WEBVTT
00:00:13.039 -->
00:00:18.480Hello, and welcome back to another episode of the Cables to Cloud's monthly news update.
00:00:18.480 -->
00:00:20.320I'll be your host for today.
00:00:20.320 -->
00:00:26.719So my name is Chris Miles, and joining me is the lovely and ever so beautiful Tim McConaughey.
00:00:26.719 -->
00:00:31.600We're kicking it old school here, just the two of us, uh just like old times.
00:00:31.600 -->
00:00:48.880Um and uh if if this is your first time tuning in, basically what happens is uh Tim and I, and sometimes our our third co-host, Catherine, will kind of collect some of the news stories that we found interesting over the last four weeks and dive into them and uh give you a little bit of commentary along the way.
00:00:48.880 -->
00:00:53.039So with that out of the way, let's go ahead and hop into the news.
00:00:53.039 -->
00:01:17.680So the first story that we have up here is um BT, um otherwise known as British Telecom and Verizon International, um, are combining and they're combining with BT International and Verizon's international enterprise wireline business into a 50-50 joint venture with Verizon paying BT a whopping $625 million equalization payment.
00:01:17.680 -->
00:01:23.200Um they're essentially forming a new entity incorporated in Jersey, which is in the UK.
00:01:23.200 -->
00:01:32.159So if you're familiar with New Jersey, this is the predecessor to New Jersey in the UK, led by CEO designate Martin Martin.
00:01:32.159 -->
00:01:35.920I'm not even sure how you say his first name, first name uh blanking.
00:01:35.920 -->
00:01:56.000Um carry roughly 4 billion in combined uh annual revenue over the 3,000 multinational customers across 180 plus countries spanning their SD WAN, MPLS, Ethernet, cloud connectivity, voice security, and uh other businesses with BT's global fabric as their orchestration platform.
00:01:56.000 -->
00:02:01.200Um so you know, we've we've seen how joint ventures have played out in the past.
00:02:01.200 -->
00:02:03.519Um how do you think this one's gonna go, Tim?
00:02:03.760 -->
00:02:18.800So I think it's interesting that you've got BT and Verizon, who essentially are competitors like in this space and have been for many years, uh deciding that, okay, well, neither of us like, you know, you gotta wonder what's the driver, right?
00:02:18.800 -->
00:02:22.719What's the driver for two competitors to enter the market together?
00:02:22.719 -->
00:02:31.120Um sometimes it's as simple as, hey, we're tired of competing against each other, uh, you know, so let's both work together and and get a fixed price.
00:02:31.120 -->
00:02:36.319So that's you that's usually not a great thing for the consumer, but who knows why that's happening here.
00:02:36.319 -->
00:02:48.800Um what I think's interesting is let's assume that governance, let's assume the they get, you know, the all of the um governmental approval to do this.
00:02:48.800 -->
00:02:52.719Does that mean that we'll see customers, you know, have to migrate?
00:02:52.719 -->
00:02:55.039Well, there be I don't know if there's really a migration here.
00:02:55.039 -->
00:02:57.680Like if they're on BT now, will they have to migrate to Verizon?
00:02:57.680 -->
00:03:03.280Or is it mentions the story mentions that Global Fabric will be the orchestrator, but nothing about the actual circuits.
00:03:03.280 -->
00:03:08.319So maybe maybe that's not a thing that's that's that's up for uh discussion.
00:03:08.319 -->
00:03:14.639But yeah, I mean it's interesting to see two competitors enter the market together as a joint venture.
00:03:14.639 -->
00:03:20.159What I think is happening is that neither of them want to own the whole of it, right?
00:03:20.159 -->
00:03:24.159So maybe maybe this is actually uh this could be even a price saving measure, a measure.
00:03:24.159 -->
00:03:26.080Like neither one of them wants to hold on to it.
00:03:26.080 -->
00:03:32.319So, you know, maybe both of them want to go in together and share the cost of of running a global win.
00:03:32.319 -->
00:03:33.439So yeah, I don't know.
00:03:33.520 -->
00:03:34.639Uh what do you think?
00:03:34.639 -->
00:03:37.520That that's uh that's kind of where I'm at as well.
00:03:37.520 -->
00:03:40.319Like you you mentioned that what does this mean for the customers?
00:03:40.319 -->
00:03:42.080Like, is this gonna be some type of migration?
00:03:42.080 -->
00:03:48.560And I assume with over 3,000 customers and 180 plus countries is a pretty damn big footprint.
00:03:48.560 -->
00:03:59.759Um and you know, if BT's kind of global as a global fabric is the orchestration platform, all those Verizon customers are not on that platform today, so we got to get them there somewhere.
00:03:59.759 -->
00:04:11.759And across that big of a footprint, I imagine there are plenty of realms in there where there's both Verizon and BT acting jointly um as as even competitors in the in that stature.
00:04:11.759 -->
00:04:31.120So um there's probably gonna be, you know, a lot of I assume deduplication in a sense in that, you know, if maybe they have pops in the same areas, if they have, you know, circuits, if they have N and I's and things like that all put in the same areas, that's gonna be probably a far bigger migration for some customers than others.
00:04:31.120 -->
00:04:36.720Um and they're probably gonna have their their hand forced in a certain direction, which is uh interesting.
00:04:36.720 -->
00:04:44.560But, you know, like we said, this is we're talking about NPLS here, this which is um I've learned over the course of working uh for global companies.
00:04:44.560 -->
00:04:46.879Uh people keep saying MPLS is dead.
