Tech Won't Save Us - The Dirty Alliance Between Tech and the Oil Industry w/ JS Tan

Episode Date: April 3, 2025

Paris Marx is joined by JS Tan to discuss his experience seeing first hand how Microsoft deployed its cloud and machine learning services to help Chevron extract more oil and gas, and the state of tech worker organizing around climate change. JS Tan is a PhD student at MIT, researching cloud computing in the US and China. He’s a a member of Collective Action in Tech and writes the Value Added newsletter.Tech Won’t Save Us offers a critical perspective on tech, its worldview, and wider society with the goal of inspiring people to demand better tech and a better world. Support the show on Patreon.The podcast is made in partnership with The Nation. Production is by Eric Wickham.Also mentioned in this episode:Read JS Tan’s “Oil Is the New Data” piece in Logic Magazine.Support the show

Perhaps the thing that I'm sort of most proud of in terms of this article is simply the fact that it became a great way for tech workers, for sort of my coworkers at the company to be able to talk about these issues in a more open way, to be able to share these issues with their coworkers who are maybe in Office or Xbox, who have really no insights into the cloud business, and be able to agitate these workers with this material. Hello and welcome to Tech Won't Save Us, made in partnership with The Nation magazine. I'm your host, Paris Marks. And before we get into this week's episode, this month, April of 2025, is an important milestone for Tech Won't Save Us. All the way back in April of 2020, I decided to start this show. We were in this new pandemic, trying to figure out what it was going to mean for the world. Many people were still in lockdowns. And finally, I knew I was going to be in one place for quite a long time. And I said, you know what, I have wanted to start a podcast for quite a long time. I see all of this boosterism in the tech

industry and particularly in the podcast space, you know, a lot of podcasts that are just giving you positive takes on tech news or, you know, are even just run by venture capitalists who are getting out their perspective. And I said, we need something different to help try to change the conversation on the tech industry to help give people a different understanding of what is actually going on here. And, you know, I think that the show has been successful at that. I think it's been successful at changing how a lot of people see the tech industry and sadly, the importance of that perspective and the importance of not falling at that. I think it's been successful at changing how a lot of people see the tech industry. And sadly, the importance of that perspective and the importance of not falling for the bullshit

is only becoming more important over time as we see the strengthening alliance between the far right and the tech industry in the United States, but beyond as well. Over the past five years, we've dug into so many different topics on the show, many more than I could ever hope to name. And of course, that continues every single week, like with the show that you are about to listen to. And every now and then, we do a special series to dig into important issues that you want to know about. Two years ago, I looked into Elon Musk and his history to help give people a better understanding of who this man really is and how he became so powerful and

influential. Especially given what is happening now, I think it's even more important than ever to understand how Elon Musk became this figure who is right alongside of Donald Trump day in and day out, it seems. And then last year, we did a series on data centers and on artificial intelligence and on why the industry is creating this expectation that we need to use more and more computation and whether we are really served by that. And so as the show celebrates five years, I need your support so I can keep making it so we can keep having these in-depth conversations every single week so we can occasionally do these even deeper dives so that you really understand

what is going on with key issues in the tech industry. And so if you want to support the work that goes into making this show every single week, and that has gone into making it virtually every single week for the past five years, you can go to patreon.com slash tech won't save us and become a supporter to help ensure that you know, I can keep doing this. And, you know, since it is a special month, since it is an anniversary month, we're obviously setting some targets. So if we get 100 new supporters at the $5 tier or above, well, I'm going to have to get to work on a new bonus series that will come out this fall, as we've usually been doing, and that will really dig into this

relationship between the tech industry and the military and the ways they are trying to integrate their own businesses into the Pentagon and the military-industrial complex and the wider consequences of that, especially as this seems to be a key part of what they're trying to do with their alliance with the Trump administration. So if you want me to work on that series, to put it together, make sure to go over to patreon.com slash techwontsaveus where you can become a supporter. And then if we hit 150 new supporters, again at $5 a month or above, I'm going to get to work on the first Tech Won't Save Us zine to celebrate five years of the show and to have yet another medium to try to spread these critical perspectives on the tech industry.

I have to admit, zines are not something that I am an expert on, but I've always wanted to make one. And so I figured, you know what, five years of tech won't save us. This is the time to figure it out. So if you want to get me to take on these projects, and if you want to support the work that goes into making the show, as I said, again, go to patreon.com slash tech won't save us become a supporter at $5 a month or above and help me hit these targets so we can make all this a reality. Now, with that said, there is still an episode for you to listen to this week. J.S. Tan is a PhD student at MIT looking at cloud computing in the United States and China. He's a member of Collective Action in Tech and writes a newsletter called Value Added. If you have been paying attention to,

you know, this field and in particular what these cloud companies have been doing for the past while now, you might remember an essay that was published in 2019 called Oil is the New Data in Logic Magazine that came from the perspective of a Microsoft worker who ended up going to Kazakhstan to try to sell an AI system to a major oil and gas company to make their operations more efficient. At the time, that essay was published by an anonymous writer. But recently, JS publicly revealed that he was the person who wrote that essay the whole time and finally put his name in the byline. So since that was public, and since I, you know, was going back over that essay and noticed it, I reached out to him and I said, JS, it's time for you to come on the show. We need to talk about this, not just your experience going to Kazakhstan, but this wider relationship between the tech industry, you know, these cloud computing services and the oil and

gas industry and how this is promoting even more oil and gas extraction at a time when we need to be vastly reducing it. So this is an incredibly important episode dealing with a topic that has resonance for all of us. So I hope you enjoy this episode. If you do want to help us meet our goals for the five-year anniversary of Tech Won't Save Us, I'll say it one final time, you can go to patreon.com slash tech won't save us to support the show. And I do need to read out some names because I've realized the names of supporters has gotten immensely long recently. And so I'll try to work through those through this month. So you can join supporters like Todd Zimmerman of Hard Left News,

