extracts that never appear in the architecture diagram.

BW estate moving across — Lever: put a sunset date on BW Bridge, which carries its own value of 1.447 per 128 GB per hour · Watch: paying for the bridge and the target at once.

Several lines of business sharing one tenant — Lever: allocate the tenant monitor's per-space consumption over a rolling 30 days, agreed before go-live · Watch: a chargeback model that treats shared overhead as free.

Procurement wants the smallest contracted block — Lever: compare the block against measured peak, not average · Watch: unused units expire monthly, so a block bought for the peak is paid for in every month, including the quiet ones.

What sits outside the Datasphere line

Federation is billed by the other platform. Querying an external warehouse or an Iceberg lake in place consumes compute on that side, on that vendor's invoice. A federation-heavy design can be cheaper in capacity units and dearer overall.

Business Data Cloud Connect has its own values: 0.185 per gigabyte for inter-region data transfer and 0.055 per 2,500 API calls. Sharing data with partners across regions is metered, not free.

SAP Analytics Cloud draws on the same pool at per-user values that fall with volume. On SAP's list, business intelligence users run from 25.60 units per user per month in the 25 to 200 band down to 10.54 above 5,000 users, and standard planning from 72.85 to 41.57. SAP's presentation of 16 April 2026 cites a minimum of 640 capacity units as the smallest viable implementation: an Analytics Cloud tenant with 25 business intelligence users (25 x 25.60 = 640). The service description guide v10-2026 does not contain that figure; its clause 6.3.2 gives 1,000 capacity units a month.

AI is a separate meter. Joule agent actions are priced in AI Units, 0.02 per Standard action and 0.05 per Advanced action (SAP whitepaper, version 10.2026), and SAP's own consultant offer page prices an AI Unit at 7 euros in a worked example. That is an example on one product page, not a list price, and it does not convert capacity units. A tenant can be inside its capacity-unit budget and still exhaust its AI Units.

Why no honest page quotes a price per user or per terabyte

The currency price of a capacity unit, the discount, the ramp, the term and any bundle with S/4HANA or RISE are negotiated per deal. A figure you find on a blog is a leaked data point or an estimate dressed as a list price. Datasphere is also not priced per user or per terabyte: the first would ignore memory, the second would ignore compute, and the rate card shows those two are where the cost sits.

The defensible position walking into a negotiation is a workload breakdown by service, a measured peak, and the rate card. That lets you challenge the attach quantity and the profile mix, which is where the money moves.

What we cannot assert

SAP publishes no currency price for Datasphere, so no per-user, per-terabyte or per-capacity-unit euro rate appears. Capacity-unit values come from SAP's service description guide v9-2026 and an April 2026 SAP presentation; worked totals are our arithmetic on those values. SAP's April 2026 presentation (640 units) and guide v10-2026 (clause 6.3.2, 1,000 units) give different entry figures that we cannot reconcile. The sap.com pricing page returned 403 to direct fetch, so its 'price on request' wording rests on a search-result excerpt. Elastic bursting at 1.3 to 1.5 times is a platform roadmap note not confirmed by SAP's guide. Contract term and renewal rules for Datasphere were not verified.

Frequently asked

How much does SAP Datasphere cost per year?

SAP does not publish a currency price; the page lists it on request. The public part is the rate card in capacity units. An always-on tenant carries the Foundation charge of 0.926 per hour, about 676 units a month (our arithmetic), before any compute or storage. Currency depends on your contract.

What is a capacity unit in SAP Datasphere?

It is the metric of the Business Data Cloud subscription. Each service has a unit of measure and a value; usage in a month multiplied by the value gives units consumed. Units can be spread across Datasphere, Analytics Cloud, HANA Cloud and other services in the same core capacity.

Is SAP Datasphere priced per user?

No. Datasphere is priced on capacity services: tenant, compute memory, storage and integration. Analytics Cloud, which draws on the same pool, is the part counted per user per month.

How does SAP measure Datasphere usage?

Per month, at the highest point reached, with hourly values for memory and storage. Units are not prorated for partial months and unused units are not carried over. SAP's own example is 256 GB of disk for 730 hours giving 46.72 capacity units.

Is there a SAP tool to estimate capacity units?

Yes. SAP provides a Business Data Cloud Capacity Unit Estimator, referenced in its April 2026 pricing presentation. SAP Community users report that it does not explain how to size compute and storage, so treat it as a starting point and validate against measured peak.

Does the Datasphere price include Joule or AI?

No. Joule agent actions are metered in AI Units, a separate currency from capacity units. SAP's published rates for its agents are 0.02 AI Units per Standard action and 0.05 per Advanced action, and the two budgets should be sized separately.

Engage the editor

This guide is written by . 27 years of SAP — Datasphere, Business Data Cloud, SAC Planning, Joule. Freelance or permanent · Hybrid / remote · Anywhere in EMEA.

Profile and contact

What this page is built on

External sources

Read next

Alert me to new SAP Datasphere missions

No account needed. We first ask you to confirm by e-mail; nothing is sent without that confirmation, and every alert has a one-click unsubscribe link. Privacy

Open missions

120 live roles and contracts in SAP Datasphere right now — see the SAP Datasphere missions · freelance SAP contracts.