NVIDIA has entered the AI Software stack

Welcome to The Alignment. More big players announcing big moves in the AI space, each one believes they are onto something. Let's find out if their plans hold water

As always if you enjoy reading our posts be sure to spread the word !

Here’s what we have lined up for you today -

1. NVIDIA has entered the AI Software stack

2. Adobe is betting big on its custom models

3. Bing and Bard launch new tools

NVIDIA has entered the AI Software stack 

NVIDIA has announced the launch of NVIDIA AI foundations, its dedicated cloud computing capabilities for organisations building Generative AI. The goal is to allow users to train, design and deploy their foundation models from scratch, “Foundation Models as a Service” for proprietary data.

Built primarily for the enterprise, users will be able to bring in all sorts of data - text, images, videos, 3D data and start building multimodal LLMs and Generative models for their custom use cases.

Getty images, Shutterstock and Adobe were among the launch partners announced so NVIDIA has to an extent de-risked fears of copyright. Support for Biology LLMs will also be provided with the company claiming that this is the first large scale AlphaFold API/fine-tuning service powering drug discovery.

NVIDIA’s ambitions are clear with this move. They’ve been hinting at a full software suite offering for a while and now they are delivering on it by becoming an enterprise focused AI capabilities provider.

Adobe is betting big on its custom models  

Adobe has launched its family of AI models called Firefly.

The models have been purpose built for media content generation. The models will be tightly integrated across Adobe apps like Photoshop, Illustrator, Express and the Adobe experience manager. Model capabilities will be improved over time. Right now a basic text to Image model plus a model that can transfer different “styles” to images is available.

Adobe claims that their model will avoid the legal pitfalls that players like StabilityAI and Midjourney are currently battling. It did not provide how it will do so apart from the fact that the models are trained on Adobe Stock, the company’s royalty free library along with other public domain media content. Adobe will allow users to train the models on their own data as well for customisation.

On the ownership of the assets created by Firefly a company exec said “that using its tools to add “creative input” to a generated image should be sufficient to allow a creator to obtain copyright. “

Firefly has distribution built in with over 600 million monthly active users on Adobe and the Experience Cloud having 12000 customers including 87% of the Fortune 100 companies.

At a time when every other company is launching a partnership with OpenAI, StabilityAI, Anthropic or other players, Adobe building models in house is a unique approach. One that it definitely has the resources to take. Let’s see how much value it can deliver !

Around the industry -

1. Luma raises 20 million Series A to build Generative 3D videos (we previously featured them)

2. Inside Notion’s strategic bet on generative AI

3. Microsoft will instantly transcribe patient notes during doctor visits

4. Bing Create - Create images from words with AI

5. Bard is opening up to many users (only in the US + UK for now)

Product Corner -

There are way too many cool products to showcase just one !

• Tripsnotes : AI Generated Travel Plans

• ScripAI : 10x faster scripts for TikTok, Reels , YT shorts

• Milo : Automate the repetitive boring parenting tasks

• Defog : ChatGPT for your data

• Youtube University : Automated notes of any video

Bringing the Zing while cloning Bing

All in 2 hours of work !