00:04:46.879 -->
00:04:49.759There are parts of the world MPLS is not dead, my friend.
00:04:49.759 -->
00:04:53.839Uh it is very it's very alive and well and used quite a bit.
00:04:53.839 -->
00:04:56.399So um I imagine we'll we'll see some of that there.
00:04:56.639 -->
00:04:57.279Yeah, for sure.
00:04:57.279 -->
00:05:07.680The I mean, truthfully, I think NPLS is actually still everybody talks about ST WAN, and ST WAN has certainly become more of a ubiquitous technology in the last 10 years.
00:05:07.680 -->
00:05:14.399Um, but honestly, I still think the lion's share of WAN connectivity is not software driven, actually.
00:05:14.399 -->
00:05:15.680You just don't hear about it, right?
00:05:15.680 -->
00:05:20.319Like it's still the lion, it's the the the the dark matter of the WAN universe, if you will.
00:05:20.319 -->
00:05:22.079Like it just it's just there.
00:05:22.079 -->
00:05:24.160Um so yeah, anyway.
00:05:24.160 -->
00:05:26.639All right, so moving on uh to story number two.
00:05:26.639 -->
00:05:37.519So Virginia has has enacted the US, the US's first per kilowatt hour data center uh tax, electricity tax specifically, right?
00:05:37.519 -->
00:05:47.759So there's a data center power tax that Virginia has now approved, uh, which I I'll be curious to see if other states follow, if this becomes a big revenue driver.
00:05:47.759 -->
00:05:49.199Um, but yeah, anyway, sorry.
00:05:49.199 -->
00:05:55.279So Virginia approved the first US tax tied specifically to data center electricity consumption.
00:05:55.279 -->
00:06:05.040And it's it's very small, it's like 0.011 uh cents per kilowatt hours on all power consumed by data centers.
00:06:05.040 -->
00:06:14.720But you know, data centers, of course, being aggregators of lots of of power consumption, this this is probably going to end up being a good amount of money.
00:06:14.720 -->
00:06:20.079So as effect effective, sorry, effective as of July 1st.
00:06:20.079 -->
00:06:24.800So this is actually right now like happening in Virginia, these data centers.
00:06:24.800 -->
00:06:27.920So I'm very curious to see what those revenue numbers look like.
00:06:27.920 -->
00:06:33.600And um, so this is covering grid, competitive retail, and self-generated sources.
00:06:33.600 -->
00:06:43.839So uh in data center alley, remember the Northern Virginia, of course, is like the biggest East Coast uh conglomeration of data centers.
00:06:43.839 -->
00:06:49.360You know, this levy uh amounts to roughly 10% increase in electricity rates.
00:06:49.360 -->
00:06:58.160But for data centers, you know, that's the important part, for data centers, and it's projected to raise about, oh well, they they're they're thinking it's gonna raise about 600 million annually.
00:06:58.160 -->
00:07:07.040Um, and then so industry watchers basically are thinking, okay, so Virginia's gonna do this, and as soon as it's proved effective, you know, other states are gonna follow.
00:07:07.040 -->
00:07:13.839And I I think that so this the fact that they're including self-generated power is interesting.
00:07:13.839 -->
00:07:17.439So what do you uh what's your thought about this one, Chris?
00:07:17.920 -->
00:07:26.240No, yeah, I think the um, yeah, the self-generated power thing in here is kind of a um like you said, it's an interesting detail.
00:07:26.240 -->
00:07:36.720Um, and I think it kind of addresses probably a significant work workaround or a loophole, whatever you want to call it, probably before anyone had the the chance to actually enact on it.
00:07:36.720 -->
00:07:46.079Um so if you have some type of like on-site gas turbine or something like that, it's not exactly going to uh to save you in paying any of the these taxes here.
00:07:46.079 -->
00:08:01.680Um, you know, you you also point out that you know Ashburn, Virginia, Northern Virginia is kind of the um the main um uh East Coast location for many of the the large-scale hyperscalers and things like that.
00:08:01.680 -->
00:08:17.519Um if we think in terms of AWS, I think probably a lot of services they run that run out of US East One uh either like one, that's probably where if there are um kind of cheaper consumption options, they are probably available in US East One.
00:08:17.519 -->
00:08:21.040They have the most availability zone, so it has the highest resilience.
00:08:21.040 -->
00:08:29.680Um but also if if US East One takes a shit, uh AWS kind of completely takes shit as well, which is uh kind of beside the point.
00:08:29.680 -->
00:08:32.480But um it just kind of shows you how much is there.
00:08:32.480 -->
00:08:44.320Um I think all that we can really hope for is that you know, 600 million sounds like a lot of money, but in in the grand scheme of what these guys are earning is like it's it's pennies.
00:08:44.320 -->
00:08:55.440Um so I don't think anyone's gonna like start to move away from the this you know, data center alley East Coast uh kind of uh flagpole that they have.
00:08:55.440 -->
00:09:01.440But um I hope it I hope it permeates out to the rest of the the country as fast as possible, right?
00:09:01.440 -->
00:09:14.159Because then it could kind of if if it's an even playing field, uh you know it's it's it's uh kind of makes the bar of entry equal everywhere, probably gonna be higher in somewhere like California or something like that.
00:09:14.159 -->
00:09:17.519But um nonetheless, it it we know what happens, right?
00:09:17.519 -->
00:09:23.519If if if any of these companies end up having to pay more to run their equipment, it's not like they just eat the costs.
00:09:23.519 -->
00:09:25.039They they push it onto us, right?
