Adobe & NVIDIA boost creative software & media tools

Adobe and NVIDIA have expanded their partnership to optimise creative software, cloud media processing and personalised customer experiences, with Adobe reporting performance improvements of up to twice the speed in selected applications.

The collaboration covers Adobe Photoshop, Premiere, Substance 3D, Frame.io and Brand Concierge. It combines NVIDIA's accelerated computing technology and open AI models with Adobe's creative and marketing software.

Photoshop, Premiere and Substance 3D are being optimised for NVIDIA's RTX Spark system, with improvements targeting AI processing, editing, colour and effects. Adobe said the changes are designed to help creative professionals complete tasks more quickly.

Creative tools

Adobe has reworked Photoshop around GPU-accelerated compositing to improve responsiveness for tasks such as removing distractions and applying visual effects. These include Gaussian Blur and Light Leaks, which users can apply directly to the canvas.

In Premiere, NVIDIA acceleration targets demanding post-production tasks involving playback, rendering, effects, colour and AI-powered features. Adobe cited Enhance Speech, which processes audio, and Object Mask among the functions benefiting from the optimisation.

For users of Substance 3D Painter, the work focuses on speeding up computationally demanding texturing tasks used in 3D production.

The applications are optimised to use the RTX Spark system's unified memory and AI computing capabilities. Adobe said the resulting improvements could help users move between editing tasks and AI-assisted features with less waiting.

"Creative workflows are more multimodal than ever, with many projects spanning images, video, audio, motion and 3D, making performance critical to helping creatives stay focused on their ideas and craft," said Deepa Subramaniam, Vice President, Product Marketing, Creative Professionals, Adobe. "By partnering with NVIDIA to optimize Photoshop, Premiere and Substance 3D for RTX Spark and accelerating Frame.io in the cloud, we're delivering the performance creatives need to bring their ideas to life."

"AI agents are helping creators unlock deeper and richer workflows using the Adobe applications they know and love," said Jason Paul, Vice President, Local AI, NVIDIA. "Powered by RTX Spark, Adobe's optimized tools deliver up to 2x faster performance to accelerate these new workflows."

Cloud processing

The expanded partnership also targets Frame.io, Adobe's platform for video collaboration and media management. The companies are using NVIDIA accelerated computing to speed up cloud-based media processing and content discovery.

The updated processing pipeline uses NVIDIA NVDEC for hardware-accelerated video decoding, alongside NVIDIA TensorRT and CUDA for content preprocessing and embedding generation. Embeddings help systems represent media content in a form that supports search and retrieval.

Moving these workloads to NVIDIA GPUs is intended to reduce the time required to process uploaded media and make it searchable. The work also targets content discovery across large libraries used by production teams.

Early results from selected Frame.io transcoding and search workloads showed approximately five times faster embedding generation, a four-to-five-fold improvement in 95th-percentile processing latency and an approximately 80% reduction in search proxy size.

The 95th-percentile latency measure reflects the processing time below which 95% of measured requests are completed. Improvements in this measure can help reduce delays for teams handling substantial volumes of media.

The results relate to selected workloads rather than every Frame.io operation. Adobe said the changes are intended to make uploaded content searchable sooner and improve the efficiency of maintaining and searching larger media libraries.

Customer engagement

The third area of the collaboration involves Adobe Brand Concierge, which supports personalised conversational experiences for businesses.

Adobe and NVIDIA have worked to optimise Brand Concierge using NVIDIA Nemotron open models. The version is expected to become available to enterprise customers later in 2026.

NVIDIA has also selected Brand Concierge to power customer interactions on its website. The system will use organisational data and content to tailor responses to customer needs, with Nemotron supporting conversational performance at scale.

The approach gives enterprises another option for deploying AI models in customer-facing services. Open models can provide flexibility in how businesses configure and operate AI-powered interactions, although the results will depend on implementation and the quality of the information used to ground responses.

The expanded work builds on a strategic partnership established in March 2026. That collaboration also includes GPU acceleration for media ingestion and intelligence in Frame.io, alongside work involving Adobe Substance 3D, Firefly Foundry and 3D digital twin workflows for enterprise customers.