Brianne from Toronto, Megan in London, Will from Orlando, Florida, Tyler from Detroit, David from Munich, Brendan from Brooklyn, Simona from London, John from Ossining, New York, Sebastian from Belgium, Andrew from Seattle, Aaron from Brooklyn, and Negan, who is sailing around the world, which sounds fantastic. And with that said, enjoy this week's conversation. JS, welcome back to Tech Won't Save Us. Thank you. It's great to be here. Absolutely. You know, it's always great to chat with you, to be able to pick your brain about,

you know, all these topics. Usually we're talking about tech workers and tech labor organizing, and you've been doing some really interesting writing on that recently, To be able to pick your brain about, you know, all these topics. Usually we're talking about tech workers and tech labor organizing. And you've been doing some really interesting writing on that recently, you know, on your newsletter with regard to like, you know, what we're seeing in China and with deep seek workers and stuff like that too. But that's not actually the topic of our conversation today. Because I guess about a month, month and a half ago, I was looking over this older piece in Logic Magazine from 2019, I believe it was,

that I feel like a lot of people have read that is a really fascinating piece, kind of following this Microsoft worker going over to Kazakhstan, you know, as this oil company over there is implementing image recognition, cloud services, all these sorts of things into the running of this facility. You know, and I feel like this piece kind of really made an impact when it came out, a lot of people read it, A lot of people were like, wow, this is the relationship between these major tech companies and these major fossil fuel companies. But the person who wrote it was anonymous. And so when I went back to review it a couple months ago, I was like, wait, there's a name on this piece now. And of course, that name was yours. So I'm kind of curious,

how did you come to write this piece? And of course, why did you decide to reveal that you're the one who wrote it quite recently? Yeah, for sure. So the timing of this piece was really just a crazy thing. When I was in Kazakhstan, on behalf of Microsoft, helping with their projects, right at that time, the Amazon open letter had just come out. It was an open letter to essentially sort of have a set of new demands for Amazon in terms of climate justice. And that was kind of part of my inspiration to sort of, you know, sort of see seeing all the tech workers across the industry standing up and willing to sort of say what's happening, protest against the companies. I think that was part of what got me interested in at that point, thinking about writing something.

And definitely when I was when I was back in the US, sort of having taken a bunch of notes of all the things that I saw there, that sort of got me thinking, okay, maybe it's time to put a piece together around this. As to why I decided to put my name out there finally, I honestly don't really have a great answer for you. I think over the past year or so, people had sort of been in some networks sharing this a little bit. And I figured I'm no longer at the company. It's not like they can fire me. So I reached out to the editors and asked them to put my name up. That's awesome. Yeah.

And so I guess at the time that you wrote it, you would have still been at Microsoft. Is that right? Yeah. When I wrote it, I was still at Microsoft. Cool. And do you want to talk a bit about like what you actually did at Microsoft to the degree that you can? You know, because obviously in this piece, you describe going over to Kazakhstan and, you know, kind of talking about the implementation of these systems. And we'll, you know, we'll get into those actual experiences. But what was it

that you were actually working on when you were at Microsoft? And, you know, was there anything that really stood out to you from your time there that you found really interesting about what was going on internally at the company? Yeah, so I mean, I was at Microsoft right out of college for a little over five years. I spent my whole time there essentially working within Azure. So initially I was in the compute infrastructure team focusing on specifically this segment called high-performance computing. And, you know, the task there was really about figuring out how to bring these massively parallel jobs into Microsoft's cloud, be able to leverage the kind of flexibility that the cloud provides. And this, for me, exposed me to really a bunch of interesting scenarios that,

you know, as someone just coming out of college, I'd really never quite encountered. And so that included things like, you know, learning about how pharma was sort of using high-performance computing in various ways. The banking industry was huge. Being able to sort of do backtesting on their various portfolios, protein folding simulations. One of the really cool scenarios was using fluid dynamics to model car design, that kind of stuff.

And so I was in that team for about half my time at Microsoft. Oh, and just one other thing. That was also the team that actually brought NVIDIA GPUs to Microsoft Azure. We went to conferences like SIGGRAPH, which is sort of the major graphics conference. And this was back in like 2016, 2017, when Jensen Huang wasn't really as big of a name. For me, he was, he was a big name because I, you know, I game and NVIDIA GPUs were always sort of big in my mind. But yeah, I remember going out to a conference with this team that I was on and sort of seeing Jensen from a distance. Anyway, and then the sort of second part of my time at Microsoft, I was still in Microsoft Cloud, but moved from the compute infrastructure team, sort of as low as you can get on the compute stack, to the very top where I joined an AI-focused team.

