Runway CEO Cristóbal Valenzuela posted two phrases on X this afternoon in response to at present’s OpenAI shock demo drop of its new Sora video AI mannequin, which is able to producing 60-second clips: “sport on.”
Clearly, the race for video AI gold is on. In spite of everything, only a few months in the past Runway blew people’s minds with its Gen-2 replace. Three weeks in the past, Google launched its Lumiere video AI technology mannequin, whereas simply final week, Stability AI launched SVD 1.1, a diffusion mannequin for extra constant AI movies.
Stability AI CEO Emad Mostaque responded positively to the Sora information, calling OpenAI CEO Sam Altman a “wizard.”
Right now’s Sora information firmly squashed any speak about OpenAI “jumping the shark” within the wake of Andrej Karpathy’s departure; studies that it was all of a sudden entering into the web search product enterprise; and Sam Altman’s $7 trillion AI chips challenge.
Runway has a necessity for velocity
In June 2023, Runway was valued at $1.5 billion, after elevating $141 million from traders together with Google and Nvidia.
In January, Runway introduced it added a number of movement controls to AI movies with its Multi Movement Brush, whereas it’s as but unclear what function set Sora gives, or its limitations.
Coincidentally, I spoke to Runway’s inventive director Jamie Umpherson yesterday about what I feel is without doubt one of the firm’s actually fascinating and intelligent non-video promotional artwork initiatives — its Gen-2 Book of Weights, an analog printed booklet of nothing however numbers that’s the first quantity (of 6834) what would be the total assortment of Gen-2 mannequin weights.
“It was a enjoyable experiment,” he advised me, including it got here from a “dialog in passing as lots of these extra inventive experimental concepts do — we have been simply discussing how humorous it will be to make the weights of the mannequin tangible, visible, one thing that you can truly eat.”
For now, nevertheless, Runway will most definitely need to be laser-focused not on artwork or philosophy, however maintaining with OpenAI and the remainder of the pack as a way to hold tempo within the video AI race.