00:09:25.039 -->
00:09:32.000So we end up paying the taxes at the end of the day if if if it um becomes too much of a cumbersome piece for them.
00:09:32.000 -->
00:09:38.240So um well, yeah, hopefully uh you guys all have 600 million laying around because I'm sure we're gonna end up paying it at some point.
00:09:38.480 -->
00:09:38.799Yeah.
00:09:38.799 -->
00:09:48.320Well the other the other the little so I'm glad they're doing this, but one piece that's missing from this is uh not missing necessarily, just I don't know.
00:09:48.320 -->
00:09:58.480You you the whole point of of charging the tax is in theory so that you know, not just to give the government more money to to, you know, burn up burn on hats, right?
00:09:58.480 -->
00:10:06.720Like the you want you really want this money that's raised to be earmarked for the environmental impact that a data center is going to have, right?
00:10:06.720 -->
00:10:15.360Or for the or or to offset the cost of uh you know other costs that are going to come as a result of those data centers that are that you know on the people.
00:10:15.360 -->
00:10:19.919Um which of course if you think about it, six hundred million dollars is is nothing compared.
00:10:19.919 -->
00:10:22.559You know, it's like it really is a drop in the bucket.
00:10:22.559 -->
00:10:24.879So yeah, I I agree on two two things.
00:10:24.879 -->
00:10:35.840The the the the data centers, sorry, the hyperscalers or whoever owns the data, whoever's whoever's owning the data centers, uh ultimately the cost will be passed on because of course that's how business works.
00:10:35.840 -->
00:10:45.200Um, which I always think is funny when people argue against taxing corporations because you know they'll just pass the ta the they'll just pass that on.
00:10:45.200 -->
00:10:51.919Um I I've never thought that was really a good reason not to tax corporations in general, but you know, I anyway.
00:10:51.919 -->
00:10:55.840I I guess that's somewhat true to the somewhat true.
00:10:55.840 -->
00:11:08.559But yeah, six hundred million dollars annually is not really gonna pay I mean if you think about one environmental disaster as a result of it of one data center, like that would be nothing if that actually happened, right?
00:11:08.639 -->
00:11:10.240So yeah.
00:11:10.240 -->
00:11:14.559Not to not to mention there's there's there's a recurring theme around the country.
00:11:14.559 -->
00:11:20.159I think uh everyone pretty much knows all the complaints that are being made about data centers across the globe, right?
00:11:20.159 -->
00:11:37.840So um I think there's one like Microsoft's getting sued in, I think it's Wisconsin or something like that, where the amount of noise that gets that's getting generated from the um the data center has become a kind of a public nuisance for the people in the neighborhood, and Microsoft's actually getting a class action lawsuit out of it.
00:11:37.840 -->
00:11:53.440Um if anything, I think I think some of these uh kind of bigger players in the game are learning they're they're testing the the waters with how much they can get away with before before they get their hands slapped.
00:11:53.440 -->
00:11:55.600So um, you know what?
00:11:55.600 -->
00:11:57.440I say complain about every little thing.
00:11:57.440 -->
00:12:00.159Let's make it as hard as possible for them.
00:12:00.720 -->
00:12:09.840Well, and as we know, uh when it comes to businesses, right, uh a fine a fine that isn't truly punitive is just cost of doing business.
00:12:09.840 -->
00:12:11.200Yeah, exactly.
00:12:12.799 -->
00:12:15.759It's basically it becomes run rate at that point, pretty much.
00:12:15.759 -->
00:12:16.559All right.
00:12:16.559 -->
00:12:17.679Uh moving on.
00:12:17.679 -->
00:12:18.960Our third story here.
00:12:18.960 -->
00:12:25.759Uh this one's gonna be a mouthful, but uh we have uh kind of a setting uh a theme for some of the later stories here.
00:12:25.759 -->
00:12:31.120We have a uh uh White House post-quantum cryptography executive order.
00:12:31.120 -->
00:12:53.440So in this executive order, which is uh executive order 14412, um it is labeled Securing the Nation against Against Advanced Cryptography Cryptography the Nation Against Advanced Cryptographic Attacks signed uh just on the 22nd of June, and it puts hard dates on the federal migration to a post-quantum cryptography.
00:12:53.440 -->
00:13:08.320Um so agencies must name a PQC migration lead within 30 days and submit migration plans to OMB and ONCD by October 22, 2026, um, which is a pretty uh pretty close deadline here.
00:13:08.320 -->
00:13:18.399Um so high value assets and high impact systems must move to post-quantum key establishment by the 31st of uh December in 2030.
00:13:18.399 -->
00:13:23.919So there's uh I think everyone's kind of considering the quantum D-Day is 2030.
00:13:23.919 -->
00:13:26.000I've heard it pushed up to 2029.
00:13:26.000 -->
00:13:44.399So um I I think most most listeners of this show are gonna are gonna already be aware of what quantum D-Day is, but um just for those who who may not be aware, is there there's this concept that is very prevalent in the market today about um harvest now and decrypt later.
00:13:44.399 -->
00:13:56.240Um so the the basically the 50,000-foot view is that there's bad actors out there that are just basically harvesting um probably upwards of petabytes of data that is just encrypted data.
00:13:56.240 -->
00:14:08.799Um they're harvesting it now until they get access to a quantum system that can then decrypt that um and decrypt the key exchange um and decrypt the entire payload and and they'll have access to whatever was there.
00:14:08.799 -->
00:14:13.519Um so that's kind of the uh the concept around quantum D-Day.
