Nashville, TN — Autodesk’s annual user conference, Autodesk University (AU), is always a barometer for where the design software giant thinks the industry is headed. This year, CEO Andrew Anagnost left little doubt: Autodesk is wagering its future on AI-native, cloud-based industry platforms.
In a sweeping 1 hr 20-minute opening keynote, Andrew and his product leaders introduced expansions of Autodesk’s three industry “clouds”—Forma for architecture, engineering, and construction (AEC); Fusion for design and manufacturing (D&M); and Flow for media and entertainment (M&E). Together, the platforms aim to move customers beyond files and isolated desktop workflows into continuous, data-rich, AI-assisted collaboration environments.
“The choices we make over the next five years will shape our work, and the world that results,” Andrew said. “We want you to make more with less—more housing, more resilient infrastructure, more factories, more stories—using fewer resources, less time, and delivering better outcomes.”
Addressing Scarcity with Technology
Andrew framed the announcements against a backdrop of global scarcity. Labor shortages, strained supply chains, and climate-related constraints, he said, mean industries must do more with fewer inputs. AI and cloud, in Autodesk’s vision, unlock “unlimited capacity” by amplifying human effort and making project delivery faster and more efficient.
He reminded attendees that Autodesk forecast this shift seven years ago. At his first AU as CEO, he promised that “the future would allow every discipline to collaborate seamlessly in the cloud, powered by AI.” That vision, he argued, has arrived in the form of Forma, Fusion, and Flow.
Forma: From Concept to Construction
The keynote’s first live segment showcased Autodesk Forma, the AEC cloud Autodesk launched in 2023 after acquiring Spacemaker. Forma, Andrew said, is becoming the first comprehensive AEC industry cloud, spanning concept, design, and construction.
Product expert Robin Franke, an architect by training, demonstrated Forma’s expanding building design tools. Starting from a massing model of the Snowden Towers—a demo dataset familiar to Revit users as it appears every time Revit is initiated—Robin converted simple blocks into detailed buildings with facades, windows, and balconies.
Crucially, she wasn’t manually modeling: “The system is modeling for her,” interjects Andrew. Robin explored options while Forma handled geometry generation in the background.
She ran a daylight potential analysis in real time. Lighter blues indicated adequate daylight, while darker shades highlighted problem areas. By pushing a building mass back, Robin reran the simulation and improved the results instantly.
The workflow underscored Autodesk’s pitch: iterative design with immediate performance feedback, reducing the labor of traditional BIM modeling.
Generative Layouts Inside Buildings
Robin then showed Forma’s new generative AI interior layout tool. Setting parameters—multifamily residential, wood as primary structural material—she clicked “Generate.” The AI produced 36 housing units with a balanced mix. With a small mass expansion and material switch, she optimized to 56 units.
All results were editable: walls and columns could be adjusted, with daylight analysis updating accordingly. This represented Autodesk’s argument that AI augments but doesn’t replace architects, keeping “creative agency” intact.
Connecting Forma and Revit
The next phase brought in Matt Arsenault, Autodesk’s “Revit champion.” Because Robin’s Forma model was connected, all contextual data—including earlier analyses—appeared natively in Revit.
Matt reran a wind comfort analysis directly in Revit, powered by Forma’s cloud back-end. Changes in one environment were immediately visible in the other.
This bidirectional link, Andrew emphasized, “pulls the digital thread across all phases of design.”
Autodesk Assistant in Revit
Matt then introduced the Autodesk Assistant, an AI agent embedded in Revit. Asked about window-to-wall ratios, it reported 19% against the project’s targets. When prompted, it automatically widened windows across the façade, maintaining performance criteria.
He pushed further: creating a floor plan, cleaning it with a view template, auto-tagging walls and rooms, and placing it on a new sheet—tasks that would normally take hours—were executed in seconds by Assistant.
“Hours of work compressed into moments,” Matt said.
Construction in the Cloud
Nick Zeeben, another product lead, showed Autodesk’s new connected preconstruction environment, uniting quantity takeoff and cost estimation.
Using live model data, he updated a concrete cost from $130 per cubic yard (from a local supplier) and applied a waste factor.
“That’s a lot of decimals,” remarked Andrew, noting the precision of the data.
Nick also demoed custom dashboards for field supervisors. Easily created dashboards displayed cost, quality, safety, and schedule metrics, deployable per project or company-wide.
On-site crews will soon capture photos and logs via mobile, with AI automatically filling forms from images. Project managers can query Assistant for risks—such as steel installation delays—and have it draft update requests to trade partners.
Tandem and Digital Twins
The construction section culminated with Adam Peter, subject matter expert, the “best title ever,” demonstrating Autodesk Tandem linked with IBM Maximo.
A heat recovery unit flagged for failure generated a work order. Tandem Insights aggregated manuals, IoT data, warranties, and performance metrics, presented spatially in the building model.
“This moves us from reactive to proactive—truly predictive digital twins,” Adam said.
Customer Case: Denver International Airport
To prove Forma’s real-world traction, Autodesk highlighted Denver International Airport’s “Vision 100” plan to handle 100 million passengers.
Denver’s digital lead Brendan Dillon, Director of Digital Facilities at DEN, explained that 300 active projects, once siloed, now coordinate via cloud-based Revit, Civil 3D, and Tandem.
