U.S. Government Looks to Invest in Quantum Startups | Tech Field Day News Rundown: October 29, 2025

Open AI and Microsoft get more mature. Amazon is discounting a little bit of their opex IBM applies textures. IBM and A MD are gonna get even more.

Quantum. Qualcomm finally squares up to Nvidia, Palo Alto soars with Prisma Air. And we're gonna take a closer look at the US government sudden interest in quantum computing in this week's episode of the Tech Field Day Rundown.

Hello everyone. Welcome to the Tech Field Day rundown. I hope you have your costumes ready because we are getting so close to Halloween.

Uh, but I hope that you're enjoying the crisp fall air on this national oatmeal day. I, of course, am Tom Hollingsworth, and I'm joined once again by my amazing co-host, Mr. Alistair Cook.

Al, welcome to the show. Always great to be here with you and, uh, enjoying my first cup of coffee of the data right now. Well, I'm glad that you are enjoying some coffee as we take a look at some of these wonderfully fun stories going on in the world of enterprise it.

And as we discussed in the opening, Microsoft and OpenAI are taking their partnership to yet another level with yet another new agreement. Microsoft now owns a $135 billion stake in AI and keeps exclusive rights to its AI models and Azure API all the way through 2032. The deal adds independent.

A GI verification allows for joint product development with third parties and lets Microsoft pursue a GI On its own open AI gains more flexibility, including access to US government customers, the ability to release open models and the freedom to use multiple cloud providers. The company's aim to keep innovating responsibly and create new opportunities for businesses and users. Al Microsoft now owns $135 billion of open ai, and yet they're allowing them to go off and do other things.

Where's the value for Microsoft in this? Well bear in mind that that $135 billion is only 27% of OpenAI, or at least the, uh, public benefit co corporation that they're transitioning to. So remember that OpenAI was not originally commercial and there's all kinds of challenges around that.

So, uh, there's some fun pieces in here. Uh, part of it is around how the original agreement between Open AI and Microsoft ran and that this, uh, artificial general intelligence declaration is a sort of watershed in that agreement. And so now there has apparently been a, a declaration that is, uh, due to be verified for artificial general intelligence from open ai.

Uh, and that changes how things all fit out together, that that really tightly bound partnership becomes a little looser on this. So Microsoft can choose to follow more, uh, uh, AI and general, uh, AI without necessarily being bound to open ai. They probably are still gonna be pretty tightly bound and open.

AI can buy its compute resources from somebody other than Microsoft. Although they've already contracted for $250 billion of Azure services. I have to be incredibly successful to need to spend some money anywhere else.

So that loosening of the all of these things is basically around how the contracts used to be and how the contracts have been renegotiated. Uh, it's just allowing both sides to be a little more free of what they do. Personally, I think we'll see them as tightly bound together as they always have been.

I don't think this is actually gonna make any material difference to how either organization operates. It just gives them the option that they might choose to. We often don't take the options that we might choose to.

We just have them because maybe we might want them. Amazon, who we covered some of their troubles last week, although much overblown troubles in my opinion. Uh, Amazon has a new piece of news for us.

They're laying off approximately 14,000 employees, about 4% of its corporate workforce. As they adopt AI to streamline operations. I may cut an additional 30,000 roles, well, that'd be another 8%, bringing it to a whole 12% of their workforce.

Uh, over time the company says AI will enable faster innovation and efficiency and, uh, reduces the need for certain positions while creating new opportunities elsewhere. That sounds like management speak. Uh, this marks Amazon's largest recent restructuring follows previous lay layoffs across its divisions.

Executives emphasize that AI adoption is transformative for the business, uh, even as it raises concerns about job displacement in the wider tech sector. I don't think you and I are at risk of being replaced with ai, but how much more of Amazon can be replaced with ai. I'd be a little bit worried about what actually has been replaced by AI at Amazon because we hear about how they want to cut all of these jobs because they overhired during the pandemic.

Okay, I'll buy that. Well, no, actually, what we really meant is, is that a lot of those roles that we hired for during the pandemic, maybe a few more people are actually able to be done by ai. Okay, weird.