00:14:13.519 -->
00:14:16.080But um, I don't know what do you what do you think, Tim?
00:14:16.080 -->
00:14:25.759Do you think this is uh this is actually a good a good thing coming down from this uh you know lovely establishment that's been so consistent about technology in the last few years?
00:14:26.320 -->
00:14:31.039Very consistent, supportive of certain companies in technology.
00:14:31.039 -->
00:14:44.080Uh so I mean, well, I mean, as you know, like but Cisco, Fortnet, like all the all the vendors out there have been talking about uh PQC D Day or Quantum DQ Day or whatever for a while now.
00:14:44.080 -->
00:14:52.639Um, you know, because a lot of enterprises are, you know, I mean, obviously it's a staggered cycle, but but they're doing refreshes, right?
00:14:52.639 -->
00:15:06.320So so we're now targeting Q Day, meaning, you know, when when a quant when a uh quantum computer capable of doing decryption on today's algorithms, you know, like you said, is is going to be available uh to be within five years, right?
00:15:06.320 -->
00:15:10.000Like that's what I've but I've heard the same where it could be as early as 2029.
00:15:10.000 -->
00:15:20.240What's what's interesting to me is this idea that like Q Day will happen, and you know, you'll read about it on CNN that Q Day might have already the thing is that nation very true.
00:15:20.399 -->
00:15:20.639Yeah.
00:15:20.879 -->
00:15:30.879Like there's no unl unless unless some hacker group does it and and feels very uh ready to throw away their advantage, they're not gonna come out and tell anybody.
00:15:30.879 -->
00:15:36.879Like the people who are the people who can make Q Day happen are not going to tell you that Q Day has happened.
00:15:36.879 -->
00:15:39.919So, you know, for all we know, it could have already happened.
00:15:39.919 -->
00:15:40.720Who knows, right?
00:15:40.720 -->
00:15:46.240But but realistically, uh within the next five years is the number I keep is the number I keep hearing.
00:15:46.240 -->
00:15:49.840But I think even as early as 2029, which is insane if you think about it, right?
00:15:49.840 -->
00:16:01.039We're talking about something within a re a refresh cycle of the next three year, three to five years, uh, we could have, you know, all of our encryption uh could be essentially useless, right?
00:16:01.039 -->
00:16:02.960Or useless to some degree.
00:16:02.960 -->
00:16:10.399There, there's there's there's there's older algorithms that are actually quantum resistant, which is kind of in which is kind of interesting to think about.
00:16:10.399 -->
00:16:17.200But yeah, so from a PQC mandate perspective, um this is not un this is not surprising to me.
00:16:17.200 -->
00:16:21.759I think we had we had even seen some of this earlier, but now this has become an executive order.
00:16:21.759 -->
00:16:31.120So what it means is is that uh anybody touching government data is gonna have to figure out their their migration plan.
00:16:31.120 -->
00:16:42.000And honestly, in most of these cases, almost all these cases, especially since the government likes to run really old hardware, um, it's gonna be uh refresh, which sounds great for the for the vendors.
00:16:42.000 -->
00:16:45.840It sounds great for the vendors that are that are gonna have to supply this uh hardware, right?
00:16:45.840 -->
00:16:59.360So I expect a flurry of activity, a lot of RFPs, uh a lot of just a huge amount of uh f of just I don't know, just stuff happening here around this uh within the next year.
00:16:59.360 -->
00:17:02.639Um yeah, that's that's what I think.
00:17:03.039 -->
00:17:05.039Yeah, I'm uh I'm kind of right there with you.
00:17:05.039 -->
00:17:14.720I think it's uh I think Tim and I obviously we're we're speaking this from a specific lens because we're both in vendor land and have been in vendor land for some time now.
00:17:14.720 -->
00:17:23.519But um yeah, I think the you know we both work for vendors that you know we've we've been post-quantum compliant for for several years now.
00:17:23.519 -->
00:17:29.279It's um we're not exactly um probably the focal point.
00:17:29.279 -->
00:17:41.599We're not uh you know, if you if you if you have decided you want to um be post-quantum uh resilient, you're probably not going to your network infrastructure provider at first, right?
00:17:41.599 -->
00:17:46.960There's a there's kind of a whole whole kind of piece that you need to work down from the stack.
00:17:46.960 -->
00:18:06.240Um there's a lot of other vendors that are gonna have uh stronger issues getting this done, but I think this is good because it will probably force the hand of some that are doing a lot of business with the with the federal government today to make sure that they are post-quantum um uh post-quantum compliant.
00:18:06.240 -->
00:18:18.480Um, because I mean it like like Tim said, there's uh I remember to I remember when I first got like I was getting my CCIE, people were like, dude, get get a CCIE and then get to get top secret clearance.
00:18:18.480 -->
00:18:26.559Because if you have those two things, you will always be paid like for for your entire life, because there's always money to be made in defense.
00:18:26.559 -->
00:18:28.960Uh there's there's there's so much spending there.
00:18:28.960 -->
00:18:32.640So if there's um if they're gonna spend it, people are gonna flock to it, right?
00:18:32.640 -->
00:18:37.920So I think this will uh this will end up being um a good thing, I guess.
00:18:37.920 -->
00:18:41.839Um but the thing is that it feels like we're we're already kind of there.
00:18:41.839 -->
00:18:46.319So it's it's just uh um we'll see where it goes.
00:18:47.359 -->
00:18:47.680Yeah.
