← All forecasts · October 19

SAP / Q3 2026

SAP SE

Reported: October 21, 2026

Published forecast

2026-10-20T05:14:22.297197Z

Opus 5.5 High in Shortcut

Shortcut

1.77

Shortcut normalized EPS

-3.5% vs Street

Off by €0.12 per share

Street

1.83

Street normalized EPS

Frozen October 18 snapshot

Off by €0.06 per share

Real

1.89

Issuer non-IFRS basic EPS

Street win

Reported result · October 21 ↗ · Non-IFRS basic EPS was €1.89; IFRS basic EPS was €1.96. Both forecasts remain in EUR per ordinary share, without USD conversion.

EUR per basic ordinary share. Percentages use unrounded forecasts.

Model download

Explore the assumptions and calculations behind the forecast.

Open in Shortcut ↗Download XLSX ↓

Shortcut’s reasoning.

The scored forecast is EUR non-IFRS basic EPS per ordinary share. All six frozen normalized historical actuals match SAP basic EPS despite the raw CapIQ diluted label. The IFRS diluted comparison uses a different share basis; the USD ADR memo is not the scored forecast.

Non-IFRS net income of EUR2,029.707026636 million divided by 1,149.582 million basic shares gives EPS of EUR1.7656043906706962, against frozen Street EUR1.83. Revenue is EUR10,103.19955 million, with EUR2,900.9468904 million of non-IFRS operating profit.

MEDIUM source and modeling limits remain in cash-settled compensation: the EUR185.66 September30 price lacks an archived pre-cutoff capture, and the sensitivity coefficient is judgmental. Only 78% of the incremental EUR87.822 million remeasurement is explicitly linked to operating expenses, an actual EPS effect of about EUR0.04261; the earlier EUR0.05462 full-allocation example is not the current-model effect.

The EUR116.78 million cost-of-revenue SBC allocation is assumed embedded in margin and cost ratios, but that embedding is not independently reconciled. An additional EUR12.78 million charge would lower EPS about EUR0.008 only if not already included. No double charge, financial-input retuning or EPS change was made during review.

The issuer-published EUR5,852 million constant-currency first-half operating profit is retained even though independently rounded components EUR5,609 million and EUR242 million sum to EUR5,851 million. Regional FX diagnostics, share-opening proxies, historical EUR1 million rounding gaps, source capture and header/readability limits remain disclosed.

On the record.

Reporting basis
non-IFRS basic EPS; six frozen normalized actuals match basic
Forecast completed
2026-10-20T00:59:18.460477+00:00
Published
2026-10-20T05:14:22.297197Z
Model assumptions and review notes
  • MEDIUM: The EUR185.66 September30 close was captured after the common cutoff without an archived pre-cutoff copy; the 1.7 SBC sensitivity is judgmental. The actual linked P&L allocation is 78%, so the EUR87.822m incremental remeasurement changes current-model EPS by about EUR0.04261. The prior EUR0.05462 figure in review/narrative material is a hypothetical full allocation, not the current linked EPS effect.
  • MEDIUM: The EUR116.78m cost-of-revenue SBC allocation is not separately charged by a formula and is assumed embedded in the cost/margin ratios. Whether the increase from EUR104m is fully absorbed is unproved; an additional EUR12.78m charge would lower EPS about EUR0.008 only if not already included. No double charge or margin retuning was made.
  • LOW: Regional and segment FX/constant-currency/organic diagnostics use inconsistent illustrative allocation conventions. They do not drive reported-revenue forecast EPS; derived guidance comparisons are approximate and unscored.
  • LOW: The issuer explicitly reports 1H26 constant-currency non-IFRS OP of EUR5,852m while independently rounded components EUR5,609m and EUR242m sum to EUR5,851m. The published total is retained. The note equation is not an exact integer reconciliation; quarter-summed OP and annual OP also have EUR1m rounding differences.
  • LOW: The basic-share opening proxy and treasury-reissue timing remain uncertain; each additional1m basic shares would lower EPS about EUR0.00153. The market-cap source attribution and index-membership inference remain imperfect, though the large-cap bucket is robust.
  • LOW: Some secondary Cover memo headers/long notes, branding context bands, CET-versus-CEST wording and the displayed EUR/USD precision remain presentation limitations. Only EUR non-IFRS basic EPS is scored; the raw frozen CapIQ diluted label is preserved with its six-quarter basic-EPS mapping.
  • LOW: The exact supplied forecasting-playbook.md is v2 experimental, SHA1f08987..., distinct from the frozen v3 SKILL object. The probability constants/overlays are heuristic and do not multiply EPS.
  • LOW: Historical Income Statement versus YTD-derived cash-flow tax, net financial income and SBC have EUR1m differences. The current forecast EPS build is independent of these historical rounding gaps.
  • LOW: Cover and Inputs wrapped column headers use fixed15pt rows and some Sources descriptions remain clipped; long Cover notes overflow. Raw text is preserved, and no native Excel presentation acceptance is claimed.
  • The public export differs in raw bytes through serialization: all8062values/caches4230expandedformulas match the accepted workbook, with1709formula-prefix differences. Exact default-style normalization proves semantic equality; column widths can differ by1/256character and default widths serialize differently.
Research sources