And so in that team, we were essentially really figuring out how to build AI scenarios to connect Azure, Microsoft Cloud, to the various different industries that Microsoft was pursuing at the time. So for instance, within my team, there was a subgroup that focused mainly on manufacturing. There's one that focused on banking, one that focused on recommendation systems for retailers, and of course, one that focuses focused on oil and gas. Of course, right. You know, a huge market and many huge markets to want to try to move into and entice these

people onto the cloud. And, you know, especially as Microsoft was trying to compete with Amazon in particular to ensure that it was getting, you know, that business, right? I found it really interesting when you said that you had these pharma companies and these other big companies that were kind of doing work on the cloud. When you're looking at major companies like that, is it kind of like they still have their own data centers where they are doing some things and then they're doing some work on kind of the public cloud, as we would call it,

of a Microsoft or an Amazon or a Google. How does that actually work? Like, how is the division between their own computation and the computation that they're kind of renting or whatnot from these major cloud providers? Yeah, I mean, it really runs the full spectrum. So you might have, you know, certainly every, well, I can't say with absolute certainty, but my feeling is that pretty much every Fortune 500 company has some mix at this point of using the public cloud as well as having stuff on-prem, especially on that first team that I was on, the high-performance computing team, where we're trying to sort of really cater to these compute-intensive scenarios. It lended itself particularly well to the cloud because it was kind of a way for these companies to

sort of burst their workloads into the cloud. So for example, let's say you're an animation studio. And as a studio, you may not have that many computers sitting in your office. Maybe once a year, you have to render out a feature-length animation. So that might be a time you want to actually utilize the cloud and sort of offload that rendering onto the public cloud. And just to be clear, of course, when you're saying public cloud and when we're saying public cloud, we don't mean a cloud that is owned by the government. We mean the clouds that are owned and controlled by Amazon, Microsoft, Google,

these large companies, right? Yep, absolutely. So you also talked about this other piece of it where these massive cloud companies are targeting particular industries and then building particular solutions, cloud solutions, AI solutions that are working specifically to try to entice them onto the cloud, but to say like, look,

we can make your businesses more efficient, all this kind of stuff. Can you talk a bit about what that aspect of it looks like? And, you know, the degree to which they're actually like working with these different industries to develop these particular cloud and AI solutions to try to entice them on to say, Microsoft, Azure, or AWS, or something like that? That's a great question. And the thing is, it's quite a large question to answer, just because there's so many touch points between, you know, let's say Microsoft's cloud and the oil and gas sector, right? One touch point is sort of a direct kind of relationship between, you know,

let's say Microsoft and Chevron. They would sort of have a large enterprise contract and sort of the salespeople on Microsoft's side would help a company like Chevron figure out how to use the various Azure services. Obviously, this collaboration can sort of deepen, as with the case of the whole story that I wrote for Logic Magazine, where we have a team of Microsoft employees going out to a Chevron office. But the kind of knowledge sharing also can vary in many other ways. So for example, Microsoft, Google, pretty sure Amazon as well, have all hired folks from the oil and gas sector into their cloud divisions

to really help develop more custom, more sort of bespoke offerings for the oil and gas sector. And so, you know, like you said, that could be sort of in terms of AI, machine learning sort of software that could sort of also be figuring out how to get existing software to work well on these cloud platforms. Another touch point is sort of bringing in open source as well. You know, you may sort of have various researchers or various

companies who are sort of providing IT services to oil and gas companies. They may spin up an open source project where they're doing something like machine learning for 3D seismic survey analysis. This is kind of a place where a company like Microsoft might chip into those kinds of projects, or at least figure out how those kinds of projects can exist on Azure. That's fascinating to hear how that works. And I think what I'd like to do is, you know, go into your experience in Kazakhstan and the story that you wrote. And I think that will allow us to kind of flesh out some of those more concrete details about how that actually

works, if that makes sense. So, you know, how did you end up going to Atirao? Would that be the right pronunciation? I'm no expert on how to pronounce that, but I think so. Yeah. Okay, cool. Yeah. You know, the city in Kazakhstan where there is a major kind of Chevron, you know, oil extracting project and whatnot. Yeah. So how did you end up going over there? And how did this kind of partnership between Chevron and Microsoft actually, you know, start? How did this happen? I'm not quite privy to sort of the origins of this partnership. If I remember sort of in the multi-millions in terms of the size of the enterprise contract, at that level, you typically have sort of pretty long relationships. You probably have an executive

at Microsoft who's sort of quote-unquote sponsoring this relationship. But when I was an engineer at Microsoft, I was at the time already thinking about some of the contradictions that were, you know, within the company in terms of the fact that Microsoft was sort of trying to portray themselves as this corporate leader in renewable energy and sort of fighting climate change. What brings to mind and how early 2020, right, they had this big, you know, your piece came out in 2019, this big push around how they were going to do this carbon moonshot to like be carbon negative and all this kind of stuff. You know, and then of course, we've seen their emissions go nuts with the generative AI stuff, and we can get into that. But that kind of contrast between, oh, yeah, you know, we care about the environment, you know, we're taking this seriously. Even the image of like Silicon Valley and the tech companies as like, you know, these are these clean environments, they're not like the dirty industrial kind of part of the economy or whatnot, you know, so we

care about the environment, things are cleaner, we do things more sustainably and efficiently. And it's like, then you look at these like kind of corporate relationships, these relationships and trying to get these cloud services into like Chevron and, you know, these major oil and gas extraction projects and these other parts of the economy. And it's like, what are you talking about? Yeah. Yeah. And, you know, even when I was there, you know, I remember the company leadership sort of had this go-to line whenever they were questioned about why they had all these fossil fuel company contracts. And the framing was always that, oh, you know, as Microsoft, we're helping them transition to renewables. But, you know, obviously that was a ton of bullshit, as I knew at that time when I was in the team and sort of seeing how these collaborations played out. Yeah, no, it's so wild. So how did you end up going over to Kazakhstan?

