Most Autodesk histories get told through product releases. AutoCAD, Revit, Inventor, Maya, Civil 3D, Fusion, Forma. I’ve watched most of them ship from closer than most people. I was a customer in the early 1990s, spent 25 years inside the company, and today I manage Autodesk software for a large civil infrastructure engineering firm while consulting on design technology on the side. Customer, employee, consultant, CAD manager. Four different seats, same obsession.
The product list isn’t the story, though. The story is that Autodesk kept changing its mind about what the design problem was. First it was getting CAD onto a personal computer. Then it was making that CAD programmable. Then intelligent objects, then models, then coordinated models, then cloud data, then industry platforms, then AI, and now operating the thing after it’s built. Line up 44 years and the pattern is hard to miss. The abstraction layer keeps moving up.
This post runs in two layers. The CEO table and a one-table timeline come first, so you can get 44 years in about three minutes. After that come the seven eras and what I think they actually mean. That second part is the opinionated part.
The CEO Eras
| Era | Leader | What Changed |
|---|---|---|
| 1982-1986 | John Walker | Founder-led startup. Autodesk begins as a multi-product company, but AutoCAD quickly becomes the business. |
| 1986-1992 | Alvar “Al” Green | Startup grows into a professionally managed global software company. |
| 1992-2006 | Carol Bartz | Focus, scale, vertical products, 3D, BIM, acquisition-driven expansion. |
| 2006-2017 | Carl Bass | BIM scales, Digital Prototyping expands, serious moves into cloud, simulation, and generative design. |
| Feb-Jun 2017 | Andrew Anagnost and Amar Hanspal | Interim co-CEOs during the transition after Bass left. |
| 2017-present | Andrew Anagnost | Subscription matures, construction expands, industry clouds emerge, AI becomes strategic, and Autodesk pushes past “Design and Make” into operations. |
Autodesk Timeline at a Glance
| Year | Major Event | Why It Mattered |
|---|---|---|
| 1982 | Autodesk founded; AutoCAD debuts | Professional CAD moves onto the PC |
| 1985 | Autodesk IPO | Capital and scale for international expansion |
| 1986 | Al Green becomes CEO; AutoLISP in AutoCAD R2.1; AutoSketch | AutoCAD becomes an extensible platform |
| 1989 | Generic CADD acquired | Expands Autodesk’s DOS CAD position |
| 1990 | 3D Studio ships | Begins the 3D visualization and M&E lineage |
| 1992 | Carol Bartz becomes CEO | Focus, professional management, global growth |
| 1993 | AutoCAD LT launches; Ithaca Software acquired | Expands market reach; Carl Bass joins Autodesk |
| 1994 | AutoCAD R13 and ObjectARX | Intelligent custom objects and vertical applications |
| 1996 | 3D Studio MAX ships | Major move into professional Windows 3D |
| 1997 | Softdesk acquired; AutoCAD R14 | A much deeper AEC portfolio |
| 1998 | Architectural Desktop and Land Development Desktop; Discreet deal announced | AutoCAD verticalizes; expansion toward high-end M&E |
| 1999 | Discreet acquisition closes; Inventor launches; Buzzsaw created | M&E, mechanical 3D, and internet collaboration expand |
| 2000 | Revit 1.0 launches independently | Parametric BIM enters the market |
| 2002 | Autodesk acquires Revit for $133M | Autodesk commits to BIM |
| 2003 | Civil 3D technology reaches customers | Model-based civil engineering begins replacing disconnected drafting |
| 2006 | Carl Bass becomes CEO; Alias acquisition closes for $197M | Maya, Alias Studio, and broader 3D expertise join Autodesk |
| 2007 | NavisWorks acquired for $26M | Model aggregation and BIM coordination |
| 2008 | Moldflow acquired for about $297M; Softimage acquired | Simulation and M&E expand |
| 2009 | ALGOR acquired for $34M | Engineering simulation deepens |
| 2011 | Autodesk 360 launches; Horizontal Systems acquired | Cloud becomes more than storage |
| 2012 | BIM 360 emerges | Cloud BIM collaboration becomes a product strategy |
| 2013 | Fusion 360 launches (June 25) | Cloud-connected CAD/CAM/CAE platform |
| 2015 | Forge announced; subscription transition announced | Cloud APIs and a new business model |
| 2016 | New perpetual-license sales largely end | Subscription becomes the commercial model |
| 2017 | Andrew Anagnost becomes CEO | Platform, cloud, and subscription strategy accelerates |
| 2018 | PlanGrid acquired for $875M; Assemble acquired; generative design in Fusion | Deeper into construction and computational design |
| 2019 | BuildingConnected closes for $275M; Autodesk Construction Cloud introduced | Connected construction becomes strategic |
| 2020 | Spacemaker acquired for about $240M; Pype acquired | AI-assisted design and construction workflows expand |
| 2021 | Innovyze acquired for $1B; Upchain acquired; Tandem launches | Water, PLM, and digital twins extend lifecycle coverage |
| 2022 | Forma, Fusion, and Flow industry clouds announced; Forge becomes APS | Portfolio reorganizes around industry platforms |
| 2023 | Forma launches; Autodesk AI announced | Cloud platforms and AI converge |
| 2024 | Wonder Dynamics and Datum360 acquired | Generative AI and granular AEC data expand |
| 2025 | Forma repositioned as end-to-end AI-native AECO cloud; Revit becomes first Forma Connected Client | Desktop products begin connecting into an industry-cloud architecture |
| 2026 | ACC becomes Forma; Autodesk MCP servers; Autodesk Operations Solutions | Cloud, contextual AI, and lifecycle operations converge |
| Aug 3, 2026 | MaintainX acquisition completed, about $3.6B | Autodesk extends into asset operations |
Now the part that’s mine.
