Artificial intelligence (A.I.) and machine learning are exciting concepts for companies and tech pros. For those who actually want to build something with those technologies, however, there are a few big constraints. Depending on your company, mission, and budget, the biggest limitation is usually cost: powerful hardware and smart people don’t come cheap. But it seems a few companies are moving to eliminate the infrastructure part of the equation. Earlier this month, at the Hot Chips 2017 conference, Microsoft unveiled what it calls Project Brainwave. This impressive bit of hardware consists of three layers with one purpose: speeding up the cloud-based delivery of deep-learning models. “Real-time AI is becoming increasingly important as cloud infrastructures process live data streams, whether they be search queries, videos, sensor streams, or interactions with users,” read Microsoft’s note on the project (which also features a handy breakdown of Brainwave’s hardware layers). The hardware stack will support Microsoft’s Cognitive Toolkit and Google’s Tensorflow, with a promised capacity for other platforms that may come online. “We are working to bring this powerful, real-time AI system to users in Azure, so that our customers can benefit from Project Brainwave directly, complementing the indirect access through our services such as Bing,” Microsoft’s note continued. “With the Project Brainwave system incorporated at scale and available to our customers, Microsoft Azure will have industry-leading capabilities for real-time AI.” Google already provides cloud-based A.I. services, including the Google Cloud Machine Learning Engine and the Cloud Vision API; other tools include text analysis, speech recognition, dynamic translation, and a jobs API (a deeper breakdown is available on Google’s Cloud Platform website). On the enterprise side of things, Salesforce’s Einstein Platform offers a selection of A.I.-based tools—albeit for those building on the Salesforce customer-centric platform. As more people use those services (and similar ones), they will only become faster and more sophisticated. Over the next few years, the number of cloud-based A.I. tools will doubtlessly proliferate. That may prove enormously beneficial for those firms that want to take advantage of things like machine learning, but don’t necessarily want to burn the time—and hire the staff—to come up with their own algorithms and platforms (and to be fair, the number of companies with the resources to build out an A.I. department is relatively small). But for those companies that want to set up highly customized A.I. platforms, or who are concerned about the security and privacy of third-party tools, the best option will likely remain in-house development teams. That may prove expensive—but if artificial intelligence really is as game-changing as many predict, that cost could be worth it. A.I. Migrating Fast Toward the Cloud
Artificial intelligence (A.I.) and machine learning are exciting concepts for companies and tech pros. For those who actually want to build something with those technologies, however, there are a few big constraints. Depending on your company, mission, and budget, the biggest limitation is usually cost: powerful hardware and smart people don’t come cheap. But it seems a few companies are moving to eliminate the infrastructure part of the equation. Earlier this month, at the Hot Chips 2017 conference, Microsoft unveiled what it calls Project Brainwave. This impressive bit of hardware consists of three layers with one purpose: speeding up the cloud-based delivery of deep-learning models. “Real-time AI is becoming increasingly important as cloud infrastructures process live data streams, whether they be search queries, videos, sensor streams, or interactions with users,” read Microsoft’s note on the project (which also features a handy breakdown of Brainwave’s hardware layers). The hardware stack will support Microsoft’s Cognitive Toolkit and Google’s Tensorflow, with a promised capacity for other platforms that may come online. “We are working to bring this powerful, real-time AI system to users in Azure, so that our customers can benefit from Project Brainwave directly, complementing the indirect access through our services such as Bing,” Microsoft’s note continued. “With the Project Brainwave system incorporated at scale and available to our customers, Microsoft Azure will have industry-leading capabilities for real-time AI.” Google already provides cloud-based A.I. services, including the Google Cloud Machine Learning Engine and the Cloud Vision API; other tools include text analysis, speech recognition, dynamic translation, and a jobs API (a deeper breakdown is available on Google’s Cloud Platform website). On the enterprise side of things, Salesforce’s Einstein Platform offers a selection of A.I.-based tools—albeit for those building on the Salesforce customer-centric platform. As more people use those services (and similar ones), they will only become faster and more sophisticated. Over the next few years, the number of cloud-based A.I. tools will doubtlessly proliferate. That may prove enormously beneficial for those firms that want to take advantage of things like machine learning, but don’t necessarily want to burn the time—and hire the staff—to come up with their own algorithms and platforms (and to be fair, the number of companies with the resources to build out an A.I. department is relatively small). But for those companies that want to set up highly customized A.I. platforms, or who are concerned about the security and privacy of third-party tools, the best option will likely remain in-house development teams. That may prove expensive—but if artificial intelligence really is as game-changing as many predict, that cost could be worth it. 
