A mashup of 3D printing, computer vision, machine learning and other advanced technologies.
It’s an exciting time at the intersection of robotics, computer vision, AI, machine learning, deep learning/neural networks and 3D printing. We roamed MIT’s campus earlier this week as we were on site for a 3D printing class. While unsuccessful in our quest to 3D-print a robot of our own, we did spot a few delivery bots on campus. Below are snippets of content we found interesting about one of MIT’s more famous spin-outs, Boston Dynamics. We are working to get founder Marc Raibert on the TEK2day Podcast. The Boston Dynamics robots operate much like an autonomous vehicle, leveraging LiDAR technology to power “vision” and spatial awareness.
Sure, Amazon’s Alexa-powered devices are nifty, but there’s only one AI king of the jungle – Google. Google benefits from more nodes on the network – i.e. more input data from which to learn – than any other single technology platform. “Google By the Numbers” or “How We Feed the AI”:
Google processes 63,000+ search queries each second
3.8 million searches per minute
227 million searches per hour
5.4 billion searches per day
2 trillion searches per year
15% of all Google searches have never been searched previously
More than 2 billion people with Chrome browsers. Chrome serves as a highly sophisticated data collection engine
People watch over a billion hours of video each day
Approximately 2 billion logged-in users watch video games on YouTube each month.
Passive to Active: as Google releases its cloud gaming offering – Stadia – many of today’s “passive” YouTube video game views will convert to “active” views. This translates to a richer data input set for Google’s data centers/AI / machine learning layer.
Earlier this week Amazon’s AWS unit and automobile giant VW announced a joint effort to build the “Industrial Cloud”. This effort delivers on the Cloud vision from more than a decade ago of driving operating efficiency by migrating large, complete enterprise data sets to the cloud. Massively scaled, secure cloud environments like AWS enable interoperability between otherwise disparate data sets. Further, scaled cloud environments are optimal for data-driven processes such as AI and machine learning given that the cloud facilitates data sharing. The greater the number of high quality data sources (inputs), the faster the machine learns. We covered the Industrial Cloud in a recent podcast.
Highlights from the Press Release:
Volkswagen and Amazon Web Services (AWS) are to develop the Volkswagen Industrial Cloud together.
The Volkswagen Industrial Cloud will combine the data of all machines, plants and systems from all 122 facilities of the Volkswagen Group.
In the long term, the global supply chain of the Volkswagen Group with more than 30,000 locations of over 1,500 suppliers and partner companies could also be integrated.
Volkswagen is creating its Industrial Cloud as an open industry platform which other partners from industry, logistics and sales may use in the future.
Production planning and inventory management is to be standardized and networked across all 122 production plants of the Volkswagen Group.
The gaming industry – the largest segment of the broader entertainment industry – continues to grow and evolve. Google is set to join Technology giants Amazon (Twitch) and Microsoft (Xbox) in the gaming space – formal announcement coming on March 19th. Not only are the industry players changing, but so is the underlying AI technology. Autonomous game design is here.
Many insurers are hesitant to invest aggressively in AI & ML initiatives given that the ROIC is difficult to quantify. This conservatism combined with market disruption bodes well for technology & service providers that offer an Outsourcing option with a demonstrable ROI. Examples include Accenture, EXL Service, IBM Global Services and SS&C Technologies. Plus -…
Financial Institutions May Recover Lost Glory If They Aggressively Adopt Technology Fee Compression As Far As The Eye Can See Long gone are the days of white-collared shirts, suspenders and fat fees. The Financial Services industry is reducing headcount across the board. No pocket of the capital markets is immune. Investment Banks, Depository Institutions and…
For the past few years I’ve poked around the machine learning (“ML”) and artificial intelligence (“AI”) space. I advised Boston-based DataRobot back in 2014 when they started to build their machine learning platform. I’ve thought about how we at CEORater may leverage ML to score CEOs and companies. Typically when we read about ML and AI it’s from the perspective of a pure-play vendor who markets and licenses its platform across multiple industries for a variety of use cases. Often the use cases we read about are focused on “power users” – people who have a PhD in Statistics or some similar quantitative background.
From a background perspective, Asset Management firms of all flavors (small, mid-sized and large, traditional, hedge funds, private equity etc.), Fund Administrators and Insurers use a variety of SS&C products and services to value assets (equity and fixed income securities, derivatives, bank loans, private placements and real assets to name a few asset classes)/ strike an NAV, settle trades and report on asset holdings. The company’s Singularity initiative will replace siloed products with a common ML-based core layer that will have modular AI and RPA services that sit on top.
Multi-tenant machine learning is a significant competitive differentiator. Some readers pride themselves on identifying businesses that have a competitive “moat”. For non-investors a “moat” is a source of sustainable competitive differentiation. Challengers who wish to compete against companies with established moats best be prepared to completely shift the paradigm and render the moat obsolete. You’re simply not going to spend your way around, over or through a moat. Brute force won’t work. If any company ever had a moat, SS&C has one in the world of portfolio accounting systems.
SS&C’s moat is about to get significantly wider and deeper as Singularity is rolled out. This is in no small part due to the multi-tenant machine learning layer. This means that as Customer X has an experience that requires a “learning”, the benefit of that learning is enjoyed not only by Customer X but also by the other customers on the platform. This multi-tenant element to Singularity’s machine learning layer is a powerful scale differentiator primarily for three reasons:
Large installed customer base: SS&C has a great many customers and users – therefore more opportunities for machine-driven learnings – the benefits of which accrue to all SS&C Singularity customers.
Purpose-built from the ground up: SS&C has incorporated machine learning into Singularity from Day One, providing the company with a significant and sustainable advantage over competitors who may try to retrofit a third-party’s machine learning layer on top of legacy products and services. Retrofitting legacy technology simply can not be as effective from a throughput and efficiency standpoint as a new, modern-architected platform.
Cost prohibitive: It’s not an insignificant dollar amount that’s required to build a modern, ML/ AI/ RPA-powered Fintech platform from scratch. To replicate Singularity from a domain-expertise and technology perspective would be cost prohibitive.
VC’s would be wise to avoid trying to disrupt this market. As I see it, the only way to replicate what SS&C has built would be to acquire the company.