Amit Gandhi, Chief Economist at Microsoft Azure and a tenured professor at University of Pennsylvania, presented at Luohan Academy's Frontier Dialogue. Steve Tadelis, Professor of Economics at UC Berkeley, discussed Amit's presentation. The following text features transcripted excerpts from. It has been lightly edited for clarity and length.
Speaker presentation by Amit Gandhi
Thank you Long. The next topic is "Top business survive and thrive in the process of zero-net transformation". The presenter is professor Amit Gandhi. Amit is the Chief Economist at Microsoft Azure and also a tenured professor at University of Pennsylvania. Amit, please take it away.
Thank you to the Academy and the organizers for the invitation, the incredible audience and panel. As mentioned, I'm working as an economist at Microsoft for several years and also here, the University of Pennsylvania. I'm coming to you today, largely with my Microsoft head on, I will add some economic points of view as we go along, but really I just want to give you a little bit of perspective into Microsoft's journey into the net-zero goals and targets for both the company and the world as a whole.
I think just start out this, zoom out for a moment before we jump into Microsoft. What is it that Microsoft is looking at as a problem. The basic science, is pretty grim. So we need to keep temperature no more than 1.5 celsius higher than pre-industrial levels by the end of the century. The earth has already worn 1.2 ℃ relative to that goal. So there's not a lot of margin left there. And if we just look at basic science who says, and this is what the net-zero initiative arises, that we need to stop adding more carbon to the atmosphere by 2050 by the middle of the century. We think about carbon math. It's simple, but it's worthwhile point that it's not zero, it is emissions minus removal. This is basic credit and debit system here. We need to hit emissions equal removals by this 2050 time period. That's the global macro scenario.
Now, where are we today? So I think it's worthwhile pointing out that the world, even the developing world is still industrializing. So we're in the midst of, sometimes known as the 4th industrial revolution. This is the coming together of our digital and physical spheres. It's powered by devices, internet of things, machine learning, artificial intelligence. They're all coming together in new and normal ways to solve new problems and build new applications.
When you think about the industrial process that sits behind this 4th industrial revolution, it's depicted here in this picture on the left.
And what is that is a data center. The data centers, modern day, industrial complex, sort of the factories of the future. And what is it that data centers produce? They produce applications, the applications they get streamed to end users through high-speed networks. We are sitting on the zoom call right now. We are consuming a stream, which is an application that's produced by underlying data center. That data center happens to be an oracle data center on a Microsoft data center.
But nevertheless, this is sort of the picture of what this 4th industrial revolution entails.
And this matters because what is powering this. What is the oil that goes into these factories? And this sort of gets into the famous expression. Many of you probably heard of this notion of "The data is the New Oil". I think it's perhaps a bit of a happening expression at this point. But when you really think about it, it has multiple layers of truth associated to it. So I think Hal very recently has pointed out that like oil, in order to be useful, a data needs to be refined. So raw data often does not solve problems, but it needs to get processed through machine learning and modeling and data science in order to get into an interpretable decision posture.
So that's kind of one prism through which the metaphor makes sense. But there's another real sense of the analogy here, which is data literally has to be burned. What's going into these data centers are servers, which are depreciating over a roughly 2 to 4-year time scale. And what goes into the servers is data. When we think about applications of the electricity, carbon footprint that they have it, it's pretty remarkable. So here's a step that comes from a recent study. It just shows at the bottom there. You can see the carbon footprint of training, a high dimensional machine learning, natural language [processing] model. This would be a model of the form of GPT three or bird-style architectures. And we can see it's about 626 thousand pounds of CO2, which is 5 times the equivalent of an average US car.
So we are literally burning resources to produce these applications. So I think that's worthwhile bearing in mind that industrialization is still happening. It's just taking a very reform. What is that Microsoft is seeing in terms of the solution? And I think this is where some of the hopeful, optimistic side of the equation starts to get realized is that there are pathways here. I think the speakers who [I] just heard from [talked about] what some of those pathways look like from the market side. This is a stylized scenario. But I think the key part to really emphasize here is that it's a portfolio approach to the solution.
And let me kind of call out three major elements that certainly Microsoft sees. When it thinks about its strategy, one is just efficiency. We have to get more efficient about emissions, which is consumption of energy resources to power our lives and businesses. But another important part is zero carbon electricity, where sometimes known as the decarbonization of the electricity grid. So we need to move electricity to renewable that do not have a carbon footprint. I'll talk more about that momentarily, but if we get to zero carbon electricity, then we need to electrify, move applications that currently rely on fossil fuels, to electricity sources. And then the 3rd element here is removal, actually physically extracting or pulling carbon from the atmosphere is another arrow in the quiver. So I think it's really important to realize that it is a portfolio approach and we have to bring all of these approaches to bear on the problem and sort of with that as backdrop.