So I'm assuming that all of your ai, uh, data centers and infrastructure and all that stuff is running at a hundred percent capacity right now and really providing value. What do you mean it's only running at like 18 to 20%? That's so weird.

It's almost like what you have isn't being utilized to its fullest extent, but yet you're cutting people because the AI is doing better than you thought. Well, if you're only running it at 20%, you're not gonna buy any more this year. Are you?

Surely not. Surely you're not gonna go out and use the money that you save from these layoffs to buy more AI data center components to augment things that are already running at what a fifth of their regular capacity. That's one of the problems that we're seeing here, is that all of the signs in this story point to the fact that Amazon is not actually laying people off because AI is doing their job.

They're laying people off so that they can buy more stuff to make AI bigger. You know what this sounds like, right? You know, there was that little kerfuffle in the Netherlands about something, something tulips.

You don't wanna be the last person not holding the bag until it's time when everybody drops the bag and you don't want to be the last person holding it. So why would you not preemptively say, we're not going to play this game. We're gonna use what we've got, and when it finally reaches a point of maturity, then we're going to maybe buy a little bit more, but not a lot more.

Well, that's because the market is basically being propped up right now by people buying like crazy, like they're going, uh, out of business. And, and I think what you're gonna see is that, and, and we know this through every bubble, every hype cycle that we've ever lived through, there are things that AI is good at, but AI is not good at everything. And once you realize that and you stop over spending on ai, you're gonna realize you probably are gonna end up needing to hire a lot of those people back that you laid off because they're the ones that were doing the real grunt work underneath the covers.

And I think it's funny that a company that started off by saying, oops, we might have overspent on hiring in the pandemic now claims that all of this extra AI stuff they're buying is in no way overspending on any of that, except the difference is is that hopefully some of these people can go out and find jobs and do something meaningful in the industry. All that AI gear that you're buying, if this doesn't pan out, I don't know what you can use it for. I dunno, maybe you can sell more books.

IBM consulting has acquired texture. It's a company that helps businesses move to and modernize hybrid cloud systems. Textures.

Tools will speed up cloud projects, reduce manual work, improve recommendations and support greener IT strategies With experience in over 100 global projects, texture strengthens IBM's platform from managing cloud transformation from start to finish, helping clients modernize faster and more confidently. Al do you think that IBM's cloud ambitions would benefit from having a tool like texture? IBM's always had a big consulting practice, and it's a consulting practice built around having processes and templates that can be reused by the IBM branded person in your town.

And so having good templates, good processes around that application migration and modernization and, and, and just general cloud strategy and hopefully probably an AI strategy. And there is vital IBM of course, international business machines, that's an English language thing, and texture is an Austrian company. It's really vital as you're starting to sort of focus on European growth, that you have local presence and local understanding that is quite different from the US Understanding our operating in Europe is very different.

It's, it's a series of much more isolated smaller enclaves of ways of doing things in regulatory and compliance environments. And so I think this acquisition is around getting the knowledge, the checklists, the processes that fit a European business P practice, uh, in addition to the existing deep knowledge of IBM operating in North American, uh, practices. I think it's a a great thing for, uh, IBM to have a broader, uh, platform for their, their, uh, consultants and to have more access to European markets.

I think a lot of US companies underestimate just how different the European market is to the US market, thinking that countries in the US are similar to states, uh, states thinking that countries in Europe are similar to states in the US and, uh, that simply isn't the way things, things work out there. So yeah, good move and helping companies move across to that migration and modernization. I know we've been talking about migration and modernization for a long time and cloud strategies for a long time, but there's still a lot of organizations that really still haven't made enough of the move or haven't made a move in a way that actually gives them business advantage, ends up being beneficial to them.

So, uh, better consultancy, better methodologies around that is always gonna be beneficial. Speaking of IBMI, BM and a MD have made big step in quantum computing by running the error correction algorithms on a MD chips instead of requiring yet more qubits. This allows the fragile quantum calculations to be stabilized using low cost, widely available non-super cool hardware, uh, making practical quantum quantum systems more achievable.

