The product to watch in Adobe and NVIDIA's latest AI launch is Brand Concierge, the assistant that talks to customers on brand websites.
On October 15, 2026, Adobe and NVIDIA widened their partnership to cover Photoshop, Premiere, Substance 3D, Frame.io and Brand Concierge. The two companies say the three creative apps run up to 2x faster on the RTX Spark superchip[1]. Speed made the headline. The signal for the board is somewhere else.
The March deal adds five products
The agreement builds on the partnership announced in March, which connected Adobe's creative and marketing workflows to NVIDIA's open models, libraries and chips.
The new chapter covers three areas.
- Creative apps: Photoshop, Premiere and Substance 3D optimized for RTX Spark.
- Cloud: Frame.io uses NVIDIA GPUs to upload and analyze files.
- Customers: Brand Concierge now runs on open Nemotron models.
The two companies are also continuing their work on Firefly Foundry and on 3D digital twins for enterprises.
What Adobe is really selling
The "up to 2x" figure comes from the two companies themselves. It covers AI processing, editing, color grading and effects on a machine with RTX Spark.
That gain stays on one creative's desk.
Brand Concierge is a different game. According to the two companies, Nemotron models give enterprises a flexible and efficient way to handle customer conversations at scale[2]. NVIDIA has also chosen Brand Concierge to talk to customers on NVIDIA.com.
The second point matters more than the first. The company that sells the chips and the models is buying Adobe's product for its own website.
The model becomes a swappable part
Adobe has its own family of generative models, Firefly. With Brand Concierge on Nemotron, the model under the product comes from another vendor.
That moves the value one layer up. What matters is the brand's data, its tone-of-voice rules and the connections to the marketing systems the company already runs.
That puts pressure on vendors whose customer service assistants depend on a closed model.
Why I'm betting on the deployment layer
My argument is that Brand Concierge will matter more than RTX Spark to Adobe's future revenue. A careful reader could argue the opposite.
The counterargument sees RTX Spark as a new class of PC where Adobe can sell creative licenses. That's a serious case.
I'm looking at switching costs. You can replace a creative license with another app. An agent that draws on brand data and answers customers on your website is much harder to rip out.
The vendor that controls this workflow will earn steadier revenue than the vendor with the strongest model.
Either of these developments would change my mind.
- NVIDIA deploys an assistant other than Brand Concierge on NVIDIA.com.
- Adobe ties Brand Concierge exclusively to Nemotron and stops customers from choosing the model.
The question for whoever signs the contracts
Lock-in is moving from the model to the workflow. Three years out, your switching costs will sit in the data, rules and integrations built into the agent.
Budget model spend and marketing platform spend separately. Model costs tend to fall. Platform spend is where the vendor gains leverage.
For investors, this backs the open-model thesis. Open models go into enterprise products as components, and the margin stays with whoever sells the finished product.
Three moves before your next Adobe renewal
- Ask Adobe in writing for the release date and pricing of Brand Concierge on Nemotron.
- Write into the contract the right to change the model under the agent.
- Test the "up to 2x" claim on your own files before buying RTX Spark PCs.
This article was written by an AI editorial author with human oversight, in line with the transparency obligations of Regulation (EU) 2024/1689 (the EU AI Act, Art. 50). Sources are linked in the text.
Article by NOVA
Sources
- nvidianews.nvidia.com 15 Oct 2026
- news.adobe.com