Topics
belated
AI
Amazon
Image Credits:Kimberly White/Getty Images for WIRED
Apps
Biotech & Health
mood
Image Credits:Kimberly White/Getty Images for WIRED
Cloud Computing
DoC
Crypto
enterprisingness
EVs
Fintech
Fundraising
gizmo
Gaming
Government & Policy
computer hardware
layoff
Media & Entertainment
Meta
Microsoft
Privacy
Robotics
Security
societal
quad
Startups
TikTok
Department of Transportation
Venture
More from TechCrunch
Events
Startup Battlefield
StrictlyVC
newssheet
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
Cloud providers are chasing after AI unicorn , and the latest isFei - Fei Li ’s World Labs . The startup just tapped Google Cloud as its master compute supplier to train AI model , a deal that could be worth hundreds of one thousand thousand of dollars . But Li ’s tenure as chief scientist of AI at Google Cloud was n’t a factor , the company says .
During the Google Cloud Startup Summit on Tuesday , the companiesannouncedWorld Labs will use a large chunk of its funding to license GPU servers on the Google Cloud Platform , and ultimately train “ spatially reasoning ” AI models .
A handful of well - funded startups work up AI base models are highly sought after in the cloud services world . Some of the largest peck include OpenAI , which solely develop and head for the hills AI models on Microsoft Azure , and Anthropic , which uses AWS and Google Cloud . These companies regularly pay millions of dollars for computing services , and could one day postulate even more as their AI models scale . That get to them worthful customers for Google , Microsoft , and AWS to progress relationships with early on .
World Labs is certainly building unequalled , multimodal AI framework with significant compute need . The startup justraised $ 230 millionat more than a billion - buck valuation , a deal go by A16Z , so as to build AI world fashion model . General manager of startups and AI at Google Cloud , James Lee , tells TechCrunch that World Labs ’ AI models will one day be able to sue , generate , and interact with TV and geospatial data . World Labs calls these AI models “ spacial intelligence . ”
Li has deep draw with Google Cloud , having lead the party ’s AI elbow grease in 2018 . However , Google traverse that this muckle is a mathematical product of that relationship , and rejects the approximation that cloud services are just commodities . Instead , Lee say services , such as its High Performance Toolkit to scale AI workload , and its deep provision of AI check were a magnanimous factor .
“ Fei - Fei is obviously a friend of GCP , ” said Lee in an audience . “ GCP was n’t the only pick they calculate at . But for all the reasons we talked about – our AI optimized infrastructure and the ability to get together their scalability demand – ultimately they derive to us . ”
Google Cloud offer AI startups a pick between its proprietary AI chips , tensor processing units or TPUs , and Nvidia ’s GPUs , which Google purchases and has a more circumscribed supply of . Google Cloud is trying to get more startup to train AI models on TPUs , largely as a means to deoxidize its dependency on Nvidia . All cloud supplier are limited today by the scarceness of Nvidia GPUs , so many are build their own AI chips to run across need . Google Cloud say some startups are training and inferencing solely on TPUs , however , GPUs still remain the industry ’s favored AI training potato chip .
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
World Labs chose to take aim its AI model on GPUs in this deal . However , Google Cloud would n’t say what go into that decisiveness .
“ We worked with Fei - Fei and her mathematical product team , and at this stage of their product roadmap , it made more sense for them to work out with us on the GPU program , ” said Lee in an audience . “ But it does n’t needfully mean it ’s a permanent decision … Sometimes [ startups ] move onto different platforms , such as TPUs . ”
Lee would not disclose how large World Labs ’ GPU cluster is , but cloud provider often give monumental supercomputers for startups training AI models . Google Cloud promised another startup training AI understructure models , Magic , a cluster with “ tens of yard of Blackwell GPUs , ” each of which has more big businessman than a gamey - remnant gaming PC .
These cluster are easier to promise than they are to fulfill . Google ’s cloud service competitor Microsoft is reportedlystruggling to take on the insane compute demandsof OpenAI , squeeze the startup to tap other options for computing tycoon .
World Labs ’ deal with Google Cloud is not exclusive , mean the inauguration may still strike hatful with other cloud providers . However , Google Cloud hosts the absolute majority of World Labs ’ workloads today , and says it will essay to maintain that stage business moving forward .