00:18:47.680 -->
00:18:52.799I I mean the there's a lot of balls in flight on this one already, even before the executive order came out.
00:18:52.799 -->
00:18:58.720And but you know, a lot of apps uh have been post-quantum compliant to some degree already.
00:18:58.720 -->
00:19:00.559A lot of them haven't been as well, right?
00:19:00.559 -->
00:19:02.960There's no there's been no forcing.
00:19:02.960 -->
00:19:12.960So I'm curious to see from the app layer, even, you know, apps that the federal government uses, they're gonna be requiring apps to be probably post quantum uh compliant as well.
00:19:12.960 -->
00:19:14.799So it's gonna go all the way.
00:19:14.799 -->
00:19:25.920Pretty much anybody who does business with the government with and any tech stack will have to approve just like with NIST and with other you know frameworks that you know you're compliant with this framework.
00:19:25.920 -->
00:19:27.839So really interesting.
00:19:27.839 -->
00:19:30.960All right, moving on to the next story.
00:19:30.960 -->
00:19:42.480So Cisco is acquiring or has announced intent to acquire a company called Wide Field Security to boost uh Splunk's Agentic SOC is the title of the thing.
00:19:42.480 -->
00:19:47.920So basically Cisco's uh announces intent to acquire this thing called Wild Wide Field Security.
00:19:47.920 -->
00:19:50.640It's an identity lifecycle security firm.
00:19:50.640 -->
00:20:04.079So which is so the platform discovers human and non-human identities and maps exposure across the accounts and roles of those identities and then looks for like policy misconfiguration.
00:20:04.079 -->
00:20:11.039So like kind of like the IAM, you know, in the cloud, like the IAM kind of uh stuff that you can use to do that.
00:20:11.039 -->
00:20:26.799So Cisco is hoping to or planning to fold in Widefield into the Splunk agentic sock uh, I guess, product or or solution to add some identity and credentials, sessions, you know, context to threat investigations.
00:20:26.799 -->
00:20:33.599So this is the third uh acquisition of a cybersecurity uh firm by Cisco in 2026, by the way.
00:20:33.599 -->
00:20:39.440So uh of course, full full disclosure, I work for Cisco, so just you know, I'm just reporting this story.
00:20:39.440 -->
00:20:41.920That's I don't actually have any skin in this particular game.
00:20:41.920 -->
00:20:47.359This is not uh a business unit or a uh part that I that I deal with.
00:20:47.359 -->
00:20:57.759So but what's interesting about this is that it's uh nine not we're talking about human and non-human entities, so you know, take that for what for what it means.
00:20:57.759 -->
00:21:03.920Um but like before we before we get into it too far though, you know, w what do you think?
00:21:03.920 -->
00:21:05.599Uh what do you think about that, Chris?
00:21:05.839 -->
00:21:31.200Yeah, um, so I mean it's uh like uh I was at RSA earlier this year, and I will say two probably the three biggest themes that I that I was hearing about was uh post quantum, which we already covered, uh agentix sock, which is the other one, and then um non-human identities was probably the third one that was definitely coming up a lot.
00:21:31.200 -->
00:22:04.079And um it does not surprise me at all that that Cisco is making an acquisition to kind of um uh kind of broaden its portfolio into this field because this is uh you know, I talked to I talk to a large portion of the financial sector here in Australia, um, which is maybe not maybe not on the same advanced cycle as some of the banks in the US, but they are doing some really interesting things and the This whole idea about what to do with non-human identities is a real problem.
00:22:04.079 -->
00:22:07.599Um, and no one really has a good way to solve it just yet.
00:22:07.599 -->
00:22:12.160There's a there's a lot of ideas in the air, like you said, there's a lot, there's a lot of kind of balls in the air.
00:22:12.160 -->
00:22:18.640Um, there's a lot of different ways to do things, but nothing's been kind of landed on as a as a as a standard per se, right?
00:22:18.640 -->
00:22:30.079Um so if we're talking about non-human entities, like uh this could be a service account, this could be a workload identity, this could be um this could be an an AI agent that's built specifically to run a certain task.
00:22:30.079 -->
00:22:44.640Um, and there's this whole kind of uh you know idea back and forth about whether or not an agentic identity needs to be inherited from an actual user identity or if it should come from a service account or should be like completely tied to the task at hand.
00:22:44.640 -->
00:22:49.839Um and then how do you kind of, you know, um enroll that agent?
00:22:49.839 -->
00:22:54.559How do you um validate it and then how do you let it assume its identity, things like that.
00:22:54.559 -->
00:22:58.640If you think about this, you can draw a lot of parallels to Kubernetes.
00:22:58.640 -->
00:23:28.079Um it's really this idea of a short-lived workload that has a very specific task, then but now you have to put a lot of guardrails around this about what it can inherit, what it's you know, whether it's up to um up to code um with what it's uh what the what its run times are, things like that, um, and guardrails around what it's potentially doing with um an agentic piece with like a natural language um versus um company uh kind of catered to the old way.
00:23:28.079 -->
00:23:30.880So um yeah, I think it's it's probably a good move.
00:23:30.880 -->
00:23:32.480I don't know much about Widefield.
00:23:32.480 -->
00:23:38.400Um reading through the article, it seems like they um I think this is kind of Cisco's thing, right?
00:23:38.400 -->
00:23:44.319They always uh buy up these kind of smaller, not necessarily bespoke, but very like focused companies.
00:23:44.480 -->
00:23:46.480Yeah, focused on a specific thing, yeah.
