Microsoft has recently rolled out a notable upgrade to its Bing Image Creator, integrating OpenAI's DALL-E 3 model into the AI-based art maker. However, the integration of the DALL-E 3 model has reportedly led to a significant increase in user interest, putting a strain on the system's performance.
Increased Access Leads to Service Slowdown
X (formerly Twitter) user, @dmitryforleader reports having waited over two hours for a single image to be generated.
I've come across the fact that Bing Image Creator is very, very slow to create images. I wait more than two hours to get the images. Why is this happening? What is the reason? #bingimagecreator #DALLE3 #AIart #AiArtwork #AI #image
— Dmitry Saxe-Coburg and Gotha (@DmitryForLeader) September 30, 2023
Similar experiences have been recounted by other users, complaining about similar lengthy generation times for the AI-created artwork.
This is what my screen looked like while I waited. @Bing wasn't ready for the popularity of DALL·E 3 today. Impressive results, but that wait was daunting (and yes, I went off and did some other chores, checking back every 5 or 10 minutes or so). #DALLE3 pic.twitter.com/STa0AHhG2l
— The Tårn (@thetarnck2) October 1, 2023
Microsoft Scrambles to Boost Servers' Capabilities
Aware of the issue and its unfavorable implications, Microsoft addressed the concern promptly. The company's new Head of Windows and Web Services team, Mikhail Parakhin, acknowledged the overwhelming user influx in a post on X. Parakhin admitted that even though they expected a hike in user interest following the DALL-E 3 integration, they did not anticipate the sudden spike in demand.
Folks, we know DALL-E 3.0 generation right now is taking longer than normal. We expected some strong interest, but we didn't expect THAT much, especially given it's a weekend. Bringing more GPUs in, will be better soon.
— Mikhail Parakhin (@MParakhin) October 1, 2023
In response to this unforeseen challenge, Microsoft plans to integrate additional GPUs into the system. Parakhin says that a significant acquisition, measured in thousands of Nvidia H100 GPU-equivalents, is anticipated.