The milestone moves IBM closer to its 2029 goal of fault tolerant quantum computing and shows the potential of combining classical and quantum computers for real world applications. Thomas, this big news or is this just more quantum is getting better but isn't really here yet? I think it's big news for this one particular aspect of it, and I do agree that being able to run the error correction algorithms on what I would consider to be standard computing hardware is a big deal.

For those of you who did not watch my conversations episode about quantum computing, basically what happens is, is that whenever you're trying to measure those qubits to figure out what the data is that locked in there, you have to have enough error correction to screen out the noise that's created because there's a lot of extraneous data that's produced when you're doing this measurement. Normally that takes a little bit more horsepower to do, as you mentioned, A lot of times it's being run on the same computer that's actually doing the work, and that requires a lot of space, a lot of electricity, a lot of liquid nitrogen. Yeah, really liquid nitrogen.

These things have to be cooled within a few degrees of absolute zero in order to be able to run. So on the one hand, you know, you've got cold fusion, which is we want to be able to run fusion at anything less than the heart of a star. Now on the other side, you've got warm quantum computing, which means we wanna be able to run it anywhere north of 273 degrees, science degrees below zero.

And so the the idea is here that if you can start getting more precision in your calculations on cheap hardware, it allows you to take the investments that you're doing to make the quantum computer itself run better. And this is the, where I've had a little bit of a disagreement with some of the news that's come out about quantum computing over the last couple of years is that some companies like Google are racing to produce a computer that has like the massive amount of qubits available to do these calculations, right? Like you, you hear about, you know, we've got a computer that can do hundreds of cubits or thousands of cubits, and every time I read one of those articles, I didn't see any mention of error correction.

And that's for a good reason. Uh, for those of you out there that are audio files, well, what happens if you wanna make the music louder, right? You, you go over to the speakers and you turn the knob all the way up.

But what happens when you turn the knob all the way up, you increase the noise of everything in the system, not just that, uh, beautiful, uh, you know, uh, Johnny Lee H****r Blues album that you're listening to, but all of the stuff in the background that was, uh, captured on the recording as well. That's why we have digital signal processors and all kinds of technologies that allow us to screen out that background noise. We're using it right now on this episode of the rundown to screen out my daughter coming home and slamming doors and walking around in the background.

But in order for that to work on a quantum computer, you have to be able to screen a crap load more data, and that is a quantum unit is crap load. But what you're gonna run into is, is that, that it becomes relatively expensive. So I'm glad that a MD was able to make this work.

I I can't wait to see what more applications come out of this. And if you stay tuned for a closer look, I think you might actually find out that we're not the only ones that are interested in seeing what more can quantum outta quantum computing. Qualcomm is entering the AI data center market with two brand new processors.

The AI 200 and the AI two 50. The first one is gonna launch next year in 2026, and the one is, the other one is gonna launch in 2027. They're designed for AI inference workloads.

The chips support major AI frameworks and will be sold as part of integrated rack systems or individually if you wanna pocket them in your local micro center. And they're looking to target cost conscious enterprises with that. This move diversifies Qualcomm beyond more than just making chips for mobile phones and positions them to challenge NVIDIA's dominance in the AI infrastructure market.

And that is a key market being driven primarily by a generative AI and large language models and the hardware that powers it, most of which comes from nvidia. The launch signals a strategic shift that could reshape Qualcomm's growth in industry standing and standing and understanding. We've been waiting a long time for a company to come by that has a credible threat to NVIDIA's dominance in the market.

Al, can Qualcomm pull it off? Well, that's a billion dollar question for Qualcomm really, isn't it? Uh, two chips turning up next year in the following year.

Yeah, it's, it's kind of feels a little like this is, um, me too. We're a little bit late to the party, but we realize that, uh, that this is gonna be a big party. It's not really, because silicon design takes an awful long time.

It's not like writing software where you can make a change to code and see that code run a few minutes later. When you are working in silicon, things take a lot longer to cycle through and make changes. So it's not that they completely missed the boat and are are starting to run very late.

It's that the engineering to do this takes a long time and maybe they weren't certain when they could complete that engineering. This announcement means they're pretty sure they're gonna be delivering some, uh, valuable turn in next 12 months. Um, I think this is really good.