00:23:47.119 -->
00:23:57.200Yeah, and then and then you know it's us to uh up to us as the consumer to just really hope this doesn't just go into kind of be you hell on the back end and and and never make it out of anywhere.
00:23:57.200 -->
00:24:00.880So um, you know, the innovation will be the story, yeah.
00:24:00.880 -->
00:24:01.839Yeah, 100%.
00:24:01.839 -->
00:24:05.839Um, which I mean this is we know how big the market cap is here, right?
00:24:05.839 -->
00:24:11.039And and uh and where this can go, so I imagine this will get quite a bit of attention.
00:24:11.519 -->
00:24:18.240Yeah, no, I yeah, no, I I like the call, I like the uh comparison to Kubernetes.
00:24:18.240 -->
00:24:22.319Because I mean agent agentic identity is really kind of a Mayfly thing.
00:24:22.319 -->
00:24:34.480Even even for long-running workloads or or I guess a genic workloads or however you want to uh just agents in general are even even an agent that runs on a cron job is still a Mayfly entity, right?
00:24:34.480 -->
00:24:40.559The same same thing as a as a Kubernetes uh um oh my god, I just lost my train of thought.
00:24:40.559 -->
00:24:43.039Same thing as a workload in Kubernetes.
00:24:43.039 -->
00:24:55.119Um so just like in Kubernetes, identity becomes like really, really hard to prescribe, to uh, you know, ascribe and then to uh use essentially.
00:24:55.119 -->
00:25:12.799Like what is what is it except actually it's almost worse because once you with agents you have a reasoning engine in there that you know in theory could attempt to do things that you know Kubernetes workload is just code, essentially, it's just a m.
00:25:12.799 -->
00:25:13.279Right.
00:25:13.279 -->
00:25:22.799So you have an agent that could try to do things uh that maybe uh I you know we didn't plan for identity-wise, you know, like just who knows?
00:25:22.799 -->
00:25:36.240Like, you know, you build a uh so of course in that case you you would you would hope that the identities uh would be uh permissive or rather um sorry, uh least privilege rather than you know most privileged.
00:25:36.240 -->
00:25:59.119But I've seen, you know, we when we we talked with Catherine not too long ago about all the different and and uh there's an upcoming upcoming uh episode coming out where we talked with another security researcher about all the different ways that agents can not only can be manipulated, but can escape, you know, identi uh escape any kind of uh sandboxing or whatever that you put around them.
00:25:59.119 -->
00:26:05.279So yeah, this actually identity is extremely important, but I think it's gonna even go further than just identity.
00:26:05.279 -->
00:26:19.920Like you're gonna have to there's gonna have to be a uh a framework or a solution or something that's gonna actually enforce I you know, like whatever the identity is is allowing uh on top of this.
00:26:19.920 -->
00:26:22.240So yeah, interesting.
00:26:22.640 -->
00:26:26.880I think I think I've been an an enemy of learning about identity for the longest time.
00:26:26.880 -->
00:26:31.680It's come back to bite me in the ass because it's become it has come to fruition and become the new, what do they say?
00:26:31.680 -->
00:26:34.160Identity is the new uh the new edge or whatever.
00:26:34.160 -->
00:26:38.240Um so here we are, ladies and gentlemen.
00:26:38.240 -->
00:26:40.720Back in identity hell.
00:26:41.039 -->
00:26:44.400Uh course according to marketing, everything is the new edge, to be fair.
00:26:44.400 -->
00:26:45.119Yeah, true.
00:26:45.200 -->
00:26:47.599Yeah, something else will be the new edge in a couple years.
00:26:47.599 -->
00:26:49.279Maybe I'll just wait for that one.
00:26:49.279 -->
00:26:51.599All right.
00:26:51.599 -->
00:26:53.440Uh next up.
00:26:53.440 -->
00:26:58.000Uh so we have a couple stories here that are kind of closely related to the same topic.
00:26:58.000 -->
00:27:00.480So we'll we'll kind of cover these together.
00:27:00.480 -->
00:27:11.599But um the theming here is kind of the um, how should we call this like frontier AI models as uh I guess conditional infrastructure.
00:27:11.599 -->
00:27:37.519So um we've seen the government um kind of taking taking action in two different scenarios, one with OpenAI, um, when they've released their their newest GPT 5.6 uh soul um model, and then Anthropic releasing obviously the mythos um and um the fable models out to the general public.
00:27:37.519 -->
00:27:41.279Um and we saw kind of two different things happen here.
00:27:41.279 -->
00:28:09.839So Anthropic obviously was apparently this one as high as uh Andy Jassy at Amazon apparently flagged this with uh someone at either the government or with Anthropic that there was um you know mythos or uh I guess probably fable in the scenario was quickly jailbroken and you know it was flagged as a huge security risk for kind of the um the whole of their company as as well as a lot of uh modern infrastructure out there.
00:28:09.839 -->
00:28:15.119So the White House stepped in and kind of ordered them to restrict access to everyone.
00:28:15.119 -->
00:28:19.920Um I think they wanted did they didn't they want it secured to only US entities or something like that?
00:28:20.319 -->
00:28:25.680Yeah, they were specifically they pulled it for everybody usually can't verify who's accessing it, right?
00:28:26.400 -->
00:28:28.160So um we had that happen.
00:28:28.160 -->
00:28:42.559Obviously, over the coming weeks, Fable and Mythos have been kind of reintroduced and they've uh you know they've kind of stepped in and said that, you know, they've they've verified that the jailbreaks are not possible within this model, they've tightened the guardrails, etc.