Era 1: PC CAD and the Extensible Platform (1982-1991)
In 1982 a capable CAD seat meant a dedicated workstation that cost more than a house, so AutoCAD on a personal computer was the founding disruption. Everything else rests on it. The IPO followed in 1985 and gave the company the money and credibility to go international.
Then came 1986, and this is the year people skip. AutoLISP shipped. AutoCAD stopped being an application and became something you could program, and the consultants, third-party developers, and in-house customizers who piled in built an ecosystem no competitor ever caught up with. I’d argue AutoLISP mattered as much as AutoCAD itself, and I know that’s a strange thing to say about a Lisp dialect bolted onto a drafting program.
The same year John Walker handed the CEO job to Al Green and AutoSketch shipped as a low-cost drafting product. Generic CADD was bought in 1989. And in 1990 3D Studio launched, which started a second lineage that eventually became 3ds Max and the whole media and entertainment business.
The breakthrough wasn’t CAD on a PC. It was CAD on a PC you could program.
Era 2: Focus, Intelligent Objects, and Vertical Products (1992-2001)
Carol Bartz arrived in 1992 and brought discipline. Autodesk had been experimenting with software that had little to do with design, and she pointed the company squarely at design markets.
AutoCAD LT came in 1993 and opened a much bigger drafting market. Ithaca Software was acquired the same year, which is how Carl Bass ended up at Autodesk. Release 13 and ObjectARX landed in 1994. ObjectARX let developers build custom objects with real behavior and embedded data. R13 itself was a real rough spot for users, and people who lived through it still bring it up. 3D Studio MAX shipped in 1996. Release 14 arrived in 1997 and stabilized the platform, and Softdesk came in the same year with a deep AEC portfolio.
In 1998 Architectural Desktop and Land Development Desktop shipped, Genius mechanical technology was acquired, and Autodesk agreed to buy Discreet Logic. That deal closed in March 1999, the same year Inventor launched and Buzzsaw.com was created as an early bet on internet collaboration.
Then, in 2000, a small company called Revit Technology Corporation shipped Revit 1.0. A parametric building model, built from scratch, at a time when Autodesk’s architecture products were all still AutoCAD underneath. Keep that in mind for the next section.
The thread through all of it: design information should know what it represents. A wall should be more than lines. A road alignment should be more than geometry. It took another decade to fully land, but this is where the idea started.
Era 3: BIM and Digital Prototyping (2002-2010)
On April 1, 2002 Autodesk closed its $133 million purchase of Revit. I’d call it one of the most consequential acquisitions in company history, and it’s worth being precise about why. Autodesk didn’t invent BIM internally, and it didn’t buy another AutoCAD-based architecture app. It bought a two-year-old product built on a different idea of how building information should work, and then rebuilt its AEC business around it.
Civil 3D reached customers in 2003 and brought the same model-based thinking to infrastructure. Surfaces, alignments, profiles, corridors, all connected, all reactive. That product has been a large part of my working life ever since.
The acquisitions kept coming. CAiCE in 2002, MechSoft in 2004, Alias for about $197 million in 2006 (Maya, FBX, StudioTools), Constructware in 2006, NavisWorks for $26 million in 2007, Moldflow for about $297 million in 2008, then Ecotect, Green Building Studio, Softimage, and ALGOR for $34 million in 2009.
Carl Bass became CEO in 2006 and pushed hard on 3D, BIM, simulation, and eventually cloud.
The shift in this era was from documenting geometry to modeling systems and relationships. Revit for buildings, Civil 3D for infrastructure, Inventor for products, Navisworks to coordinate the mess.
Era 4: Cloud Foundations and Subscription (2011-2016)
Two pivots here, one technical and one commercial.
Technical: Autodesk 360 services showed up in 2011, BIM 360 in 2012, and Fusion 360 launched commercially on June 25, 2013. Fusion was the biggest internally built product bet since Inventor, putting CAD, CAM, CAE, and data on a cloud-connected architecture. Forge followed in 2015 and opened the cloud as APIs. Acquisitions filled gaps along the way: Horizontal Systems, T-Splines, HSMWorks, Delcam, Shotgun, Solid Angle.
Commercial: on February 4, 2015 Autodesk announced that most new standalone seats would go subscription-only starting February 1, 2016, with suites following in the next phase. It was the biggest business-model change since the founding, and plenty of customers pushed back. Some still do. It also rewired the company’s economics and its relationship with how people actually use the software. One correction I’d make to the common version of this story: subscription did not turn Autodesk into a SaaS company. AutoCAD, Revit, Civil 3D, and Inventor stayed on your desktop. What changed was how you paid.
