Telecommunications Industry Therapy · 2026-06-01 · 18 min
Key moments - from our scoring
Substance score
33 / 100
Five dimensions, 20 points each
The TIA-942 standard has been the foundational framework for data center design and operations since its first publication in 2005, establishing itself as the primary voluntary consensus standard in the industry. Tom McGarry walks through the four-tier reliability classification system (Level 1 for basic enterprise IT rooms through Level 4 for fully redundant, unplanned-outage-resistant facilities) and explains why data center operators pursue third-party certification - primarily to demonstrate service availability commitments to customers and regulators. The certification program operates through eight independent certification bodies worldwide and has completed over 1,000 certification activities across 800+ unique data centers in 60+ countries. However, AI's emergence as a distinct data center category is forcing rapid evolution. Unlike traditional colocation or edge/modular data centers, AI data centers operate at dramatically different power densities (100+ megawatt racks versus 10-15 megawatt traditional racks), require liquid cooling instead of air conditioning, and demand exponential increases in cabling infrastructure. These changes risk undermining redundancy and availability - core tenets of 942. In response, TIA approved an AI Addendum in February to address high-density compute environments, with particular focus on maintaining concurrent maintainability and handling the structural load implications of heavier equipment. The conversation is essential for data center operators, infrastructure designers, and technology vendors involved in power, cooling, and networking solutions who need to understand evolving compliance frameworks.
Level 1 is a basic enterprise IT room with minimal redundancy; Level 2 adds some redundant components for planned maintenance; Level 3 (concurrently maintainable) includes redundant power, cooling, and cabling paths allowing equipment maintenance without service interruption; Level 4 is fully redundant and can withstand unplanned outages by automatically switching to backup systems.
Certification demonstrates to customers and regulators (like India's government) that a data center maintains a specific availability level through third-party validation, reducing contractual risk and enabling operators to market their infrastructure reliability with credible, neutral credentials.
AI data centers operate at 100+ megawatt power density per rack (versus 10-15 megawatt traditionally), require liquid cooling instead of air conditioning, and demand massive increases in cabling - changes significant enough that TIA approved a separate AI Addendum rather than wait for the next standard revision.
Anyone can join the TR42 committee through TIA's website; particularly sought are organizations involved in power systems, liquid cooling, data center operations, and infrastructure who want to influence standards alignment with their solutions.
TIA has agreements with eight independent certification bodies worldwide (audited yearly) who assess data centers against the four reliability levels; the program has certified over 1,000 activities across 800+ facilities in 60+ countries, with certificates posted on TIA's website as proof of third-party validation.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode provides a serviceable orientation to TIA-942 tiers and the emerging AI addendum, with a handful of grounding data points, but the bulk of the runtime is definitional explaining that would be familiar to anyone in the data center space. Very little per-minute yield beyond the certification volume stats and the liquid-cooling framing.
AI has added a basically third type of data center, which is an AI data center. And it's different enough that we need to now start looking at them very differently
they haven't really been installing them in a way that would make them fully redundant and highly available
There is no contrarian or first-principles argument in the episode; it is a linear walkthrough of an existing standard and its certification program, with the AI addendum discussion the closest thing to a fresh angle. Even that is framed observationally rather than analytically.
a year and a half, two years ago, I would basically tell you there's two types of data centers
We were a first mover on a dedicated standard for data centers
Tom McGarry is a legitimate domain expert with decades at Verizon and now running a standards development organization, giving him real institutional credibility. However, he is a standards administrator rather than an operator who has designed or run large-scale data centers, limiting the practitioner depth a B2B operator most values.
worked at most of my large part of my career at Verizon, most of that in the mobile group back in the 80s and 90s
about five years ago I joined TIA. I had always been involved in standards, but from the participant side, not from the standards development organization side
The episode earns some credit for naming concrete certification numbers (1,000+ activities, 800 unique data centers, 60+ countries, 8 certification bodies) and regulatory specifics (India's government-contracting requirement), but the rack power figures appear to confuse kilowatts and megawatts without correction, and most technical claims stay at a descriptive rather than evidenced level.
we've done over 1,000 certification activities in over 800 unique data centers in more than 60 countries
your typical rack today would have 10, maybe 15 megawatts of power. They're talking about 100 megawatt rack these days
Questions are pre-scripted, leading, and never challenged; the host allows a likely unit error (megawatts vs. kilowatts) to pass unchallenged, never digs into tensions or trade-offs, and the episode ends with an explicit recruitment pitch that goes completely unprobed. It reads as a sponsored explainer rather than a substantive interview.
Tom, I can imagine with all of the rapid advancements in technology and demand for this data, the standard will continue to grow and absorb these things
What Support is the TR942 standard body looking for to enable this advancement?
Computed from the transcript - who did the talking, and the words that came up most.