What was that about? And, you know, why were Microsoft workers actually going to an oil project like this? What was their role? Yeah, so I mean, this whole trip, if I remember correctly, was sort of an initiative from the sales side of Microsoft. I mean, this kind of a setup is not uncommon for sort of these large enterprise contracts where you have a sales team, they want to sort of have someone directly within Microsoft's R&D side to go with them to sort of make the sales pitch. And so whether that's just sort of bolstering their credibility or being able to run sort of more technical demos, it's a very common pattern for companies like Microsoft. And so when the opportunity came up, I kind of jumped at it.

So the sales team kind of approached my team asking if there was an engineer who wanted to sort of engineer with sort of computer vision type specialties or expertise. I had a little bit of that. I wasn't working on any scenario that they had in mind, but I wanted to sort of have that kind of front row seat and seeing what's happening with this collaboration. Yeah, fair enough. And, you know, now it allows us to learn more about what was going on, you know, in Microsoft at that moment. But so when you went over to Kazakhstan, like, what did you actually see there? You know, what was the pitch that Microsoft was making to the Chevron project for how these technologies were going to change or improve, you know, the way that a facility like this was actually run?

Yeah. I mean, so I think there's sort of a couple of large scenarios. One of the big scenarios, and this is a scenario that I wasn't working on previously, another sort of sister team within my larger org was. So this was the problem of figuring out where in the earth to actually do the drilling. And so this is a phase of the oil exploration, or that's called exploration. And the idea is really just figuring out where to drill and sort of establishing the right drilling strategy. So if you think about it, oil isn't just kind of like sitting anywhere on the surface of the earth, right? It's buried deep on the ground. It's in sort of various pockets or reservoirs that are hard to detect. And naturally, oil companies will want to sort of be able to drill exactly where these reservoirs are to be able to optimize and save costs. And in many ways, the stakes are quite high for them, right? Drilling isn't cheap. You've got to transport a ton of heavy equipment, massive rigs. You have to carry out sort of the

entire drilling process in extreme precision. So if a company guesses wrong, right, if they drill in the wrong place or they hit a reservoir that turns out to be smaller than they expected, it can cost them millions of dollars. And that's why figuring out where to drill is extremely, extremely critical. The way that companies have traditionally dealt with this problem is to essentially locate oil and gas reservoirs through this technique called seismic surveys. Basically, they send out sound waves down into the earth. Those sound waves bounce back. Different layers of the earth will sort of reflect those sound waves differently. And so by analyzing the timing and the shape of those reflections,

you can sort of get a rough picture of what's happening under the surface. But the thing is, the amount of data that we're talking about is massive. The surveys cover huge areas, often the scale of tens or in some cases, hundreds of kilometers. And they collect data and you end up collecting data at a very, very high resolution. And so one of these surveys alone can be sort of at the petabyte scale. And so the trick is figuring out or interpreting this data. And that's where it gets really challenging. Oftentimes it takes teams of geoscientists using high powered supercomputers, you know, all this expertise essentially to figure out where the promising spots to drill might be. And even after all this seismic, you know, even after using supercomputers to parse through this data, my understanding is

that it can take in the scale of months to make any real determination of where to drill. And so, you know, that's really where one of the big AI or machine learning scenarios came in for us. The idea is to be able to use machine learning techniques to help these geoscientists parse this massive data set so that they can go from months of interpretation to days and really help these oil companies figure out where to drill faster. And so the idea then is like, you're not only speeding up the process, so you're not spending this much time kind of deliberating, but also I guess there's an assumption at least that you're probably going to have a more accurate reading on the other end in the sense that you'll have a better idea of where the most profitable kind of well is for you to try to to drill at and start extracting yeah exactly

yeah interesting obviously if you're someone who's concerned about the climate uh that is going to stand out as uh you know a real concern if if you have this company that's promoting itself as caring about the environment and renewable energy and all this kind of stuff. Meanwhile, they're helping these oil companies not only more quickly find wells to exploit, but to do so more cheaply, more efficiently. That's not exactly, I think, what we want to be seeing right now. Yeah. So that's one piece of it, right? The identification, the exploration. What about when it comes to kind of existing wells that are already operating, like, you know, this one that you visited in Kazakhstan, how would, you know, a project that is already kind of up and running benefit from these kind of tools that Microsoft

is putting together and trying to get them to adopt? So that essentially brings us to sort of the production phase of the drilling process. And their efficiency is really key. So there are little things that these companies want to do, such as making sure that they're drilling in the most optimal way so that the erosion of the equipment itself is being optimized. So in this phase, the cost of machine failure was extremely high. If one machine goes down, it could sort of have this kind of butterfly effect where various processes on the oil field have to essentially stop. And so a really important thing for these companies to be able to do is make sure that their machinery essentially operating for as long, or the other way to say is to have as much uptime as possible for their equipment. And so to do that, you want to make sure that you can sort of predict when a

failure might occur. I think that does give us a good idea of the kinds of implementations that we're looking at here, right? And this is the kind of thing that we've heard about with these types of projects in the past, right? The notion that, you know, these cloud companies are selling the major oil companies on this notion that, you know, you're going to be able to do things more efficiently. That means you're going to be able to extract more oil and you're ultimately going to be able to have a more profitable operation because yes, you're paying for these cloud services. You're paying for these machine learning services as you're saying, but you know, on the net, you're going to be able to make more money or have greater savings than the actual cost of using these tools,