What is Microsoft's decision on this very pressing problem? First of all, Microsoft has a very robust, sustainability strategy today. And I want to emphasize that sustainability of Microsoft means more than just carbon. Sometimes it gets forgotten that it's carbon, water, waste, and ecosystem more broadly It happens to be that carbon is the most mature and developed of those domains. But sustainability of Microsoft means all these things. And they're really given equal attention at some level. But carbon has certainly made the most hand. Last year Microsoft put out a pretty pre-remarkable goal for itself, to which it's floating the company accountable to be carbon negative by 2030.
So that's Brad Smith on the left, the president of Microsoft on the left, [promised] to remove all of our historical carbon emissions by 2050. We have something on the right, which is complemented by 1/3 pillar, which is $1 billion climate innovation fund. That is reinvesting back into market places and technologies for helping address and solve this problem.
And as we think about it, I want to, give some complexion and color into what that exactly means. So what does it mean to be carbon neutral? And the carbon math here is really important how we think about the accounting. And so when we think about emissions from company perspective, there are three distinct scopes, scope 1, scope 2, scope 3. Scope 1 is directed to direct emissions. Just what your own business activities emit into the atmosphere. Scope 2, includes scope 1, but it adds electricity consumption that goes into running at the business. So that's a multiplier on top of scope 1. So we go from Microsoft has today about 100,000 metric tons of scope 1, about 4 million metric tons of scope2. And scope 3 is emissions from the entire supply chain. So it includes the carbon footprint of all the input purchased materials, outsourced activities, contractor vehicles, business travel, et cetera. And that's gonna be another multiplier on top of scope 2. So today for Microsoft, that's about 12 million metric tons. And so when Microsoft aims at a net-zero goal, it actually has scope 3 in mind, the full scope of all emissions, it's really the key to the solution. And another sort of important, we don't talk about carbon neutrality anymore. This is about net-zero. And that's really important because neutrality implies the purchase for investment in offsets. And that's not tied to getting us to the net-zero goal.
And let me just give you a picture of what that looks like. From a Microsoft perspective, this is a pathway of how Microsoft tends to get to its goal of net zero emissions by 2030. Black is the actual net emissions. The scope 3, output emissions are the blue bars. We can see them peeking and now they're starting to go down. I'll kind of explain a little bit more as to how they go down. The dark green were the offsets that Microsoft used to employ as part of its carbon neutral strategy. And you can see in the light green that's now getting replaced by active investments in carbon removal from the atmosphere.
Not today, carbon removal is 1% technology or engineering based on 99% neutral-based solutions to get to the end of the century goal.
Let me mention what do you mean just pause by just and by saying kind of. Now we sort of understand the strategy, which is mentioned one thing really fast in which renewables are a key source of how emissions are removed. That's we are now pretty much almost index to about 60 % renewable today. We're looking at about 100 % renewables for our electricity consumption in 2025.
Let me close by saying, why is Microsoft doing this? And if you were to ask Satya Nadella on the phone, he'll tell you something along these sides, "At a time when many are calling attention to the role technology plays in society, our mission remains constant. It grounds us in the enormous opportunity and responsibility we have to ensure that the technology we create always benefits everyone on the planet, including the planet itself".
This is an example of emission-based approach to the problem. And you will see many companies and buyers this mission-based approach the problem, which is sometimes called stakeholder capitalism, which is to say, the modern enterprise does not think about shareholders as the only stakeholder it is shareholders, customers, suppliers and partner, employees. You can see saying that the stakeholder is the planet itself. One of the key ideas of stakeholder capitalism is that all the stakeholders have mutually aligned interests in the end. And let me let me say that I think that's a very powerful view, but I don't think it actually departs from just the basic economy to the situation.
So let me give you a little bit of the dismal science view on this. One reason Microsoft is doing this is that because it's feasible and desirable. So Microsoft is partly doing it because it can do it. It has immense resources. It has technological capabilities to experiment, optimize and implement sustainability programs. It acquires learning and knowledge transfers to the rest of the economy through multiple channels, and touchpoints. But the other reason, the desirability part of the question, I'll just conclude when I run through these points is that sustainability of the day is a long run differentiator.
If we think about going to a supermarket and scanning your cereal box today, we might look at nutrition, we might look at the picture of the serial box in the future. We're gonna be able to see the entire supply chain history of that cereal box was produced. And the key idea here is that sustainability is a source of product differentiation at the end of the day.
And that the theory that it is gonna drive spending behavior as spending power moves to millennial and generation Z. The last point is that sustainability focus just it's good for business, it improves efficiency. I have not come across a single sustainability program. And Microsoft where it didn't just make business sense to do it, because a lot of it goes into succeeding and sustainability. It's just running your business smarter, faster using the power of data ad AI. AndI'll just close by saying you can't affect what you can't measure. A lot of hard part with sustainability is just being able to measure your business activity. But being able to measure your business activity and it's carbon impact is a major shift into that digitalization. Digital transform modern company. And I think that's gonna be a critical, essential part of the journey. Both in Microsoft, and the economy.
