Thomson Reuters has built seven datacentres containing internal clouds across the globe to feed information into client systems 20 times quicker.
Investment firms can now link directly into a Thomson Reuters cloud which is connected to trading venues. This cuts the time taken for data to reach traders.
The project, known as Elektron, saw Thomson Reuters develop the core software and middleware in-house. This includes offshore development in Bangkok and Beijing.
The project is a reaction to the growth of algorithmic trading. This highly automated trading means investment firm computers need to be connected to real-time information.
The information is collected and standardised in a cloud next to the trading venues then fed to the client company via dedicated high bandwidth links.
The clouds will also reduce customers' costs by hosting their infrastructures and reduce complexity by standardising information from many sources.
Until now data would travel from the trading venues to hub datacentres in three global locations where all the information is put into a standard format. It would then be sent to the customer.
Mike Powell, global head of Enterprise Information, described this as a hub and spoke model. "If you send all the data to a central hub you are increasing latency."
Every 200km reduction in fibre cuts one millisecond. A millisecond is a relatively long time when trades are today being completed on trading venues in microseconds.
Powell said, "Customers have started saying that the traditional way does not provide the performance required for algorithmic trading."
He added that increasing numbers of clients were connecting direct to trading venues, with Thomson Reuters' support, to get information quicker which can be expensive to set up.
So far the company has built seven datacentres in close proximity to the world's major trading centres. These are in New York, Chicago, Tokyo, Frankfurt, Singapore and two in London.
It is planning to build more near other major financial centres such in India and Brazil.
Powell said standardised information will reach customers an average of 20 times faster and that the investments required by customers to set up data feeds will also be slashed.
"If you want to set up your own network for direct feeds from trading venues it will take months but the hosting model makes it quick."
PJ Di Giammarino, CEO at think-tank JWG-IT, said there was increasing demand for better visibility of market data because more and more people relied on it.
"Not just for high frequency trading but also overall market intelligence to enabler companies to run models," added Di Giammarino.
He said the project would help to give investment firms that do not have an infrastructure access to the data they needed.
Thomson Reuters will soon announce details of another major IT project when it launches its video on demand news service. The service will give customers access to videos of news reports which can be found through searches for words used in a broadcast.
This, Powell said, is part of building a community and companies such as banks will be able to add their broadcasts to the service.