Verdigris – Artificial Intelligence for Energy Efficiency

August 5. 2017. 6 mins read
Table of contents

In a recent article, we wrote about 11 Smart City Solutions, we briefly touched on energy by highlighting a company called AutoGrid which creates “virtual power plants” of sorts. In this article we’re going to deep dive into the topic of using artificial intelligence (AI) for energy efficiency. It’s one of the 5 use cases in a recent paper by McKinsey which talked about how much money is being poured into AI these days:

Source: McKinsey&Company June 2017 Discussion Paper

You don’t think about your usage of electricity in the same way that you don’t think about breathing. Electricity has now become a part of our lives and we are constantly consuming it, in most cases for free. With the exception of the electricity bill you receive for your home (unless you’re lucky enough to be off the grid), you don’t pay for electricity directly. Electricity is now just a cost of doing business, and it’s priced into the products you buy and the services you use. With every single business out there using electricity, there is an opportunity for some serious cost savings if we can use artificial intelligence (AI) for energy efficiency.

Let’s try and think about the different ways we can apply artificial intelligence in the energy industry, in particular, electrical energy. Here are the three main areas that can be addressed with an example given for each:

  • Electricity producers – AI algorithms that can optimize the position of solar panels to maximize generation
  • Electricity suppliers – Using big data from “smart meters” to forecast demand
  • Electricity consumers – Using AI algorithms to optimize how we use electricity more efficiency

It’s that last use case that we’re particularly interested in. Companies like Google have already demonstrated their AI prowess by using Deepmind to reduce the energy used for cooling data centers by 40%. If someone says they can use AI to help you save electricity, then that claim should be accompanied by a chart that looks like this:

Source: World Economic Forum (Google’s Machine Learning Algorithms for Energy Efficiency)

Now that’s a compelling chart. For those of us who don’t have the resources of Google at our beck and call, there are plenty of startups we can turn to that are using AI for energy efficiency on the demand side. One such startup is Verdigris.

About Verdigris Technologies

Click for company websiteFounded in 2011, Silicon Valley startup Verdigris Technologies has taken in $16.1 million in funding so far from investors that include $5 billion electronics manufacturer Jabil (NYSE:JBL) and $198 billion telecommunications giant Verizon (NYSE:VZ). Their latest round of funding, a $6.7 million Series, closed in October of last year. They’ve also attracted the attention of Nvidia (NASDAQ:NVDA) who blogged about how Verdigris “trains its models on Pascal architecture-based NVIDIA GPUs” and that this “helps Verdigris train models 20 times as fast as on CPUs“. That same article goes on to say how “buildings gobble up about 70 percent of the world’s electricity — and waste 60 percent of it” which amounts to about $100 billion of wasted electricity.

Update 11/11/2018: Verdigris took in an additional $5 million funding round at the beginning of this year bringing total funding to $21.6 million.

So what does Verdigris do exactly? A key part of their solution is a piece of Internet of Things (IoT) hardware that can be installed by any qualified electrician and immediately starts relaying information from 42 on-board sensors up to “the cloud” where all the magic happens. The installation takes about 30-120 minutes and here’s the installation guide if you’re interested in the technical details. Below you can see the data transmitter that talks to the cloud using Verizon 4G/LTE or even your own WiFi network:

AI for Energy
Verdigris Data Transmitter

They say that the best startups come about when the founder is addressing a pain point that they personally experienced and decided to solve for. Verdigris is no exception. The founder’s unexpectedly high electricity bill couldn’t be explained by “smart meters” on the market today so he created an “electrical map” of his home which uncovered a faulty fuel pump that was causing a current overload. That’s the same idea behind this solution. The sensors create the “electrical map” and then collect data over time which results in an average 30-50% savings on your monthly electricity bill. There’s also the added benefit of that “electrical map” telling you when things stop working, like light bulbs. Imagine how handy that might be for hotels, like this one.

People spend 92% of their time in buildings, and Verdigris wants buildings to take care of people instead of the other way around. Imagine how this tool might be used by property managers to gain insights about each tenant that occupies a floor in a commercial building. Dashboards like the one seen below can help identify trends that can lead to cost savings:

Artificial intelligence for energy efficiency
Verdigris Dashboard

If you’re going to upgrade to some green technology like low-energy AC units or environmentally friendly LEDs, then you can literally see the savings materialize right before your eyes – at least you better, else you have some pretty compelling evidence that you can present to the salesman who sold you those units. The system is currently certified for use in 38 countries including the United States, European Union (32 countries), China, Malaysia, Mexico, India, Canada. As for pricing, you’ll need to contact them for details. We’d hope they adopt a pricing model like Afiniti with little or no upfront costs and then a monthly charge which is easily covered by all the savings you’ll realize after adopting the solution.

Verdigris has a number of case studies you can peruse which show how real-world implementations of this technology are actually solving problems and creating efficiencies. For example, Orchard Hotel Group in San Francisco implemented the solution in two hotels and immediately identified the following problems:

  • A broken chilled water pump that maintenance checks had missed, leading to overloading of the paired pump
  • No activity on stairwell pressurization fans, required by building code
  • Minimal activity on corridor exhaust fans
  • The standby heater for the emergency generator shut down for two weeks

In addition to that, the hotels upgraded all their light bulbs to realize immediate savings based on experiments they were able to run. They also found out that a vendor sold them a fan which promised 30% savings but in fact only 18% was realized. The 15-month payback they were promised by the vendor was in fact, much longer. Here’s how they did it:

The hotel was also able to compare two very similar properties to discover the effects of a newer keycard installation on energy usage.

In another hotel, Verdigris examined their commercial kitchen and found that a pizza oven, the light in a walk-in refrigerator, hood lights over a stove, and the kitchen TV were using more than $7,000 of electricity each year which could be eliminated by installing motion sensors. In the same kitchen, a dishwasher heating element was staying on at all hours costing the hotel $5,000 a year in unnecessary electricity usage.

Verdigris certainly isn’t alone in their quest to reduce electricity consumption in commercial buildings. One of the barriers to entry they’ve out in place is US Patent 8,694,291 which protects the method they use to achieve the granularity needed for their “electricity maps”:

System and method of waveform analysis to identify and characterize power-consuming devices on electrical circuits

Still, there are many ways to skin a cat, and other companies like Alpiq and Open Energi are also looking at using artificial intelligence for energy efficiency on the demand side. According to research firm Verdantix, global facility management company ISS World has partnered with IBM Watson unit to deploy its AI across 25,000 buildings within its portfolio. Elevator company KONE will also use IBM Watson to monitor and optimize its portfolio of elevators, escalators, doors and turnstiles in commercial buildings across the globe.

As it turns out though, the total addressable market or TAM that these companies are looking to address is simply massive. According to the U.S. Energy Information Administration, there were 5.6 million commercial buildings in the U.S. in 2012 accounting for 87 billion square feet of space. Electricity accounts for 61% of all energy consumed by those buildings, a number that’s been steadily increasing over time. Here are the types of buildings sucking up the most energy:


With big bold numbers like that, it’s no wonder that one of the angel investors in Verdigris has said that he believes this will be his most successful investment to-date.


Leave a Reply

Your email address will not be published.