MBD vs 2D: The Short Version
MBD vs 2D is whether 3D PMI is the deliverable, or the drawing still wins. inteliScience runs that choice with methods, quality, and IT, not as a ribbon toggle. December 2025 NX added PMI Annotation Plane Prediction. The engineer still confirms. A DACH plant that must ship a PDF is not behind. It is honest about the contract.
| Question | Drawing master | MBD master |
|---|---|---|
| What purchasing signs | Sheet, revision in the title block | Model revision plus PMI |
| What CAM/CMM consume | Views, notes, a DXF | Semantic PMI, if the downstream stack can read it |
| What NX already does | Drafting views, templates | PMI objects, model views |
| What AI added (Dec 2025) | Inherit on some views | Annotation plane prediction |
| What still fails | Stale views | Pretty planes, wrong GPS |
| Who should pick | Methods + quality + the customer | Same people, plus IT for Teamcenter state |
The Problem: A New Button Is Not a Policy
“We switched to MBD” often means someone turned on PMI and still emails a PDF. Or the opposite: a 3D master that purchasing cannot open.
This article is the decision. It is not how to place a section view. It is not aerospace FAI packaging. Those are other jobs.
Siemens’ AI-enabled design write-up (28 January 2026, December 2025 release) is blunt: setting annotation planes and model views by hand is tedious and inconsistent. Plane Prediction assigns a plane per PMI object during automated authoring. Out of the box or from your own data mining. Inconsistent PMI, they say, costs downstream manufacturing. We agree with the pain. We do not treat the prediction as a pass.
Live Share on the June 2026 Designcenter release lets more than one user edit PMI in the same session. Governance gets harder, not easier.
When the drawing should stay the master
Keep 2D when the customer, the notified body, or the shop system still requires a sheet. Sheet metal plants that nest from a flat view plus a bend table are in this camp. Inherit Flat Pattern PMI is a drafting preference, not MBD. The unfold itself is NX sheet metal.
Keep 2D when the CMM program is still written from views. Semantic PMI into inspection is a project, not a checkbox.
We advise against a big-bang “drawings die on 1 October.” Run one commodity as MBD, keep the rest on templates, measure the rework.
When MBD is the better master
Move when CAM, CMM, and suppliers already consume 3D, and the GPS rules fit in Check-Mate or NXOpen, not in a Word file. Aerospace programs that already treat PMI as characteristics are the obvious case. That pipeline, including FAI, is the NX Open aerospace automation story, different keyword, different buyer.
Designcenter Copilot (documentation-grounded; Siemens has named models such as GPT-4.1 and Claude Sonnet 4.5 on the Copilot stack) can tell you where the PMI command lives. It does not know your datum scheme.
Our clear recommendation: MBD vs 2D is a contract and a Teamcenter state, then a template. Tools second.
# After PMI exists: count PMI objects on the work part (NXOpen, NX 2606).
import NXOpen
session = NXOpen.Session.GetSession()
work = session.Parts.Work
pmi = work.PmiManager
print("PMI manager:", pmi is not None)If that manager is None you are not in a PMI-ready part. Buying a Copilot seat will not fix it.
What We See in Practice: MBD Pays Off When You Commit
The critics are right about one thing. Many companies that “have MBD” ship PDFs anyway, because their suppliers cannot read semantic PMI. We see it in DACH supply chains every month: a fully annotated model in Teamcenter, and a drawing exported from it because the RFQ, the supplier’s viewer or incoming inspection still expects a sheet. That is not a failure of MBD. It is a supply chain that has not moved yet.
Where the reader is your own shop, the calculation changes completely. Companies with in-house manufacturing control the whole chain: CAM, CMM, work instructions, the viewer on the shop floor. For them MBD pays off strongly and fast, because nobody downstream has to be convinced. The model that was designed is the model that is machined and measured.
MBD only works with commitment. Half MBD, a model with PMI plus a drawing maintained by hand next to it, is the worst of both worlds: two truths, two revisions, two things to check. Either the model is the master and every sheet is derived from it, or the drawing stays the master and PMI is a helper. Deciding is the work. Drifting between the two is what costs money.
Take the long view. Not long ago we drew everything by hand. Then 2D went digital, and we kept the paper logic on a screen. Everything in the model is the only consistent next step. Why maintain a 2D sheet and a 3D model, two artefacts for one part, when the model can carry every dimension, tolerance and note itself?
The model is also easier to read. Model views attach each annotation to the geometry it belongs to, so a machinist or an inspector sees the tolerance on the face it applies to instead of interpreting three projections and a section. Fewer readers misinterpret, fewer questions come back from the shop, and understanding the part no longer requires a drafting education.
Our recommendation from that experience: if you manufacture in-house, commit to MBD and derive whatever sheet a customer still demands from the model, never by hand. If you depend on suppliers who read PDFs, make the model the master internally, ship the derived PDF as long as the contract requires it, and write the date by which suppliers must read PMI into the next framework agreement. The transition then has a direction. Without a date it never ends.
What Plane Prediction does not decide
OOTB machine learning has not read your house GPS. Automated authoring without a check produces neat, wrong planes. 2D-to-3D PMI as a delivered service is not an inteliScience offer. A how-to on placing PMI is traffic, not a lead.
Checks come after the choice: every PMI has a tolerance, a datum, no orphans. The engineer confirms. The model does not “pass.” Ready to implement those checks, not the slogan? The service page in Related Topics is the how. This page stays the choice.
What to look for before you pick a master
- The contract. If the RFQ names a drawing, the drawing is still master.
- The reader. CAM, CMM, and the supplier must consume 3D PMI, or you are decorating a model.
- Teamcenter state. Which revision is released, model or sheet.
- A checker. Plane Prediction is not Check-Mate.
- A pilot commodity. One part family, not the whole vault.
- No drawing generator. Templates and completeness checks only.
Frequently Asked Questions
What does MBD vs 2D mean in NX?
Whether Product Manufacturing Information on the 3D model is the legal / process master, or the drafting sheet is. NX can do both. Your customer and your CMM decide which one you may drop.
MBD vs 2D for a DACH job shop?
If the RFQ still asks for a PDF drawing, keep the drawing master and put PMI on the model as a helper. Switching the master without switching the contract creates two truths.
Does PMI Annotation Plane Prediction replace a checker?
No. December 2025 NX suggests planes. Your ISO GPS or company PMI standard still needs a deterministic check and a human confirm.
How much does an MBD program cost?
The NX PMI license is a Siemens quote. The real bill is templates, Teamcenter release rules, and downstream readers (CAM, CMM, suppliers). Treat it as a methods project, not a weekend of Model Views.
MBD vs 2D versus “we generate all drawings with AI”?
We do not sell a drawing generator. Automation here is templates, view checks, and PMI completeness. The sheet is either required or it is not.
Next Step
Want a 60-minute read on whether MBD vs 2D is even allowed on your contracts? Book a free strategy call. Book a strategy call about MBD vs 2D