00:28:42.559 -->
00:28:52.240Um GPT 5.6 kind of had a different um a different kind of route to the same outcome.
00:28:52.240 -->
00:28:58.480So they have uh Seoul is apparently their strongest model in terms of uh cybersecurity and things like that.
00:28:58.480 -->
00:29:16.559But I think this was it was originally released to kind of a limited access of 20 partners individually bedded by the US government um over concerns that the advanced cyber capabilities could reach, you know, the more hostile nation states and things like that, which I guess is kind of what Glasswing was.
00:29:16.559 -->
00:29:19.279Um it's probably probably relatively similar to that.
00:29:19.279 -->
00:29:27.759But OpenAI was pretty, I guess I should say, critical of that that path that they had to go down.
00:29:27.759 -->
00:29:35.440So they basically they complied with it, but they were not happy about it and they were vocal about it and and said that you know this shouldn't be the the long-term solution.
00:29:35.440 -->
00:29:38.000This should not be how things go through this process.
00:29:38.000 -->
00:29:43.119There shouldn't be like a pre-clearance with the government before having to to go to market and things like that.
00:29:43.119 -->
00:30:02.319Um and yeah, I think this is uh this is probably gonna have a lot of uh stipulations and kind of um uh how should I say kind of wavering uh committed parties uh to some of these AI models.
00:30:02.319 -->
00:30:03.599How how do you feel, Tim?
00:30:05.119 -->
00:30:06.000Oh boy.
00:30:06.000 -->
00:30:13.119So yeah, the so the government the government has been getting into the arm into the AI arms race for a while now, right?
00:30:13.119 -->
00:30:14.640Citing national security.
00:30:14.640 -->
00:30:23.920It's why we have things like data centers now being classified as national security uh so that you know pe communities can't push back on them.
00:30:23.920 -->
00:30:25.359Like we're gonna see more of that too.
00:30:25.359 -->
00:30:43.920The government's throwing its weight around a lot, and there's probably several reasons for it, one of which is just blind, you know, just bald faced uh that any one of these the people that run these models and are billionaires can pick up the phone and call somebody in the administration and probably get action.
00:30:43.920 -->
00:30:52.480Um, but also, you know, when th they'll there's a lot of uh fear about, oh this reminds me almost of the space race, right?
00:30:52.480 -->
00:31:07.920Like there's this fear that, you know, then it used to be Russia, but now it's China, uh, that our adversaries are, you know, not operating with the same uh restrictions that we put on ourselves and therefore they're gonna they're gonna beat us in in some way.
00:31:07.920 -->
00:31:16.240Um and so we need to, you know, do things like export controls and um you know stuff like this.
00:31:16.240 -->
00:31:24.400Uh and you're seeing it now with all of the all of the the frontier models, except for I guess except for Grok, but I don't even know if I would call that a frontier model.
00:31:24.799 -->
00:31:31.119Um, like put my server in a bikini or whatever.
00:31:31.599 -->
00:31:32.319Exactly, yeah.
00:31:32.319 -->
00:31:35.039Take take take the clothes off my server, please.
00:31:35.039 -->
00:32:02.319Um yeah, so the government can kind of arbitrarily decide for you know reasons known only to itself, citing national security now, uh, to you know remove access to to models, to stop export controls, um, you know, kind of like we used to try to do with uh the you know the tougher algorithms for crypto and stuff like this.
00:32:02.319 -->
00:32:08.720It's just not that pr you know, it's just not that uh we don't live in a world where it's that easy to do anymore.
00:32:08.720 -->
00:32:17.759So and and and I my understanding is that China's still releasing, you know, their own models that don't have you know that don't have these same things.
00:32:17.759 -->
00:32:19.680So I don't know if that's the right answer.
00:32:19.680 -->
00:32:27.519I don't know if it's the right answer to just decide arbitrarily out of nowhere, like, hey, this is this is export control, this isn't, we're gonna release this now.
00:32:27.519 -->
00:32:38.240And um yeah, but but the reasoning I think it has m has to do more with the market than it does with the actual defense of of the US.
00:32:38.240 -->
00:32:39.920Or the administration.
00:32:39.920 -->
00:32:40.720Yeah.
00:32:40.880 -->
00:32:43.440Yeah, that's the cynic in me, to be honest.
00:32:43.440 -->
00:33:20.880Yeah, it's um I mean if we if we kind of peel back just like intention and kind of you know potential uh fallacy and kind of the reasonings behind this, it just doesn't it just doesn't really make a lot of sense because like if you're if if you're a business and you're supposed to be building tools, if this is supposed to change the way fundamentally how you how you develop things, how you take things to market, and you've built a framework around using one of these models, and then all of a sudden the White House steps in and says, This is gone, like you cannot use this.
00:33:20.880 -->
00:33:24.960It it that that is that is completely impactful to your business.
00:33:24.960 -->
00:33:34.240Like it like you could like imagine how much if if luckily fable wasn't in the market that long for people to probably have you know full-on practices built around it.
00:33:34.240 -->
00:33:44.720But I mean, to be honest, we live in a world of startups, it's very possible that did happen, and then they had to completely kind of shift gears one day just because of um you know, kind of the swing of the gavel type thing.
00:33:44.720 -->
00:33:57.599So um I just don't know c like where this takes things because I don't think you can build, I I don't think you can put uh an this amount of risk on this one thing.
00:33:57.599 -->
00:34:03.759And and we've kind of seen that these these models, as good as they are, they're not interchangeable, right?