right? I'm assuming that's the kind of, you know, pitch that they're making. Yeah. I mean, the thing is that the oil and gas industry has sort of been like a longtime user of high-performance computing or sort of just, you know, they've been a long user of compute-intensive things, you know, and they've used supercomputers to varying degrees to do things like, you know, certainly handle their massive data sets and obviously analyzing that. So being able to run complex 3D simulations, create detailed geological models. And these are things that go far beyond the typical capabilities of your standard computer. And so for this sector, I think this

is part of the reason why the cloud is such a valuable type of service for these companies. Oil companies like Chevron are the ideal customers in some ways, because at least in terms of their compute demand, you can see these huge fluctuations. So it might be that when the geoscientists come in, they need to run an experiment, they need to run, you know, some kind of analysis. And when they decide to do so, the sort of service that they have on-prem may not be enough to do that workload in a timely way. And so for them, it's extremely beneficial to be able to burst that workload out to the cloud. Yeah, absolutely. And just to be clear, when you say on-prem, you mean like the actual computational infrastructure that they have themselves. And then of course, you know,

they're taking advantage of this additional stuff from the Microsofts, the Amazons, you know, depending on the oil and gas company and who they have those relationships with. One of the things that really stood out to me as you were describing that is this notion of like, you know, how we see these companies, right? As we were talking about with Microsoft not too long ago, but we think of like the oil and gas company, the Chevron, and we imagine these big refineries and oil extraction and like, you know, it's very dirty and very industrial. And, you know, we have this particular vision and we don't think about like, as you're saying, you know, the aspect of this business, that is like high performance computing, 3d models, where you basically have the people in, you know, the offices, the kind of techie offices that are that are doing this kind of work.

And then on the opposite side of this, we have like, you know, the Microsoft's and the Google's and the Amazon's. And we imagine those, you know, kind of sterile office spaces, you know, maybe they have some ping pong tables and stuff like that, as we hear about with these tech companies, right? But, you know, we don't think about, you know, the wider supply chain and those other parts of what goes into this industry. And I don't know, it's just interesting to me, like how there's a lot of overlap here, right? But the industries themselves can be framed in very different ways.

And so when we think of them, this particular imaginary can be very beneficial to the tech industry with how, you know, we kind of publicly or how like the public often thinks about how these companies actually work, right? Yeah, I mean, I think what's the phrase you hit the nail on on the head, right? Like, you know, the way I sort of imagine most people sort of thinking about big tech and big oil is that it's kind of strange to sort of put them in the same sentence even, right? Culturally, they are so different. You know, tech companies, as we're talking about before, love to sort of brand themselves

as sort of these champions of sustainability, racing to outdo each other even. But the twist is that, as we're saying, you kind of have to come back to the business model of the cloud. And that's where you kind of have to come back to the business model of the cloud. And that's where you kind of see big tech and big oil working hand in hand and actually that relationship even getting stronger over time. And I think it's worth sort of dwelling on the point of the cloud business model for a second, because, you know, I don't know if this perception still exists today. But certainly when I joined the industry, you know, I had this idea that, OK, the cloud is sort of this disruptive technology where it's essentially startups who are the primary customers. The big sort of moment for Amazon was when Netflix decided to go all in on using AWS as their primary infrastructure. And so there was this sort of perception that the cloud is this kind of disruptive forward looking thing, as opposed to this thing that is all about supporting legacy

companies, right? And that's really where big oil comes in, right? If you're a cloud provider, if you're Amazon, Google, Microsoft, you would much rather win over the Fortune 500 companies than to win over the hundreds or thousands of various startups across the country, because that's essentially where the money is. That's where the really major IT spending is. And among the Fortune 500s, some of the biggest ones are certainly in the oil sector, right? I think I wrote in the logic piece that six of the top 10 companies in the world by revenue were oil firms, at least at the time of writing. Yeah, totally. And, you know, if they're not any longer, it's just because they've been displaced by the major tech companies for the most part, right? So there's a clear desire to want to work together on both their parts when they're such,

you know, large companies, right? Especially if they see that there can be a kind of mutually beneficial relationship being formed there. I wanted to pick up on what you were saying there about the business model, right? And then we can come back to some other aspects of, you know, what you saw in Kazakhstan and kind of working on this more broadly. But I think that this is really interesting, because there has obviously been a lot of discussion about data centers and the cloud over the past couple of years, in particular, as people have seen this boom in generative AI and, you know, companies like OpenAI and Microsoft, promoting this as like the future of how we're all going to use digital technology and all this kind of stuff. And, you know, that has been associated with this massive build out of

new hyperscale data centers and a lot of investment going into, you know, the expansion of cloud infrastructure. But like you were talking about there, you know, there's a whole aspect of this that is about supporting traditional industry, you know, that is about bringing that over to the cloud. I wonder how you see the actual breakdown of, you know, who uses the cloud and what all of this computation within, you know, Microsoft's cloud, Amazon's cloud is actually used for, you know, because it feels like the perception that a lot of people might have, especially recently, is that, you know, it's generative AI driving all of this usage of this cloud infrastructure. But actually, there's so much more that these companies are actually doing with it. So I wonder how you kind of see the distinction or, you know, what the actual

primary use cases of the cloud actually are and what's driving not just the current use, but also the expansion that we're seeing. So I think maybe the first thing I'll say is that based on sort of the media reporting, it seemed like a lot of these companies sort of had like this light bulb moment where they were like, okay, we need to start investing in data centers and start building out this infrastructure. The reality is that these companies have been doing that for a much longer time, well before ChatGPT, right? When I was working at Microsoft, I think that number, at least for Microsoft alone, was already in the tens of billions per year.