By the end of this era the cloud had stopped being a place to store files and become an architectural layer.
Era 5: Generative Design and Connected Construction (2017-2020)
Andrew Anagnost took over in 2017 with the subscription transition to finish and the cloud strategy to push.
Then Autodesk crossed a line it had respected for 36 years. For its whole history the company helped you design things before they were built. In 2018 it paid $875 million for PlanGrid and moved onto the construction site. BuildingConnected followed in 2019 for $275 million, extending upstream into bidding and preconstruction. Assemble in 2018 and Pype in 2020 rounded out the set, and Autodesk Construction Cloud pulled it all under one name in 2019.
Generative Design went into Fusion 360 in 2018, and the Spacemaker acquisition in 2020 (about $240 million) brought AI-assisted site planning that later became the core of Forma.
After this the AEC business was no longer confined to the design office.
Era 6: Industry Clouds, Lifecycle Data, and AI (2021-2024)
Innovyze cost a billion dollars in 2021 and brought water infrastructure simulation and operational analytics. Upchain added cloud PLM. Tandem launched as a digital twin platform.
The big reframe came in 2022: industry clouds. Forma for AECO, Fusion for design and manufacturing, Flow for media and entertainment, with Forge renamed Autodesk Platform Services as the shared layer underneath. Autodesk renames things a lot. This one mattered more than most because it described a strategy, not just a name.
Forma launched in 2023. Autodesk AI was announced the same year. Wonder Dynamics and Datum360 joined in 2024.
The desktop application stopped being the only center of gravity. Connected lifecycle data became the layer tying the products together. Revit, AutoCAD, Civil 3D, and Fusion still get enormous investment and still do the actual work. They’re just no longer the whole story.
Era 7: Connected Clients, Agentic AI, and the Lifecycle Loop (2025-2026)
This is where it gets interesting for me, because it’s where I’ve been spending my own time.
AU 2025 is where the pieces lined up. Forma was positioned as the end-to-end, AI-native AECO industry cloud. Autodesk said ACC would move into Forma. Revit was named the first Forma Connected Client. Granular cloud data and desktop-to-cloud interoperability became stated architecture rather than roadmap talk. If you only look at 2024 and then 2026 the convergence looks sudden. It wasn’t. 2025 is the hinge.
In March 2026 Autodesk Construction Cloud moved under the Forma name, putting design and construction in one AECO strategy. On April 9 the Autodesk Product Help MCP Server launched, a read-only service that exposes documentation across 110-plus products to AI agents through Model Context Protocol. By mid-2026 the Product Help and Fusion MCP servers had reached general availability, and Revit, AutoCAD, and Civil 3D had started exposing live application and model context through public MCP Tech Previews.
On May 28 Autodesk announced two things together: an agreement to acquire MaintainX, and a new Autodesk Operations Solutions group built from Tandem, FlexSim, Fusion Operations, and Factory Design Utilities. The MaintainX deal closed on August 3 at about $3.6 billion. It’s one of the largest acquisitions in company history and it isn’t a design tool. It’s maintenance and asset operations software.
So the perimeter is no longer Design and Make. It’s Design, Make, and Operate.
I think MCP may turn out to be the sleeper here. It moves AI from generating answers to interacting with actual design environments and structured project data. An agent that can read your model, your application state, and trusted documentation is a different thing from a chatbot, and Autodesk is now openly describing its platform work as infrastructure for agents. The direction is right. The interesting part will be watching how far it goes.
The 12 Transitions That Shaped Autodesk
- 1982: CAD moves to the PC
- 1986: AutoLISP makes AutoCAD a programmable platform
- 1992: Bartz brings focus and scale
- 1994-1999: Geometry becomes intelligent objects
- 2002: Revit commits Autodesk to BIM
- 2006-2009: Autodesk becomes a broad 3D, simulation, and media company
- 2011-2015: The cloud becomes an architectural layer
- 2015-2016: Subscription rewires the business model
- 2018-2019: Autodesk moves onto the construction site
- 2022-2026: Products reorganize around industry clouds and connected data
- 2023-2026: AI moves from embedded features toward assistants, context, MCP, and agents
- 2026: Operations closes the lifecycle loop
The Larger Pattern
Across 44 years Autodesk kept widening what counts as part of the design problem. First it was making a drawing. Then the drawing became programmable. Geometry became objects, objects became models, models became cloud data, cloud data became platforms, and now AI is starting to work across those platforms with real project context.
With the move into operations, design information no longer has to die when the building is handed over or the product leaves the factory. The loop that’s forming is design, make, operate, learn, design again.
Autodesk started in 1982 by changing where design happened. Forty-four years later it’s trying to change how long design information stays useful. The next Autodesk platform may not be an application or a file format at all. It may be the persistent, computable context around a physical asset for its entire life.
John Walker would have had opinions about all of this. He usually did.
Cheers,
Shaan
