Blog
1 October, 2024
Hour One’s Journey with TeraSky and Google Cloud
Hour One has revolutionized AI video creation by migrating its workloads to high-performance Cloud GPUs. This transition has enabled faster video content generation, improved image quality, and the development of sophisticated AI models, leading to significant business growth.
Achieving Remarkable Growth with Google Cloud
Since migrating to Google Cloud, Hour One has experienced:
- 20x business growth
- 1.8x improvement in inference speed with Cloud GPUs
- 28% reduction in inference costs
- More realistic and expressive AI-generated avatars
As video content demand surges—with 91% of consumers expressing a preference for more video content from brands—Hour One is perfectly positioned to meet this need. Founders Lior Hakim and Oren Aharon established Hour One in 2019 to democratize professional-grade video production, enabling businesses to create cinematic videos quickly and affordably, even without technical expertise.
Overcoming GPU Challenges
Originally relying on a third-party cloud provider’s GPUs, Hour One faced capacity limitations as its customer base grew. This bottleneck threatened their ability to produce high-quality videos efficiently. Recognizing the need for more robust GPU infrastructure, Hour One turned to Google Cloud.
“We needed more GPUs to power our system. Without them we wouldn’t be able to produce high-quality video, our customers would have longer waits, and it would take longer to train our models. We couldn’t scale until we solved the GPU issue. Luckily, we solved it fast with Google Cloud,” said Amir Konigsberg, Founding Director and Head of Strategic Partnerships at Hour One.
Partnering with TeraSky for Seamless Migration
In 2022, Hour One migrated its infrastructure to Google Cloud GPUs, with TeraSky providing technical support and project management assistance. This partnership ensured a smooth transition and allowed Hour One to fully leverage Google Cloud’s capabilities.
Enhancing Video Production with Cloud GPUs
With the enhanced processing power of Cloud GPUs, Hour One achieved faster and higher-quality video rendering. The large capacity and 99.9% availability of these GPUs have enabled Hour One to consistently deliver more videos to its expanding customer base.
“Inference time matters because every time a request is made by a customer to generate a video, it relies on Google Cloud to power that request,” explained Konigsberg. “If the GPUs can’t keep up, there’s a delay. Also, if we want the virtual human characters to look more realistic, more expressive, we need more machines to process the data. With Cloud GPUs our videos are produced faster, our characters are more lifelike, and our customers are happier.”
Cost-Efficient and Scalable Computing
Google Cloud’s diverse range of GPUs allows Hour One to choose the most cost-effective options for various workloads, optimizing their compute resources. Running Cloud GPUs on Google Kubernetes Engine enables Hour One to scale its infrastructure dynamically, adapting to demand fluctuations and ensuring cost efficiency with per-second billing.
This strategic approach has resulted in a 28% reduction in inference costs, savings that Hour One passes on to its customers while reinvesting in further system improvements.
Driving Business Growth with a Global Reach
With its robust cloud infrastructure, Hour One has scaled its operations significantly, growing its customer base twentyfold. The upcoming availability of Hour One’s AI platform on the Google Cloud Marketplace will further extend its reach, making its services accessible to a global network of potential customers.
“Now other Google Cloud customers looking to include generative AI into their workflow will be able to access Hour One’s services directly from Google Cloud,” Konigsberg noted. “Leveraging the reach of Google Cloud is going to be a significant boost to our go-to-market efforts as we grow our business.”
Future-Focused Innovation
Looking ahead, Hour One plans to leverage its partnership with Google Cloud to make its core foundation models available to other companies through the Model Garden on Vertex AI. This initiative will enable businesses to integrate Hour One’s proprietary models into their AI projects, driving further innovation and growth.
“Thanks to Google Cloud infrastructure, Hour One has been able to create a solution that never existed before,” said Konigsberg. “Previously we were worried about how we could support 1,000 customers. Now, with Google Cloud, we’ve been able to grow from thousands of customers to ten times that number, without worrying about scaling our infrastructure. All we’re focused on is improving our product, winning more customers, and keeping them happy.”
In partnership with TeraSky and powered by Google Cloud, Hour One continues to push the boundaries of AI video creation, delivering exceptional value to its customers and driving the future of video content.