“Previously, you crossed your fingers and hoped everything worked out,” Brendan said. “Now we can coordinate all projects and identify problems before they happen.”
His advice: every organization needs an innovation champion, must institutionalize knowledge, and should share openly with peers.
Fusion: Design and Manufacturing in the Cloud
Shifting from construction to products, Autodesk spotlighted Fusion, its cloud for design and manufacturing. This year, Fusion formally integrates PLM (product lifecycle management), aiming to unify data from kickoff through manufacturing.
Air Fryer Demo
To illustrate, Autodesk’s team collaboratively designed a consumer air fryer, perhaps the most popular kitchen appliance in recent years.
“Before this presentation, I had no idea what an air fryer was,” says Andrew.
Heather Kerrick, Director of AI Software Development at Autodesk, demonstrated AI-powered onboarding: adding Andrew to a Fusion project required a simple prompt to Assistant, which issued an invitation and role assignment without leaving context.
Next, she used generative AI to produce the air fryer’s initial CAD geometry.
“This blank sheet problem is solved,” Heather said.
The system iteratively sculpted surfaces, handles, and feet, outputting fully editable CAD—not mesh approximations. Downstream tasks like splitting, shelling, or manufacturing were immediately possible.
Mechanical and Product Design
Daniel Graham, Senior Director of Product Management at Autodesk, highlighted two features:
- AI Sketch Fixes: Automatically closing gaps and maintaining constraints.
- New Constraint Command: Simplifying 3D assembly positioning, supplementing Fusion’s joint tool.
Both, he noted, were direct responses to user requests.
Microsoft Integration
Daniel also demonstrated Fusion integration with Microsoft PowerPoint via Autodesk Assistant. AI-generated design variants (a “too fancy” rose gold, matte gray, black ABS), placed them in the context of a kitchen, rendered images and auto-populated a presentation deck.
“What used to be tedious rendering and formatting is now automated,” he said.
Inventor and Fusion Connectivity
Jeremy Stadtmueller, Director of Product Management for Fusion and Inventor, showed mold design linked to Fusion. When Daniel enlarged a hole in Fusion, the change updated automatically in Inventor’s mold cavity. Inventor’s tooling is then uploaded directly back to Fusion’s manufacturing environment, with linked data (not files) ensuring accuracy.
This workflow underscored Autodesk’s push to integrate desktop stalwarts (Inventor, Vault) into cloud-native Fusion.
Customer Case: SwissDrones
Andrew welcomed SwissDrones, a Swiss company producing unmanned helicopters for infrastructure inspection, to the main stage.
Head of Airworthiness Alexander Lemus does his best to ensure the aircraft SwissDrones flies has an “equal amount of landings and take-offs.” He was joined by design lead Pol-Victor Gisquet, who described how Fusion manages their entire lifecycle—from design to CNC composite cutting to configuration control.
Their showcased drone: 7.6 feet long, 3 feet wide, with a maximum takeoff weight of 190 pounds.
“It’s like a micro helicopter,” Paul said. “A big toy—but one made entirely with Fusion.”
Fusion’s Manage Extension ensures compliance with aviation standards, while its CAM optimizes carbon-fiber part cutting.
Media and Entertainment Flows
Autodesk’s third cloud, Flow, integrates AI for media and entertainment. Flow is a rebranded Wonder Studio, from Wonder Dynamics, which was acquired by Autodesk in 2024.
Mexico-based Boxel Studios used Flow to accelerate production of Superman & Lois.
The studio reduced rotoscoping and mocap (motion capture) overhead by using Flow Studio’s AI to map performances directly from live video.
Nikola Todorovic, co-founder of Wonder Dynamics, introduced as the mastermind behind Flow Studio, demonstrated the program’s AI-based motion capture (mocap), which utilizes 25 models to capture motion, lighting, and compositing data from video.
Real human actors on one window are tagged to digital characters in another. The AI is able to generate the character, replacing the human actor, for the whole scene, outputting not just footage but fully editable 3D data (camera tracks, alpha masks, emission maps).
That used to be weeks of work, Nikola said, that could be done in hours. Unlike “black box” AI tools, Flow outputs data that artists can refine in Maya.
A Character Creator tool generates 3D characters from text or image prompts, ready for insertion into live-action or animated scenes.
Motion Maker in Maya
Lance Thornton, a former Pixar and DreamWorks animator, now at Autodesk, unveiled Motion Maker, which is now integrated into Maya.
By setting a few keyframes, he generated a realistic running motion for a robot character named Andrew.
“Just a coincidence,” he says to real-life Andrew.
“Looks like he ran through the gate,” says real-life Andrew, giving Lance a chance to direct character Andrew to turn the corner and avoid the gate straight ahead.
The quickly generated scene shows Andrew not only disappearing as he rounded the corner but also noticeably banking as he did so.
An animator typically produces four seconds of footage per week due to the time-consuming nature of frame-by-frame character animation.
Autodesk also previewed Face Animator in Maya. Given an audio clip, the AI generated lip-sync and facial motion for a character in seconds.
Closing Vision
Andrew sees a future where projects become neural networks of data-rich nodes, continuously learning and adjusting across applications and industries.
“Over four decades, we have transformed how things are designed and made,” he said. “Our goal is to give you the platform to create a better world in this next era—working at a scale and pace you’ve never seen before—by unlocking unlimited capacity.”