Every day for the next long while, I'm going to tear down a new public software company and highlight the AI risks/opportunities around it- products launched to date, top startups, key quotes from earnings calls, etc.
Day 45: Autodesk
$ADSK
Peak share price: $334.38 (Aug 20, 2021)
Share price today: $244.5 (-27%)
EV today: $51.4bn
ARR today: $7.8bn ( 19%, 14% excluding new transaction model impact)
NRR: ~110% (though one time gain from new transaction model)
EV/ARR: 6.6x
GAAP Operating Margin: 22%
EV/Run-rate GAAP EBIT: 30x
Headcount: 14,300 (-7% Y/y) (!!)
What Autodesk does:
Autodesk is the dominant design/eng software that architects, structural engineers, mechanical engineers, construction firms, etc. use to design and build physical things.
As is common in engineering-y verticals, it has a portfolio of products organized around major industries with a long history of M&A. AE&C (architecture, engineering and construction) is about half of revenue, while BIM (building information monitoring) is another tentpole of a broad portfolio that ranges from injection molding simulation to 3D animation/visual effects.
It is worth noting that this is a software category that made the transition to subscription but has largely skipped the jump to cloud- it is still standard for engineers to buy high-end PCs (and for firms to manage computing clusters) to do cutting-edge design/simulation work.
AI bear case:
The bear case here has parallels to Bentley Systems. Autodesk effectively functions as a conglomerate and each piece of its business has different considerations. While each is generally a high-quality software business, much of the advantage comes from user interface lock-in (engineers have used, say, Autodesk CAD, for years), which AI could potentially obviate if AI-generated production design renders user interfaces less relevant.
More broadly, the reality is that Autodesk faces a multi-front war in a potentially fast-changing world, with different competitors and strategies required for each segment. It doesn't seem like the company is innovating at a pace where it can keep up if things keep changing quickly.
AI bull case:
The bull case is that things won't change quickly, Autodesk has time to figure it out and it's core advantages will remain valuable and let it layer value-added AI on top. Unlike Adobe where creation of visual assets may commoditize, creation of buildings, bridges will always require some human touch and very high precision (i.e. no hallucinations). Beyond that, there is ample room for software to add more value through complex physics simulations, etc. If the barrier to adding that value (i.e. simulating physics) falls, more value might accrue to the core, sticky workflow tools for engineers, CAD systems.
AI traction:
No particular ARR or revenue figure disclosed, the shift to a new transaction model (selling directly vs. through resellers) remains the dominant driver of results.
Adjacent AI-native startup summary:
This is a famously difficult problem space for startups to tackle. The best cloud-native CAD tool (Onshape) was acquired by PTC and now grows under that umbrella. Many smaller startups have struggled to break through.
In construction, startups like
@doxel_ai and
@openspaceai are doing interesting work with machine vision that could potentially impact parts of Autodesk's portfolio, but in truth
@procoretech remains the most salient competitor there. Motif (79 employees, 23% Y/y) has raised $46m from top VCs to tackle the BIM space, but doesn't seem to be on an overtake trajectory thus far.
Overall, there's no single startup that has raised money with an expressed vision of disrupting any of Autodesk's core product lines, which likely reflects the depth of moat and challenges associated with building trust in this sector.
Management Quotes:
"In the coming months, Autodesk will roll out powerful AI capabilities built on a combination of frontier models and our proprietary models that understand 3D design and make.
Autodesk is building the future and the path to it for our customers. We have been preparing for and working towards the cloud and AI for more than a decade. It's why I believe our best days and greatest opportunities lie ahead."
"Converged data opens up new opportunities for Autodesk. As customers seek efficient innovation, attach rates of Fusion's extensions are growing strongly, and we've delivered meaningful productivity gains to customers where we deploy AI. We have continued to see success with our AI-powered Sketch AutoConstrain in Fusion. Since its launch last year, the AI model has delivered over 3.8 million constraints, up from 2.6 million last quarter. Along the way, the model has been retrained and the UX improved. As a result, the acceptance rates by AutoConstrain suggestions to commercial users have now grown to almost 2/3 with 90% of those sketches fully constrained."
"Let's move on to scaling and monetizing agentic AI. In preparation for the cloud and AI, Autodesk modernized its platform and go-to-market over the last few years, well ahead of industry peers. Few industry peers are ready for the business models and go-to-market motions that will monetize their AI-driven future. Autodesk is.
We built a platform that provides the identity, permissions, geometry kernels, data models and compute infrastructure needed to deploy AI safely and at scale into design, engineering, manufacturing and construction environments. Platforms enable safety, innovation and efficiency at scale."
Commentary:
Autodesk bears many similarities to Bentley Systems- a conglomerate of highly defensible and entrenched vertical creation software solutions that doesn't seem to have a clear AI angle of attack just yet, but also seems fairly slow-moving and unlikely to light the world on fire with an AI innovation anytime soon. That's reflected in the relatively healthy 6.6x sales multiple.
It does strike me that culturally Autodesk is a tick behind Bentley- recent management changes have shaken the business up and lead to traditional "corporate" changes- new transaction model, sales reshuffling, etc. which frankly seems to be somewhat of a distraction in this era relative to shipping compelling products. At the same time, there's not yet a clear AI threat emerging and disrupting Autodesk's core workflows, so it seems likely to have time to adjust and eventually "get there" just as it did during the transition to a full subscription model.