When you store and deliver data at Shutterstock\u2019s scale, the flexibility and elasticity of the cloud is a huge win, freeing you from the burden of costly, high-maintenance data centers. But for the New York-based provider of stock photography, footage, and music, it\u2019s the innovation edge that makes the cloud picture perfect for its business.\n\n\u201cThe speed of innovation is really starting to accelerate,\u201d says Jefferson Frazer, director of edge compute, delivery, and storage at Shutterstock, which is headquartered in the Empire State Building. \u201cIf you\u2019re not keeping up, you\u2019re getting left behind.\u201d\n\nAdvancements in analytics and AI as well as support for unstructured data in centralized data lakes are key benefits of doing business in the cloud, and Shutterstock is capitalizing on its cloud foundation, creating new revenue streams and business models using the cloud and data lakes as key components of its innovation platform.\n\nThe company, which customizes, sells, and licenses more than one billion images, videos, and music clips from its mammoth catalog stored on AWS and Snowflake to media and marketing companies or any customer requiring digital content, currently stores more than 60 petabytes of objects, assets, and descriptors across its distributed data store.\n\nBut it\u2019s the ability to tap sophisticated analytics and AI in the cloud, combined with the \u201cdemocratization of data\u201d enabled by data lakes, that is not only accelerating innovation at Shutterstock but also facilitating new products and services, Frazer says.\n\n\u201cThe expectation from developers is that they can go faster than they\u2019ve ever gone before and that every part of the lifecycle around this data needs to be elastic, scalable,\u201d he says. \u201cNothing can be held back from giving everyone in your business democratized equal access to this information so they can leverage it to do their part of the job.\u201d\n\nThe challenge for any enterprise, he says, is finding a centralized path to access disparate stores.\n\n\u201cWe think we found a good balance there. We use Snowflake very heavily as our primary data querying engine to cross all of our distributed boundaries because we pull in from structured and non-structured data stores and flat objects that have no structure,\u201d Frazer says. \u201cThen coupling with AWS\u2019 strong authentication mechanisms, we can say with certainty that we have security and restrictions around who can access data.\u201d\n\nThis level of development is very complex and only possible with a skilled CIO who has a deep understanding of all business processes and new cloud technologies as soon as they are made available, Frazer says.\n\nCloud-first, cloud-fast\n\nFrazer believes Shutterstock CIO Hugues Hervouet has just the right blend of tech know-how and business acumen to pinpoint which parts of the company are using data to its full potential, which could capitalize more, and where opportunities for expansion and cross-functional use reside. \n\n\u201cOur CIO is particularly invested in pushing forward data consumption,\u201d Frazer says. \u201cAs soon as new cloud features come out, they are immediately consumed.\u201d\n\nHervouet himself says he is driving his developers to innovate faster and develop new classes of applications as soon as new cloud capabilities are released. Shutterstock\u2019s current focus, for instance, is generative AI \u2014 considered by many to be a bleeding-edge application.\n\n\u201cWe have been able to reallocate engineers to work on value-add activities, such as implementing a generative AI solution that enables our customers to create compelling images using the platform by describing what they are looking for in just a couple of sentences,\u201d Hervouet says, noting this enables customers to find and create content they need much faster.\n\nFrazer says Shutterstock has a dedicated team building AI algorithms and new machine learning models that are integrated into all aspects of the customer lifecycle, such as an engine that learns from customer consumption patterns and makes recommendations. To do so, the team leverages tools from AWS and Databricks, as well as custom Jupyter notebooks.\n\nFor Shutterstock, the benefits of AI have been immediately apparent. Storage intelligence, for example, has reduced the duplication of images, an issue that occurs after acquisitions. And generative AI has helped reduce the time required to prepare custom images for customers.\n\n\u201cWhat we\u2019ve seen from the cloud is being able to adapt to the complexities of different data structures much faster,\u201d Frazer points out. \u201cPrevious tasks such as changing a watermark on an image or changing metadata tagging would take months of preparation for the storage and compute we\u2019d need. Now that\u2019s down to a number of hours.\u201d\n\nShutterstock is also working with OpenAI, using their \u201cmodels to generate content now trained off of our datasets,\u201d Frazer says.\n\nOptimizing for innovation\n\nAnalytics in cloud is also proving key to Shutterstock operations. The company relies on Amazon QuickSight and Athena to add visualizations and perform deep queries on its data to ensure optimal performance across the application lifecycle, Frazer says.\n\n\u201cAnalytics doesn\u2019t just stop at performance,\u201d he says. \u201cWe want to understand everything that the customer is doing on our website. Why didn\u2019t they click on this button? The customer hovered for two seconds and didn\u2019t click that type of data. That is invaluable when optimizing your site.\u201d\n\nOther services such as Amazon CloudFront enable Shutterstock customers to enhance their content-on-demand networks, and Lambda \u2014 a serverless compute service that runs code without having to provision or manage servers \u2014 benefits Shutterstock customers wherever they are in the world, he says.\n\nFor Shutterstock, the cloud has led to faster innovation, but few enterprises are capable of exploiting sophisticated features out of the gate and ought to proceed cautiously with advanced cloud services, says IDC analyst Dave McCarthy.\n\nWhile \u201cthe cloud gives enterprises access to the latest technologies with the ability to provision new resources in minutes, cloud providers are releasing new capabilities faster than enterprises can consume them,\u201d McCarthy says.\n\nGartner analyst Arun Chandrasekaran adds that accelerated innovation in the cloud offers a high risk\/reward ratio \u201cdisruptors can leverage\u201d and creates a dynamic work environment to attract top talent. But there are pitfalls to innovating too quickly, particularly if the enterprise lacks a cohesive strategy and direction, he says.\n\n\u201cIt can lead to too much experimentation and lack of clear business value from such projects,\u201d Chandrasekaran says, as well as \u201cpotentially lower reliability and more firefighting than true innovation.\u201d\n\nEven those organizations with the talent to tackle cutting-edge technologies in the cloud can be slowed by the nature of their business environments, McCarthy says. \u201cMany companies find themselves in a hybrid architecture where they have one foot in the old world and one in the new,\u201d he says. \u201cThat creates some unique challenges in how to manage both environments consistently.\u201d\n\nStill, the drumbeat for innovation marches on. \u201cCIOs need to think of digital transformation in the context of continuous innovation,\u201d McCarthy says. \u201cIt should not be considered a one-time exercise, but rather an ongoing process where new technology becomes embedded into the business as it becomes available.\u201d\n\nFor Shutterstock, that process is a facet of the company\u2019s culture, thanks to strong IT leadership, a robust cloud infrastructure, a diverse toolset, and talent, Frazer says.