And so, you know, while it's certainly accelerated since ChatGPT, you know, we sort of just want to remember that there's sort of been this longer history of investment in cloud already. In terms of what the cloud is being used for at this point, you know, I've been out of Microsoft for the past five years, so can't say based on my own experience, but you know, a lot of these high-performance computing scenarios continue to exist to this day, right? A lot of

the people that they've hired, a lot of the technologies that they've built, a lot of the software solutions that they've worked on, and certainly a lot of the partnerships that they've developed over the past 10 years around scenarios like the analysis of seismic surveys continue to this day and continue to drive a lot of the consumption of the cloud. When it comes to whether or not generative AI is being used by these companies, my guess, and I'm not certain of this, but my guess is that they're using it for some of the very basic scenarios that you might expect. For example, being able to fine tune a model on HR documents so that employees can search for things more easily. Scenarios like that, whether or not generative AI has fundamentally changed things like oil exploration or oil production, that I would be

a bit more skeptical of. So then the question is, well, you know, if it's not fundamentally changing the business models of these oil companies, why the investment in this infrastructure? And I think that in itself is, I mean, something that you certainly commented on in the past. But one thing I'll add is that, you know, building all these data centers and buying all these chips is in a way not a bad investment for companies that are already so cash rich. These capital assets, their value is not going to evaporate if AI goes bust or if the bubble pops. So it's sort of a somewhat safe asset to some extent. Does that mean it's not an overinvestment? I mean, I personally am on the skeptical side. Yeah, you certainly get vibes of the 90s and the massive fiber build out with some

of what we're seeing with all the money flooding into data centers right now, which is also not to say that it won't become useful, you know, in the future, even if it is being overbuilt at the moment, right, which is something that we frequently see there. So I wanted to go back to what you were saying about, you know, your experiences in Kazakhstan, and kind of, you know, the more concrete ways that these technologies were being deployed. Because one of the things that I think really stood out to me, and that I feel like really stood out to you as you were, you know, having this experience and having these conversations there, because so much of your work has, you know, focused on workers in particular, is how, yeah, you know, they were interested in how, you know, these cloud tools

and these machine learning tools could be deployed around oil and gas extraction and production and exploration and all these sorts of things. But they also had this workforce of thousands of workers there who they wanted to make sure they could get a better handle on. So what did you see there in terms of the desire to want to use these, you know, these technologies to increase control over the workforce? Yeah, I mean, I think that's an aspect of this article that I wrote that certainly got a lot of people reading it to some extent. My experience with this was quite a strange one because for most of this trip, we were sort of in this big conference hall type situation where it was 40 plus people and we were sort

of presenting these scenarios to a variety of people in the room, you know, whether that's sort of American execs, you know, who work at Chevron to sort of data scientists and also folks who are local to Kazakhstan. However, there was this one meeting that I ended up joining and this was sort of more of a closed door meeting. This was a meeting where it was,

my impression was primarily the executive level, although it's hard to keep track of all the various hierarchies with these different companies. And what was striking was that there were no Kazakh people in that room at that meeting. It was kind of the executive level who were basically Americans. I believe there's one Russian person as well, and us from Microsoft. And it was in that meeting where I started to hear some of the desire to sort of use machine learning and AI technology for essentially labor disciplining purposes, right? So, you know, one of the examples that I wrote about was being able to use machine learning technology to figure out if workers had sort

of been stealing equipment based on their GPS location. I'm not sure if there was really precedent for this. They certainly didn't talk about any precedent for this, but, you know, the same technology could just as well be used essentially for, you know, making sure that these workers are sewing up on time, are sort of getting their, you know, whatever KPIs they're assigned to them. And so the way I saw some of that was very much about disciplining, disciplining labor. You know, another example that sort of comes to mind was sort of something that was said a little bit more offhand, right? It wasn't sort of one of the bullet points in the meeting that we're having, but, you know, one of the Chevron folks were saying that, oh, is there something that we can do sort of in terms of data analytics and data processing

where we can essentially flag things that the cosmic workers or other employees in the organization, you know, flag things in their email system, basically, that they thought were suspicious or whatnot, which to me basically read like a way to surveil their emails. It's incredibly concerning to hear those things, right? Because I feel like often, you know, especially in the work that we do, we talk a lot about how technology has been deployed over, you know, the past decade, but, you know, much longer than that as well, to increase that surveillance over workers, and as a result, to decrease the kind of power and leverage that workers actually have. And, you know, often that focuses on like, you know, what Uber is doing, what Amazon is doing, and these sorts of companies.

