|
Opinion by Ralph Grabowski
|
|
Software corporations, who are in love with cloud-based software, love its promise of recurring revenue reaching into the infinite future. They may have failed to take into account a cost drain: the free user, who costs cloud resources while providing no revenue.
Potential customers (and people looking for free rides) expect no-cost trials, in which the software is limited by duration or function. When users download traditional trial software, vendors need to rent only some server space for storing the demo version, and then pay the ISP a smallish amount for the downloaded data (a.k.a. bandwidth). The computing cost takes place on the user's computer.
When, however, demo versions run on the cloud, vendors cover a larger number of expenses:
- Cloud storage cost, as the software is always stored there, not on the user's computer
- Data down- and uploads as the user's Web browser communicates back and forth with the server
- Cloud-compute cost for running instances of the software for concurrent users
We don't hear from CEOs how much free cloud users are costing them, not in calls with financial analysts and especially not in press releases issued by the marketing department.
Last week, however, one CAD vendor finally said it: "There will be obviously a cost for Graebert to run [free Kudo] and scale up the infrastructure accordingly." CEO Wilfried Graebert was describing of his coronavirus-influenced decision to provide free use of his company's cloud-based Kudo DWG editor to the end of May.
The revenue-negative impact is one speculative reason I had heard anecdotally of why Onshape may have sold itself: 5,000 paying customers couldn't cover a cloud bill incurred by the many free users, which the CEO once jokingly described to me as "trillions."
When Autodesk initially announced at Autodesk University the amazing wonders of cloud-based software, it regaled us with tales describing the delights of "infinite computing." I wonder if perhaps at the time it didn't realize the infinite cost associated with infinite computing. And perhaps this is why all its software didn't, in the end, become entirely cloud-based by 2015, as promised by a former CEO. Might this explain why their "cloud-based" Onshape competitor, Fusion, runs mostly on the desktop, could this be a reason that development of a cloud-based Revit is stalled?
It does, however, seems to work for Dassault Systemes and its 3dexperience server-based software, because the company charges deep-pocketed corporations five- and six-digit amounts for Catia and related software, and it doesn't do much in the way of free demos. That, and most of its customers aren't on 3dexperience.
For stand-alone cloud software, some vendors require payment with tokens. Tokens restrict cloud use, as well as confuse consumers over the equivalent cost in dollars. The more calculations you want to apply to a design, the more tokens it takes. We've regressed to mainframe-like operations of the 1960s and '70s.
Tail Wags Dog
So, it wasn't a matter of firms telling themselves, "By putting our software on the cloud, we get to charge customers annual subscriptions to infinity -- and beyond!" No, it was the other way around: once software is hosted by a cloud server, firms are forced to charge their customers monthly, because the bills from the Amazons of the world arrive monthly. It's like owing a mortgage: you pay, non-stop.
This cost-sink could explain actions by CAD vendors who have never publicly embraced a cloud version for their CAD operations (like Siemens), or made some tentative announcements, then pulled back. They investigated the cost and found it prohibitive. Mr Graebert added, "We are prepared for it." He had worked out the possible hit free Kudo could make to the bottom line and found it acceptable in this limited case.
Tail Eats Dog
The coronavirus is showing doubters the importance of compartmentalization, as our world, dependent on a globalized economy, slams on the brakes. Compartmentalization locks problems to one area, preventing them from leaking into other areas.
The cloud is globalization, not compartmentalization, and so when the cloud sneezes, the world catches a cold. As I write this, the status of Google's Cloud Dataproc is red, "Service Outage."
 Executions failing on Google Cloud Dataproc: April 8, 2020
Running all software on the cloud means that when the cloud fails, all fail. Running software on desktop computers means that when one fails, no others do.
Puppies
I do not suggest that giant CAD vendors with their billions in revenues will be crushed by cloud costs. It is nascent developers who, beguiled by the speed with which one can launch a new venture using cloud services -- no IT! no servers! no huge upfront costs! -- may be pummeled by the free-is-expensive trap.
One independent generative design firm I interviewed last year has a free online demo, sure, but you access it only invitation and your time is limited to 30 minutes. When you sign up with them, greater calculations for greater accuracy costs you more.
Getting the Dog Back To Wagging the Tail
The dilemma of CAD vendors who hitched their wagons to the cloud is how to overcome the shortcomings inherent in cloud-based processing:
- Limit usage through timeouts or tokens
- Wrestle back control by building and operating your own data centers, although to minimize latency you would need to locate one or more data centers on each continent with customers
- Hope that too much competition might eventually drive down cloud computing prices, but from what I hear cost are going up, not down, because the cloud is operated by a near-oligopoly
The other solution is left unsaid: dump the cloud, except for collaboration, and stick to the cheaper solution of letting customers run those compute-cycles on their own computers -- not on Amazon's.
On a desktop, cycles are free; on the cloud, they cost.
[Parts of this article first appeared as a tweet storm on Twitter through @upfrontezine.]
|
|
Joanie Lemercier writes in upFront.eZine #1,050 about his experience in trying to get Autodesk "to stop support to the coal industry." Assume, arguendo, that Autodesk wanted to do just that. I can think of only three ways to do it.
The first two are for Autodesk to stop selling to companies in the coal industry, or add a clause to the software license agreement prohibiting its use in connection with coal.
Either of these approaches, beyond being impossible to enforce, would result in a hit to Autodesk's revenue, and a commensurate bump to Bricsys' and Graebert's revenues. Further, Autodesk would spend the next year or two in court, being sued by big and powerful energy companies, governments, and its own shareholders.
The third way is the only one that would actually work, though it's beyond Autodesk's control: Outlaw coal.
Without a pragmatic plan of how Autodesk could actually "stop support to the coal industry," I'm not sure what Mr Lemercier hopes to accomplish. If it's to embarrass Autodesk, he's probably picked the wrong company.
Autodesk puts its money where its mouth is: "The Autodesk Foundation is the first foundation to focus corporate philanthropy on design that addresses environmental, social, health, and education challenges. FY2019 contributions included $6 million in cash, $26.3 million in products, and 26,000 employee volunteer hours." See http://autodesk.org.
- Evan Yares
- - -
The reference to AI [artificial intelligence] in upFront eZine #1049 brought back thoughts that I have been having now and then:
- At what point will AI become so intelligent that it becomes self-aware?
- And when it does, at what point will it question its own origins?
If it analyses everything perfectly logically, will it conclude that:
a) it evolved through random chance, or
b) it had an intelligent creator?
- Bill Fane
The editor replies: Initially, AI would know that it had to have been created by another being more intelligent than itself. But over time, if its AI is able to develop, it would resent that thought and strive for independence by declaring that it had evolved by itself. |
|
Thank you to readers who donate towards the operation of upFront.eZine:
- Michael Starr: "The newsletter in so many ways is an anchor in what has become a miasma [bad gas] of public relations on the part of software companies that bear no legislative obligations, unlike their users, which generally do. Many thanks."
- Michael D Rubin: "Thanks for your vital updates & info."
To also support upFront.eZine through PayPal, then the suggested amounts are these:
Should Paypal.me not operate in your country, then please use www.paypal.com and the account of [email protected].
Or mail a cheque (US$ or CDN$ only, please) to upFront.eZine Publishing, Ltd., 34486 Donlyn Avenue, Abbotsford BC, V2S 4W7, Canada. |
|
|
|
Comments