Discussant presentation by Steve Tadelis
Thank you very much, Amit. Good to see you after a while and learn about Microsoft's efforts. I was asked to comment on this full disclosure. I'm not an energy economist. The good news is that almost no one on this call seems to be. I guess we're all trying to figure out what's going on here. And when I thought about the presentation and what on it, big tech and net-zero. How is this going to imply anything about the world?
Surprisingly, I did a quick search and only less than 4 months ago. There was this article from Bloomberg Green, April 7th, 2021, where it states that of the ten largest US companies by market value, only four have announced plans to reduce their emissions by net-zero by 2050. As it turns out, they all happen to be technology companies, Apple, Microsoft, Amazon, and Facebook. It turns out that Google claims have already been achieving that target, being fully offsetting its emissions since 2007. It is interesting that we have this move.
And questions come to my mind. What kind of impact is this going to have? Since this is really a global issue with the familiar "commons" problem. So the prisoner's dilemma, the tragedy of the commons, many names to the same phenomenon where we're all suffering from global warming, namely, gave a great overview of just some recent anecdotes.
Of course, once you have more than three or four anecdotes that all have happened in very recent history, it starts looking like a data set. And the first thing that comes to mind is: Can Big-Tech really move the needle, or is it a form of greenwashing?
One of the things that Amit described towards the end is that, in fact, for many of these efforts, it seems to be what's called doing good and doing well. In the social responsibility angle, namely, companies are doing good for the environment. But by doing so, they're also saving costs, and so on and so forth. So there may indeed be just pure incentives for companies to pay a little more attention to their footprint. There's more and more research that's coming out about the so-called behavioral IO, which means companies act in ways where they're really not maximizing profits. They're focusing on a few goals, a few KPIs. But maybe as people start really looking under the hood of how their emissions and footprint in terms of the environment, they are affecting things or maybe ways to hit that double bottom line.
So at the same time, we have to remember something about the scope. So I did a little bit of searching. It turns out that Apple has given about 25 million tons of emissions. Kazakhstan, which is No. 20 on the list of the mission producers, has about 390 megatons as far as I know. This means that you need something like fourteen Apples to be one Kazakhstan. Now, on one hand, that's actually Apple has quite a big footprint. And I'm sure Amazon is quite substantial given the logistics and everything that goes around. But when you look at the Top 20 countries, that makes all the big tech in the world look like a tiny dwarf. The question is how could this have an impact?
Another question that struck me of interest is following some stuff I read by my good friend and colleague Severin Bernstein at the Energy Institute in UC Berkeley. He's written some lovely blocks about related issues. Also, stuff I read by Jim Bushnell from UC Davis, who's also part of the Institute. The question is: Are zero targets helpful or harmful?
As economists, we know that what we'd like to do is, to have the social marginal cost equal to the social marginal benefit. And it may indeed be going beyond net-zero. The kind of effort that I've referred to, is more effective than net-zero. We know from, again, the behavioral incentive theory, that when you set targets and you don't meet them, it could cause a backlash. It could cause people to be unhappy with what's going on, and maybe even abandon these efforts. There's some nice jargon that goes on here with the net-zero impact. But at the same time that may be a little too simplistic, we want to be more nuanced.
The next thing that comes to mind is what are the distributional impacts? Something that I learned recently is that, in California, more folks are using the solar at the home level and they push it back into the network and so on and so forth. But what's really happening is that the people who could afford it are wealthy people. It is just like, maybe among businesses that could afford to go down this route, even if it does do good and do well in the longer term, are the larger cash and heavy businesses.
Now, what does that mean? It means they reduce the marginal use of the network. But the network has a tremendous amount of fixed costs that need to be covered. This is why these industries are typically natural monopolies and are regulated. What happens as the wealthier and Big-Tech, starts moving off the grid, so to speak, imposing the cost of the grid on whoever's left, which typically seems to be poor people. There's an interestingly regressive tax aspect of the wealthier moving on to net-zero.
And to conclude, I want to leave more room for discussion to think about what might be a channel. The Big-Tech could really have an influence on moving the needle given that they are a drop in the bucket. Big-tech is in the news and a lot of Big-Techs receive headlines. Maybe Big-Tech can influence public opinions. And if public opinions move enough, it will loop into politics and policy change. This kind of effort that I described is where I want to feel optimistic. Microsoft and other companies, as we see from the Bloomberg article, are doing things that may have a much bigger impact than their footprint by themselves.
At the same time, I look at what's going on with vaccines in the United States, where there's so much discussion. And the way people respond is just mind-boggling. I'm not sure if public opinion is the channel that could indeed move the needle. I still want to be hopeful that it can.
 Bloomberg Green, Tech Companies Are Setting the Most Ambitious Net-Zero Goals, April 7, 2021
For more information, please visit Luohan Academy's youtube channel: Luohan Academy