And then, you know, to read this piece, and, you know, to hear about the meetings that you had, and the discussions that you had, where, you know, they're really not looking at this, you know, just as like, you know, Amazon in its warehouses, or some new tech company or whatnot, you know, this oil and gas company that operates in a country like Kazakhstan, and is looking to deploy, you know, not dissimilar technologies, and potentially even in a more intrusive way, when we're talking about scanning emails and things like that, in order to expand the power and surveillance that they have over the workers there. And it's like, it very much shows directly, I think, which is maybe one of the reasons that it stood out to

people that this is not just something coming for some new tech companies, but is going to be rolled out. And you can see that there's this desire to roll it out in so many other different types of companies where they can get away with it. Yeah. And, you know, I'll add that in the case of Kazakhstan, in the case of Fatura, this particular city that neighbors the oil field, there was also this kind of really interesting dimension in terms of, you know, it's Chevron, one of America's

biggest companies, you know, setting up an office out in Kazakhstan where they're able to sort of take advantage of the fact that, you know, labor is going to be cheaper there. They're going to be able to take advantage of possibly weaker labor laws, possibly weaker environmental protection regulations. And it was quite visible in sort of just being in that space, right? Etorah, the city itself, was not, by any means, did not look like a developed city. There were sort of these kind of moments in the city where there were development, but those were sort of the bungalows for the American expats who lived there or, you know, sort of the hotel and some of the neighboring amenities around the hotel where clients or other Chevron related people like myself at that time were coming to visit.

And the spillover was, from what I could tell, extremely limited. That sort of economic spillover onto that city was extremely limited, you know, which sort of goes to show that that was the kind of relationship that an American company like Chevron was setting up in a place like Atturo. Yeah, no, that's a really good point. And I feel like it's something that we see time and again, right, with so many of these resource projects, especially located in parts of the global south, but even just in rural areas and stuff like that in the global north as well, right? You can very clearly see how the benefits are not redounding to the people who are nearby, right? So, you know, we've talked a lot about how Microsoft was trying to sell these technologies,

how a company like Chevron was looking at potentially rolling them out. You know, and obviously when a lot of readers read your piece and learned about this, there was quite a reaction to it, right? I feel like this is one of the pieces that people who are engaged in this space often really remember, right? Because there is so much concrete detail in it as to, you know, the types of meetings taking place, the kind of things that, you know, Chevron executives and Microsoft workers were talking

about. I wonder what you made of the response to the piece when it came out and, you know, how people were talking about it and kind of the impact that it made. Yeah. I mean, I think maybe the way I'll answer this is to start with, you know, whether or not it changed Microsoft's trajectory. So, I mean, maybe it's worth talking a little bit about Microsoft's commitment to sort of green initiatives as a whole. So when I was at Microsoft, you know, one of the things that I remember was Brad Smith,

president of Microsoft, chief legal counsel, maybe his title has changed, sort of making this big public pledge to sort of green the company's cloud business. And the idea was to power more of our data centers, more Microsoft data centers with renewable energy. I think one of the first major goals was to hit 50% renewable energy by the end of 2018. And to be fair, Microsoft was able to hit that target. And so I think the following year, they essentially raised the bar to some higher target some later year. Then in 2020, they made an even bolder announcement, which is that they would go carbon negative by 2030.

And so this meant not just reducing emissions, but actually removing carbon from the atmosphere that they've admitted. So sort of across everything from data centers to their supply chain to sort of the offices themselves. And as part of that pledge, they had said by 2050, they were going to remove all of the carbon the company had ever emitted since it was founded. So this pledge came out, I believe, less than a year after my article was published. And that was, you know, I'm not going to say that that was because the article, that was because a lot of things that was happening in the industry at the time. So that included the fact that Amazon was also sort of making comparable pledges, partly moved by the fact that there was a sustainability working group at really useful way to sort of, one,

educate other co-workers in the company about what was happening at the company. You can imagine a large company like Microsoft, 150,000 plus employees, most of which do not work within the cloud division. Getting them to understand what was happening in Microsoft's cloud was part of the struggle. And that eventually was able to sort of contribute to this working group to help push for more demands. And one of the things that came out of that was that there was a commitment by Microsoft. Part of what the sustainability interest group inside the company did was to continue to put pressure on Microsoft's leadership to rethink these partnerships. And part of how they

did that was to use my article as a way to make the point that these collaborations were actually resulting in worsening outcomes. Eventually in 2023, Microsoft put out a new set of principles to guide how it will work with fossil fuel companies moving forward. And sort of the key point there with that sort of renewed commitment was that they would only support extraction projects for companies that have made public a net zero commitment. So that sounds pretty good, right? That sounds very close to some of the original demands that sort of people like me have been pushing for as early as 2018, 2019. But so Karen Howe wrote this great piece for The Atlantic on sort of the same topic. She notes that that kind of promise actually doesn't mean much in practice. Most of these so-called sort of net zero pledges from oil companies only count emissions from

their direct operations. So that's like whether their offices are running on green energy, whether their cars are running on green energy. Yeah, whether Chevron owns some EVs or something. Exactly, right? And it completely ignores the much bigger picture of the fact that they're producing so much emissions. So, you know, in terms of the impact of this article, I would say it's quite a challenging thing to assess in terms of how the company responded. But I think sort of these pledges and commitments aside, I

think perhaps the thing that I'm sort of most proud of in terms of this article is simply the fact that it became a great way for tech workers, for sort of my co-workers at the company to be able to talk about these issues in a more open way, to be able to share these issues with their co-workers who are maybe in Office or Xbox, who have really no insights into the cloud business, and be able to agitate these workers with this material in hand. I love that. Yeah. And I think that's so important, right?