We do need to see some credible competition for Nvidia. And I think the very wise to target inference, inference is the phase we were actually running an application that delivers some value. A lot of the time we're seeing a lot of focus in the AI industry around what do you do for creating foundation models and what do you do for training?

And it becomes a second thought that we're actually gonna need to get some business value for all this money we are spending and that business value comes from inference. So yeah, seeing Qualcomm focusing on inference and focusing on what I characterize as an, as an engineered system. So a rack scale infrastructure full of inference hardware.

Uh, I think this is something that we will see more of. Uh, we last week at, um, at Cloud Fields Oxide computer showed us that same rack scale infrastructure idea. I think, uh, there's wasn't full of Qualcomm chip, but maybe some of the later ones will be.

Uh, Qualcomm delivering this as, as rat scale infrastructure for running your AI workloads on. Seems like a really good move. Others in the market, uh, a MD has made announcements.

Intel of course has had, uh, their accelerators in the market for a little while as well. So it's not that Qualcomm is the only challenger for Nvidia here, it's just that we haven't seen anybody unseat Nvidia from being that primary accelerator for our workloads. Palo Alto Networks is expanding its AI tools with Prisma as two and Cortex Agent X to support AI applications and automate security tasks.

So AI for security and security for ai, uh, Prisma two secures AI models and agents throughout their lifecycle where Cortex agent lets teams build and manage AI agents to handle threat detection response and policy enforcement automatically. These updates help organizations stay ahead of moving cybersecurity challenges as AI uses rapidly increased both by the companies that are targets as well as the bad actors Who would like to use AI to attack you. Uh, is Palo Alto the best place for all of this?

Is it a necessary place for all of this? I think it is. If you're Palo Alto's investors because they, they want to see something being done with agentic ai, right?

I mean, LLMs were so last year or was it the year before? Whatever, it doesn't matter. New, new things, new new stuff, right?

A agents, what are agents? Um, there's stuff that does things for me and stuff, but they need to be secure. 0 comes into play.

Uh, we're gonna secure your models and secure your agents so nobody can do anything with them than they're not supposed to do. You know, this is, uh, like the scene in the movie where the, the positronic net brain has some kind of a security mechanism and, and it'll keep you from, I don't know, corrupting it or you can just wait for it to do it on its own because we're still kind of at that point where AI likes to make stuff up. Uh, the other thing, agen X is all about making and building these agents.

Um, you may remember this from, uh, your early days in, in working in computers. This was called, uh, writing a program. And, and it's weird because that's really what you're doing is you're writing yet another program that runs on your network.

And yeah, it looks for threats, it gives you suggestions for the responses. If there's a policy enforcement violation, it can take care of all those. I want you to go out and do a favor for me.

I want you to survey all of the security analysts and operations people in your organization and ask them how comfortable they are with an AI agent handling threat remediation and policy enforcement for them on their own. Just, just curious, do you get like, um, oh, I think that's a really good idea. Um, I would love to be more efficient in my, my, uh, role or do you get Oh my God, no, not a chance because I know which one you're more likely to get.

I understand that a lot of people are really excited about the, the concept of what an AG agentic AI system can do. And I love that these are the same people that wanted to take a lot of these zero trust, uh, tools and just turn them on and see who screamed the loudest to determine what needed to be adjusted. Did you know that the screen test is actually a thing in it where you implement a policy change and you wait for people to scream before you adjust it?

Yeah. Are you as horrified by that idea as I am? 'cause 'cause I, I get it.

Everybody has to have AI in their system, right? And if you're a security company, you better be securing the AI that people are wanting to use and you better be coming up with ways to build AI to do stuff for you. And I can't wait to see if Agentic AI can make it into 2026 as the hot new thing, or if we're gonna discard it off to the wayside and move on to some other thing that everybody says we have to have in our product.

Boy, I really hope that we don't get burned on this one, but if we do, there's an agent for that. We had a story that we wanted to take a closer look at, and we've alluded to it because it involves the US federal government and the fact that they're considering taking a stake in several quantum computing startups like Adam Computing, D-Wave Ion Q tti, and Quantum Computing and Exchange for federal funding of at least $10 million each. This program is gonna be managed through the Commerce Department's chips program.

