Topics
tardy
AI
Amazon
Image Credits:Runware
Apps
Biotech & Health
mood
Image Credits:Runware
Cloud Computing
DoC
Crypto
enterprisingness
EVs
Fintech
Fundraising
Gadgets
Gaming
Government & Policy
Hardware
layoff
Media & Entertainment
Meta
Microsoft
privateness
Robotics
security department
societal
Space
Startups
TikTok
Transportation
Venture
More from TechCrunch
Events
Startup Battlefield
StrictlyVC
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
get through Us
Sometimes , a demonstration is all you necessitate to understand a product . And that ’s the pillow slip with Runware . If you guide over toRunware ’s website , go into a prompt and hit enter to generate an image , you ’ll be storm by how speedily Runware sire the persona for you — it takes less than a moment .
Runware is a newbie in the AI illation , or reproductive AI , startup landscape . The company is building its own servers and optimize the package layer on those server to remove bottlenecks and amend inference speeds for effigy contemporaries models . The inauguration has already insure $ 3 million in funding from Andreessen Horowitz ’s Speedrun , LakeStar ’s Halo II and Lunar Ventures .
The company does n’t want to reinvent the wheel . It just want to make it spin faster . Behind the scenes , Runware fabricate its own server with as many GPUs as potential on the same motherboard . It has its own usage - made cool organization and manages its own data centers .
When it come to running AI model on its server , Runware has optimized the orchestration bed with BIOS and operating organisation optimizations to ameliorate cold start times . It has grow its own algorithms that allocate hindrance workloads .
The demonstration is impressive by itself . Now , the company require to practice all this body of work in research and development and turn it into a business organisation .
Unlike many GPU hosting companies , Runware is n’t run to engage its GPUs based on GPU fourth dimension . alternatively , it believe companies should be encouraged to speed up work load . That ’s why Runware is offering an image generation API with a traditional price - per - API - call fee structure . It ’s base on democratic AI models from Flux and Stable Diffusion .
“ If you reckon at Together AI , Replicate , Hugging Face — all of them — they are selling compute based on GPU time , ” co - father and CEO Flaviu Radulescu told TechCrunch . “ If you equate the amount of metre it takes for us to make an image versus them . And then you equate the pricing , you will see that we are so much cheap , so much quicker . ”
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
“ It ’s going to be impossible for them to match this execution , ” he add . “ specially in a cloud provider , you have to lead on a virtualized environment , which adds additional delays . ”
As Runware is looking at the entire inference pipeline , and optimizing hardware and software , the company hopes that it will be able to use GPUs from multiple vender in the near time to come . This has been an important try for several startups as Nvidia is the well-defined leader in the GPU quad , which means that Nvidia GPUs tend to be quite expensive .
“ powerful now , we expend just Nvidia GPUs . But this should be an abstract of the software level , ” Radulescu said . “ We can flip a model from GPU memory in and out very , very fast , which allow us to put multiple customers on the same GPUs .
“ So we are not like our contender . They just load a framework into the GPU and then the GPU does a very specific case of chore . In our case , we ’ve develop this software result , which admit us to alternate a model in the GPU computer storage as we do illation . “
If AMD and other GPU vender can create compatibility level that work with distinctive AI workloads , Runware is well positioned to build a intercrossed cloud that would swear on GPUs from multiple vendors . And that will certainly help if it wants to rest cheaper than competitors at AI inference .