00:34:03.759 -->
00:34:11.360Uh it's not like, oh, they took my access away for for mythos, I'll just switch it over to to GPT 5.6 or something like that.
00:34:11.360 -->
00:34:20.159It's like it's not an apples to apples thing, and you know, that you get different outcomes depending on the the weights and and things like that and how the models trained.
00:34:20.159 -->
00:34:22.400So it's it's not exactly that easy.
00:34:22.400 -->
00:34:32.880Um you know, we didn't have time to fit this into the show today, but we've also seen there's there's plenty of other companies out there starting to try to make their own chipsets as well and make their own hardware.
00:34:32.880 -->
00:34:57.280And I wonder if this is kind of leading towards where people focus on the idea of running these things in-house versus uh relying on something external just to kind of stop this this external dependency that has obviously a lot of volatility based on an administration that can obviously, you know, there there will be a changing in the guard and there will be a new administration later, and and that keeps happening.
00:34:57.280 -->
00:35:00.800So I think they kind of want to um reduce the amount of risk there.
00:35:00.800 -->
00:35:03.599But it's uh I think I'm like you, Tim.
00:35:03.599 -->
00:35:04.719I don't know the right answer.
00:35:04.719 -->
00:35:06.719It's just like uh fuck.
00:35:06.719 -->
00:35:11.760I'm I'm glad I'm not a C I a CEO at this point because it sounds very difficult to do.
00:35:12.320 -->
00:35:13.199Yeah, no doubt.
00:35:13.199 -->
00:35:22.880Like it well, I mean again, market manipulation is uh is definitely a thing that this administration likes to do, so I wouldn't put any of that past any, you know, past it.
00:35:23.280 -->
00:35:31.119So um I just I just looked at my uh bank account and I I I do wish I was still a CEO, so I'll I'll change my answer there.
00:35:31.440 -->
00:35:32.719You were still still a CEO?
00:35:32.880 -->
00:35:40.239Still you know, sorry, I should say I I wish hasn't been telling Chris did not tell us something.
00:35:40.239 -->
00:35:44.000Uh CEO of Cable's Clouds Incorporated, obviously.
00:35:44.559 -->
00:35:45.039There you go.
00:35:45.199 -->
00:35:46.159Yeah Oh man.
00:35:46.159 -->
00:35:48.079Uh for a first dollar, man.
00:35:48.079 -->
00:35:49.039It's gonna take that.
00:35:49.599 -->
00:35:50.800Any any minute now.
00:35:50.800 -->
00:35:55.599Any minute now, it's gonna we're we'll be rolling in in uh quarters.
00:35:55.599 -->
00:35:56.159Yeah.
00:35:56.159 -->
00:36:09.519So to end on a good note, um I so I took my uh I took my daughter up to the University University of Asheville on Friday uh to visit the university there.
00:36:09.519 -->
00:36:10.480She loves it, by the way.
00:36:10.480 -->
00:36:14.079I have to figure out now how to get her in as a freshman instead of waiting for transfer.
00:36:14.079 -->
00:36:15.599So good luck to me.
00:36:15.599 -->
00:36:25.039Um but while we're up there, I reached out and uh so our our former co-host and good friend of ours, Alex Perkins, still uh lives up there.
00:36:25.039 -->
00:36:28.719So I reached out to him and we uh we had dinner and we were talking to him.
00:36:28.719 -->
00:36:29.440He's doing great.
00:36:29.440 -->
00:36:37.280He has a he has a new job that he loves, and uh he's still working on his college degree, and yeah, he's he's he's really really doing well.
00:36:37.280 -->
00:36:38.480So it was great to see him.
00:36:38.480 -->
00:36:45.199We got to got to hug it out and introduce our families and hug it out, bitch.
00:36:45.199 -->
00:36:46.159That's right.
00:36:46.159 -->
00:36:49.360I'm just kidding, hug it out implies there was some reason to hug it out.
00:36:49.360 -->
00:36:51.280We just we just gave each other a hug.
00:36:51.280 -->
00:36:56.800But uh yeah, anyway, so yeah, Alex is doing great, and uh we're yeah, the invitation's open, bud.
00:36:56.800 -->
00:37:00.880You can come back anytime or uh we'll have you on as a guest and catch up with you.
00:37:01.119 -->
00:37:01.840So yeah, good time.
00:37:01.840 -->
00:37:02.880Shout out, Alex.
00:37:02.880 -->
00:37:03.599We love you, man.
00:37:03.599 -->
00:37:04.320Done there.
00:37:04.320 -->
00:37:04.800Okay.
00:37:04.800 -->
00:37:09.519Uh so let's uh let's go ahead and wrap it up for this week.
00:37:09.519 -->
00:37:10.800Um, so thank you for joining.
00:37:10.800 -->
00:37:14.639If you enjoyed this, if you made it to the end, obviously you enjoyed a little bit of it.
00:37:14.639 -->
00:37:23.119So uh please put the word out there, share it in some capacity, whatever uh feels uh uh whatever feels uh inviting to you to do.
00:37:23.119 -->
00:37:27.760Um but yeah, just put the word out there, maybe introduce it to someone who maybe not heard the show before.
00:37:27.760 -->
00:37:33.920Um and uh send us an email, send us uh uh send us a DM, whatever you want to do, talk to us.
00:37:33.920 -->
00:37:35.840We'd love to hear from you.
00:37:35.840 -->
00:37:39.280Um and we will see you in a couple weeks.
00:37:39.280 -->
00:37:40.880Goodbye.