And obviously there's the one part with how it informed these workers within Microsoft and, you know, I'm sure to a certain degree within Amazon and Google and, you know, these other cloud companies as well, but also the broader public as to what is actually going on here in these relationships that exist so that they can be aware of it and they can try to push back against it, right? Because if there aren't stories like that, if there aren't people explaining what is actually happening there, it can be really difficult to know. And so I guess, you know, as we start to round off our conversation, I was wondering, you know, you talked about kind of being inside, having these conversations with co-workers, how those conversations potentially progressed

afterward. How did workers within Microsoft and within Microsoft's cloud division feel about working with these oil and gas companies? I imagine there were some people maybe who didn't mind and other people who had issues with it, but what did you see there? And, you know, how do you feel that that, you know, tech climate activism has progressed in the past few years, especially as we see these companies, you know, increasingly cozying up to the Trump administration and these kind of climate denialist talking points. Yeah. I mean, I would say at a company like Microsoft, even before going out to Kazakhstan and writing this article, within Microsoft, I was involved with a bunch of organizing efforts over various issues. So climate was certainly one of them. I helped organize a bunch of my

coworkers to sort of join the climate walkout. I think that was in 2019. And that was sort of a cross sector. I think it was actually like beyond cross sector, but it was a huge walkout. And so that was one of the things we're organizing around. There are other things including, you know, military contracts, collaboration with like the HoloLens goggles and the Department of Defense to help sprained soldiers or something like that. And so I'd sort of been involved in a lot of these efforts to sort of push back against these problematic contracts. And I would say sort of across the board, it was quite a challenge to push particularly employees at Microsoft, perhaps in a way that maybe differs a little bit from, say, a company

like Google, partly because, you know, Microsoft's average employee is a little bit older, especially within the cloud division. There's sort of been a lot of migration from Microsoft's enterprise services into Microsoft's cloud. And oftentimes they overlap, right? And so some of these employees had sort of been privy to the fact that there had been sort of these big contracts with banks, with oil and gas companies, with da-da-da. And so it was hard to sort of get buy-in across the board. That said, there were definitely a lot of passionate people at Microsoft who saw something like this article and was infuriated. And so after I ended up leaving, there was that sustainability

working group that continued to push and push leadership and bring in more people into their community to help make sure that the voice of workers was being heard. That's so great to hear some positive aspects of it. You know, even though it's not surprising to hear that there will also be workers who are not as interested in these sorts of things, too. Right. And I guess picking up on what I was asking before and recognizing that was a really big question, we're at a moment where it felt like there was this climate activism happening at the tech companies that in some cases was actually, you know, making a difference, right? Getting the companies to announce more ambitious pledges. But it also feels like in the past few years, we've seen a number of climate activists be fired for kind of pushing for the companies to do better. And now we see this alignment between,

you know, a lot of these major tech companies and the Trump administration and, you know, the particular politics around climate change that have emerged from that. And so I wonder, you know, what you think of the state of, you know, climate activism within tech today, and if there are still some bright spots, as far as you see it. To answer this question, I think we have to zoom out a little bit and sort of see the tech worker movement as a whole. So sort of in Trump one, in the first administration of Trump, a lot of the activism was sort of centered around specific issues, whether that's climate, whether that's military contracts, whether that's sexual harassment, LGBTQ plus rights. And from the perspective of the media, it was very apparent that these various issues

were taking off, that employees were protesting at a scale that was essentially unprecedented for the tech industry in the US. Over time, that has sort of transitioned from sort of these issue campaigns where employees like myself would sort of write these open letters, send it out to the New York Times and sort of have this kind of shame and blame approach to get companies to change their behaviors. And the strategy shifted essentially to sort of organizing in a more covert way, in a more underground way, partly due to the fact that the initial strategy was met with a lot of retaliation from employers. And so over time,

that did shift towards these new strategies of labor organizing, where, you know, the issues shifted in part to sort of have that wider base of support from employees. And the goal was to sort of work towards a legally recognized union. And so for me, the big shift was not so much in the past several months with the new Trump administration is the fact that in sort of 2022, 2023, there was a series of layoffs, you know, across the tech industry. And, you know, that kind of completely changed the labor relations in the tech industry, such that, you know, where workers used to sort of have a lot more labor market power, they now didn't. And, you know, as a result, there was a lot more fear in terms of organizing,

in terms of, you know, really a result, there was a lot more fear in terms of organizing, in terms of, you know, really trying to do any activism within these companies. Certainly with the new Trump administration, that has worsened. But to me, the major shift was slightly before that, when there was this kind of fundamental change in the labor relations and tech. And so how that applies to climate organizing, I think climate organizing is one of these issues that sort of did go from being out in the open, being sort of what tech workers were writing open letters about, participating in walkouts about, to being part of a more underground, a more covert approach to build worker power within these companies. And then unfortunately, I think that's sort of deteriorating over the past couple of years as labor relations within the tech sector has shifted.

Yeah, it's a disappointing trajectory, but I think that puts it in so much better context, you know, to understand what happened a couple of years ago and how that feeds into what is going on now. JS, I feel like I always learn so much when I talk to you about these things. You know, usually it's about tech organizing more generally, but I think it's great that you wrote this piece, you know, over five years ago now. I think it really, you know, helped to catalyze some important conversations around the relationship between the tech industry and the oil industry. And I'm really happy that, you know, even five years later, you know, you were able to come on the show. And, you know, I think that this will continue to inform a lot of

people about what is going on and hopefully, you know, continue to foment action to try to change these things, you know, and make sure that we're not continuing to extract oil and gas and, you know, making that even easier to do through technological means. But again, thanks so much for taking the time. Always great to have you on the show. Thanks. It's a pleasure. JSTAN is a PhD student, a member of Collective Action in Tech, and writes the Value Added newsletter. Thanks. It's a pleasure. your arm.com slash tech won't save us and making a pledge of your own. Thanks for listening and make sure to come back next week.