You remember that one. And the plan aims to support us quantum companies provide needed capital for them, and potentially earn returns for taxpayers reflecting the government's growing role in strategic tech investments. Al we've seen the US federal government buying a stake in companies as of late, including Intel and others, but do you think that they're going out on a limb here by trying to invest in quantum startups?

Well, it's not a lot of money they're talking about here. I mean, it does say at least 10 million, but I mean 10 million's, a pretty small amount of money when you're building brand new kinds of infrastructure. Um, just as a, 'cause we were talking about liquid nitrogen cooling before for, um, quantum computing.

I was looking at that, and that's, that's probably one of the costs these companies have and costs about a buck a gallon to make liquid nitrogen. Uh, so, you know, 10 million liters, uh, 10 million gallons of liquid nitrogen, that's quite a lot of liquid nitrogen, but that's just a consumable. That's not the thing that they're actually building here.

So, uh, unless the amounts being invested here are a lot larger and don't see that, it's actually gonna make a, a huge amount of difference to the financials of these companies. The validation that the US government thinks that they're strategically important enough to the, to the country, because fundamentally that's gotta be why you're investing here. Uh, speculative investment is the, the job of the venture capitalists, the government's job is to, uh, look after the country as a whole.

And so investing in these companies is really putting a, a little sticker on them saying, uh, tested and approved by the US government more than, I think making a big change to the, the business model in here. Uh, absolutely these companies will be very happy to get that sticker and to be able to show that sticker off to the, uh, venture capitalists who are going to be the ones who front up the billions of dollars it's going to take to commercialize a quantum system. Uh, of course this is all still proposed and maybe, and, and talked about and thought about, uh, and it's partly because these American companies are, are struggling to secure funding.

And so that sticker saying approved by the government has, has gotta be a good thing. How much return are we going to get to see? I'm not sure that the, the returns are going to be near, and like a lot of these, uh, early investments, there's gonna be a lot of failure along the way.

So you wanna see a big return from a relatively small number of these investments that you make, just like any other venture capital, uh, does follow that precedent of taking a stake in, uh, Intel recently, uh, that was a much larger investment. It was a nearly 10% stake in Intel, uh, as well as some of the securing of rare earth, um, materials that have been, uh, um, that's the Pentagon side, making sure that they've got a secure military supply over areas. It's, it's not quite the same as a, a central government investment.

So yeah, there's some, some prior art of these investments, but, uh, I don't know to, is this a good use of your taxpayer money? No, no, it's not. Uh, I I, I'll go out on a limb here and say, uh, $10 million for a lot of these companies, relatively speaking is a drop in the bucket.

It, it covers a lot of operational costs, really. And when you consider that a lot of these companies are already kind of backed by very large organizations anyway, why on earth would the US federal government want to basically kind of toss and change at people? Well, a lot of these companies are private, which means they don't have to disclose anything about what they're doing, what they're working on, what their direction is.

And so if I buy into them a little bit, that gives me a look into the company, including their financials and their projected plans, right? Because if as an investor they have to brief me on it, even if I own a pittance of the actual overall company, now you, you wouldn't think that that would give me the ability to, I don't know, influence the company's direction, right? Like that would be wrong is if the US federal government stepped in and kind of directed the company to do certain things, to research in certain ways to create opportunity for us, uh, interests to be coming out on top.

No, I I, that's a conspiracy theory, right? I'm, I'm sure there's some group of political people out there that believe that the government should keep their hands off of businesses, right? You, you can go look up which one that one is, and then giggle when you realize who was asking for this.

I think that the US government really does need to keep their hands off of this. What they need to do is they need to use the National Science Foundation as the vehicle to move money into these things because it creates a layer of insulation. Because you said that a lot of these companies are gonna wait for that little sticker that says the US government invested in us and we must be great, right?

Then. I'm gonna go up to Sand Hill Road and I'm gonna convince all of those people that they wanna invest in me. And what I'm seeing when I see that little sticker on the, the prospectus is so there's an outside actor with more power than me that can step in and basically invalidate my investment.

If they decide, you know what? We don't like the way that this is going. We're gonna create regulations that's gonna allow you to not work on this anymore.

