Finding a pot of gold, any pot of gold, at the end of the rainbow-colored cloud
I recall when a large CAD vendor first enthusiastically announced it was going to embrace the cloud. Well, it was going to get us to embrace the cloud. This was at a user conference at which we were told that the cloud represented “infinite computing.” The ceo even used the phrase a few more times over the following years in interviews, adding that the only way to get all of his company’s software would be online, by 2014-5.
It’s 2023, and the CAD company has spent the last decade struggling to cloud-ize all its software. The flagship design software becomes a pale imitation of itself when users fire up the cloud version. The new timeline is 2028. It’s a mystery why it's so hard.
The Mystery Explained
Except it is not. It's no longer a mystery why a Salesforce or a Gmail can be good enough running in a Web browser, but CAD cannot be. The first two concern themselves with accuracy and interactions in text; CAD is concerned with accuracy and interactions with 3D graphics, and so there is so much more complexity between the computer screen displaying the Web browser and the code running on multiple servers thousands of miles away.
We know now that cloud computing is not infinite; it suffers from more constraints than desktop computing. One of the first cloud processing jobs offered by this particular CAD vendor was rendering. I submitted a 3D model, expecting the results back in seconds -- or perhaps even more quickly! Several days later... I finally received the email telling me the image file was ready for me to download.
It turns out that, 23 years after Alibre released the very first cloud-run CAD (and then, in an act of foreshadowing, gave up on it), that there are just a couple cloudCADs that are pretty good, including Onshape for 3D modeling and Graebert Kudo for 2D drafting. Both were developed by scrappy underdogs, fighting to prove themselves against the big four -- Dassault, Siemens, Autodesk, and PTC.
CloudCAD suffers from customer inertia (CAD is best done with desktop computers, an inalienable fact) and from inert complexity. The complexity revealed by CAD vendors themselves, as they unveil a cloud strategy, abandon it after a couple of years, and then replace it with the new one. We saw zigzagging like these ones:
- Dassault Solidworks -- from Solidworks V6 to X-series to 3dexperience
- Autodesk -- from Quantum to Project Plasma to Forma/Flow/Fusion
- PTC -- from Atlas to +
Only Siemens is smart enough to not put its CAD programs in the cloud.
The Insanity of Cost
On top of the lack of customer interest and inability to pin down a cloudCAD system, CAD vendors suffer from insane cloud costs. They hope to lock in customers through the cloud (and so keep prices increasing annually); the ceo of PTC said this week, “You can't switch from it; you can't stop using it; you can't stop paying for it -- I mean, unless you're winding your business down." But so do the Amazons and Microsofts who provide cloud services to cloud-obsessed CAD vendors: they lock them in, with attendant price increases. Goose, meet gander; pot, meet kettle.
The high cost of cloud computing is why CAD vendors are also zigzagging their way through billing tactics: subscriptions, tokens, pay-per-use. Tokens are a particularly opaque system of payment, as CAD vendors charge varying numbers of tokens with no clear link to the $$$ cost.
As a result, there is an entire sub-industry devoted to reducing horrendous cloud bills charged by Amazon, et al, and even developing tactics in devolving the cloud. One software vendor changed his $3-million annual Amazon bill to $60,000 by switching to Dell’s highest-end workstations, on-premise. Amazon this week reported growth slowing for its cloud division, from increasing 40% annually to 20% last year: "Businesses that would use the service are being more cautious and looking for ways to save money.”
What this means for end-users is that their intuition is correct: CAD runs best on the desktop. As the ceo of Autodesk himself said in an interview last year, there's just never going to be a Revit on the cloud.