Capital markets firms are increasing their use of cloud computing at an ever-accelerating rate, with the industry planning to consume and deliver more services via the cloud over the next 12 months, according to a survey conducted by Google and research firm Coalition Greenwich. However, how and where buy-side and sell-side firms are using the cloud differs significantly.

For example, while 90% of the buy side is consuming cloud-deployed data in some way, only 67% of sell-side firms currently use cloud for any data.

Between February and April of this year, Coalition Greenwich conducted interviews with executives at 102 organizations spanning exchanges, trading systems, information providers, aggregators, hedge funds, asset managers, investment managers, commercial banks, and investment banks to better understand their approaches to market data consumption and distribution in the cloud, as well as the use of trading systems in the cloud. Respondents were primarily based in North America, followed by EMEA then Asia. 

The research found the priorities of different industry segments vary. For example, while the buy side’s top uses of cloud are for portfolio management (79%), real-time market data (61%) and trade order management (52%), the sell-side’s current top uses of cloud are for data analytics (88%), risk management and risk analytics (75%—compared to 46% for the buy side), and then a large drop in usage to the third use case, market data (38%).

In other areas, only 13% of sell-side respondents reported using cloud for trade execution, while 25% each reported using cloud to support data-intensive functions such as back-testing and quantitative strategies. Similarly, on the buy side, 21% reported using cloud for back-testing and portfolio simulations.

David Easthope, senior analyst at Coalition Greenwich, says that “back testing is indeed a strong use case for the buy side, but it is simply dwarfed by the other use cases like portfolio management, real-time market data, and trade order management. There are just so many firms using cloud for that. Same for the sell side, the other use cases are just so much more widely observed. Not every firm is going to back-test strategies to the same extent. It depends on the investment mandate and trading strategies.”

Accessibility was cited overwhelmingly by buy-side firms as a reason for using cloud, while sell-side firms—with average data budgets of $140 million, three times those of buy-side firms—saw cost reduction as a key driver for adoption.

“The thing that struck me the most was the reason why people are using cloud for market data,” says Philip Moyer, vice president of strategic industries at Google. “In the past, speed was the reason that people bought services. Now, it’s accessibility—and I equate that to agility. It’s not just about cost—firms need to be able to scale up and down. For example, crypto—it’s a $1.7 trillion asset class that trades all over the world, not on a single exchange. Firms need to be able to go where the markets are, and to be able to distribute services anywhere.”

Similarly, if a firm enters a new market with different risk requirements, or if risk requirements in its home market change, it can easily spin up systems to perform risk calculations. “You can do, for example, risk calculations more frequently and pay for it like electricity—when the switch is off, you don’t pay—so you don’t have fixed costs associated with starting a new business,” Moyer adds.

The same is true for exchanges and data providers, which Moyer sees stepping up the provision of data via Google’s cloud. Of exchanges and trading platforms surveyed, 83% are currently providing access to data via the cloud, with 100% planning to do so over the next 12 months, focusing initially on derived data, reference data, and end-of-day datasets.

AI as cloud driver

Another key growth area being enabled by greater cloud adoption is the use of artificial intelligence and machine learning, with 50% of exchanges surveyed already offering data services powered by AI/ML, and 50% planning to offer AI-powered trade execution and trading analytics over the next 12 months. Among end-user firms, 55% of the sell side and 14% of the buy side currently use AI/ML in the cloud. According to the report, the sell side is taking advantage of cloud’s increased processing power and on-demand computing to leverage AI/ML to support functions including investment research, transaction cost analysis, and risk analytics, whereas cloud providers still need to better demonstrate the benefits of AI/ML in the cloud for the buy side.

“With AI, we are seeing a rise in the use of rich algorithms to make decisions for settlement, clearing, and for valuation of assets. So I think there is a significant opportunity for the buy side, as well as for exchanges and banks,” Moyer says. However, he adds that key requirements when using AI/ML are data performance and consistency, as well as security to safeguard sensitive data and calculations.

“Most financial services firms have requirements about whether their data traverses the public internet,” Moyer says, adding that Google Cloud owns and operates one of the largest fiber-optic networks in the world, with hundreds of points of presence around the globe connected by undersea cables, which not only enables firms to spin up access to new geographical markets and experiment with new business, but does so with low jitter and high security.

“In many cases we’re able to go end-to-end without going over the public internet. And we have encryption when data is at rest, when it’s in motion, and when it’s in process, meaning that if you’re doing risk calculations or creating trading ideas around the world, we can provide a fully encrypted experience,” he says.

And though AI/ML is eighth on firms’ current and planned uses for public cloud, firms highlighted a range of areas where they expect that it will help them improve efforts in the future, from faster regulatory/compliance/risk reviews (28%), cleaning and organizing data (27%), explaining predictions (15%), faster training of ML models (12%), and faster time to production (10%).

 Lowering the cost of computing for training and deploying ML models was ranked sixth with 8%, with the research concluding that firms see AI/ML as a way to deliver “more rapid responses” to strategic challenges and opportunities, such as risk initiatives and “commercially lucrative use cases.”