We're gonna pull your government contracts for this thing because we don't like the way that you have, uh, ignored us on this one little thing that we asked for over here. You're probably sitting there saying to yourself, there's no way they could do that. They've done that.

The, the current administration has done that many times recently, and there's nothing stopping a future administration from doing that on either side of the aisle. If you think that I'm trying to, to play a certain kind of role here, anybody could do that. And that little 10%, $10 million investment gives them the foothold that they do because if the government hadn't invested in them, government wouldn't have any idea what to do.

And so I think that this all needs to be run through the National Science Foundation. It needs to be run through outside parties that are not directly involved with the day-to-day operations of the US Federal government. Quite honestly, I think it might be time for the, the federal government to get out of the business of investing in companies hoping to get a payoff.

'cause I will tell you that as a taxpayer, I'm never gonna see that money. In fact, I don't know that the federal government's gonna see that money. I I don't have any visibility into where it goes and I'm a stakeholder in the US Federal government.

You're not. But I am. I do wonder also whether that insight into what they're doing might really be something where the funding should have come from dapa.

That it's actually the, the Defense Department, sorry, ministry of War that, uh, wants to look inside these, these quantum companies and possibly see the, uh, military applications for it. Because I think maybe that's the, the interesting insight on that rather than ncf. Although of course I'm sure there's some, uh, some spokes who, uh, are part of or at least observing what happens with NSF.

Of course, if you'd like to observe what's going on in the future, we highly recommend you check out the AI Field Day that is running October 29th, why that's today and tomorrow October 30th, Stephen is out in the Bay Area with a great lineup of companies and delegates as usual. And of course, you should probably be watching that live stream as well. Then we've got a little moment of break before you have to travel, Tom.

That's right. I am gonna be out in Silicon Valley on November 5th and sixth with Networking Field Day. We are gonna have a great lineup of companies and they're all gonna be talking about ai.

Surprise, surprise. com and check out the lineup. We've got the schedule posted.

Uh, gonna be some great conversations, some wonderful delegates that are joining me for this one. Uh, it's kind of fun, uh, but it's not the most fun that's gonna be had in the month of November. I think that might belong to you, Al.

'cause what are you doing the week after, The week after I get to spend some time at KB Con North America and Atlanta? So we'll have a day of presentations at KubeCon. We've got, uh, south Works and Traffic Labs and VMware by Broadcom presenting all looking at cloud native Kubernetes stuffs and some of the operations around them.

So we'll be on, uh, November the 11th. I, of course, I'm gonna spend some more time at KubeCon learning about all of the interesting innovation and vendors that are there. But that week you're gonna be somewhere as well.

Do the people at home get to know where you are spending your November? Uh, actually I'm gonna be working at the Commvault Shift event that's going on in New York City Now, Steven's hosting it, but I'm actually gonna be a delegate for this one. com before you know it.

Um, but I've got a great group of people that are gonna be joining me there. I get to go to Times Square. Um, maybe I go get to visit the whole all guys, who knows.

But I also get to learn about data protection and all the cool stuff that Commvault's been working on, and that market is exploding so fast right now. Uh, like the, the investments that are going on, the acquisitions that are happening, uh, I'll have a lot to say and Steven will have a lot to say as the host of that event. So make sure you check out our website for more details.

We also hope that you will join us in the future for more great episodes of the Tech Field Day rundown. We're here every Wednesday, sometimes with the two of us as co-hosts, and sometimes we invite new people. Uh, there will be somebody co-hosting next week because I'm gonna be out at Networking Field Day.

But no matter what happens, make sure that you catch us when we are streamed on Wednesday. Uh, make sure you subscribe to us in your favorite podcast application, because that's a good way to listen to us when you're mowing the lawn or, um, painting the walls. Just don't pick yellow.

Evidently. It is the slowest color to dry, as I was told by my wonderful producer, Corey. Um, but whatever you choose to do, we hope that you leave us a rating and a review so that everybody knows how great we are around here.

But we're only great because you're great, because you continue to listen to us and you keep showing us stories that we definitely wanna cover. We'll be back next week with more great information and a little bit of fun. But until then, take care of yourselves.

Have a happy Halloween, and we'll see you in November.