In this episode, we sit down with Tom McGarry, VP of Standards at the Telecommunications Industry Association (TIA) , to explore the ANSI/TIA-942 standard - the Telecommunications Infrastructure Standard for Data Centers. Tom walks us through what the standard actually covers: the minimum requirements for data center design, construction, and operation across telecommunications, power, physical security, and mechanical infrastructure. He explains the four resilience ratings that classify data centers by their level of redundancy and fault tolerance, and why those ratings matter to the people relying on these facilities. A major focus of the conversation is TIA's certification program. Tom describes how independent verification bodies conduct certification, what motivates data center operators to pursue it, and how the process works in practice. We dig into the downstream effects too: how certification builds trust with end customers and gives regulators confidence in a facility's reliability and security. The discussion then turns to the future.
Transcribed and scored by The B2B Podcast Index.
Speaker A: Welcome back to industry Therapy. Today on our show we are joined by Tom McGarry, Vice President of Standards at uh, Telecommunications Industry association, to Talk about the TIE942 data center infrastructure Standard.
Speaker B: This podcast is provided by the Telecommunications Industry Foundation. This podcast as well as much more additional content can be found@tifonline.org Please welcome your hosts, Michelle Kang and Scott Stecker.
Speaker A: Welcome to the show, Tom.
Speaker B: Thanks, Michelle. Appreciate it. Thanks for inviting me.
Speaker A: And since this is your first time on the show, can you tell us a little bit about your background leading up to your involvement with the standard at tia?
Speaker B: Yeah, my background is, uh, worked at most of my large part of my career at Verizon, um, most of that in the mobile group back in the 80s and 90s, um, which was sort of the Wild west, uh, which is great, great time to be involved. Um, and, uh, worked at an organization called NuStar, which provided services to telecom providers. So my background is telecom, um, dealt quite extensively at my NuStar career with data centers. And, um, and about five years ago I joined TIA. I had always been involved in standards, but from the participant side, not from the standards development organization side. So, um, the past five or six years has been, uh, my first experience, um, running a standards development organization. Been fantastic and very interesting.
Speaker A: Wow, that's excellent. So for listeners who may be new to the space, how would you describe the different types of data centers and where the tie 942 standard fits into that landscape?
Speaker B: So, you know, it's funny, a year and a half, two years ago, I would basically tell you there's two types of data centers, um, I guess what we call today traditional data centers, what you imagine as colocation data centers, colocation sites and hyperscaler data centers, um, and EDGE or modular data centers, small data centers that are closer to, um, where the data would be going. Um, but, um, very quickly, uh, AI has added a basically third type of data center, um, which is an AI data center. And it's, it's different enough that, you know, we need to now start looking at them, um, very differently. Um, the 942 standard covers sort of your general data centers. It has a very specific section on edge modular data centers. So it covers both of those and they're starting work soon, um, on addressing the AI data centers.
Speaker A: Okay, that's interesting. And TIA 942 has been around for more than two decades. What makes it such a mature and widely adopted standard in the data center industry?
Speaker B: So it started in the early 2000s. We published our first version of it in 2005. So obviously work started before then. Um and we uh, were a first mover on um, a dedicated standard for data centers. Um the uh, we also are the first and really the main voluntary consensus standard. There are some organizations out there that have what they call standards but they're private organizations creating private documents. They don't seek um, broad input like we do and we're required to as an ANSI accredited standards development organization. Um, so this really started at what I consider the first wave of data center growth back in the early 2000s was when data centers really started becoming um, a big deal and the industry decided that they needed something that um, helped people determine how to design, build data centers and then um, also how to design them to be reliable um, and have high availability. And you know that's really sort of the two things that the standard at a high level covers is how to design, build it and operate it, um and um, how to do it in such a way um, that you can maintain availability of the service, um even in times of unplanned outages.
Speaker A: Now what I think is particularly interesting about the standard is its rating system. So can you walk us through the TIA 942 rating system levels 1 through 4 and explain why these reliability classifications matter to operators and their customers?
Speaker B: Yeah, so as you mentioned there are four levels. Um, one being the, the least amount of availability and reliability, four being the most. Um one is sort of a basic design which is what you would think of as an enterprise IT room or a telecom room. Um so it has very little um redundancy. Um number two, level two is some redundant components. Um so this is you can switch over to some equipment if there is an outage or if you have planned maintenance. Um but um, and this you would consider sort of a regional um it uh room or telecom room or one that has critical applications that need some level of protection. Number three is concurrently maintainable which means that there are um redundant um equipment and multiple paths to, for power, um cooling, cabling. Um and really three allows you to do maintenance, take some equipment out, um, turn some other equipment on and do maintenance on the, the out of uh service equipment. Um and number four is multiple um paths, fully redundant. And this can withstand a unplanned outage. Three is planned outages when you want to do maintenance and four is unplanned outages. So you have a router go down, you have a power supply go down, it will automatically switch over um, to the one that's there. Um so data centers that we do certifications on, they're all looking for three and four. They don't want to be rated one or two.
Speaker A: That's very helpful. TOMS and TIA also runs a certification program with independent certification bodies. What motivates data center operators to pursue certification and how does this process work?
Speaker B: Primarily the. Let me talk a little bit about how it works. So we have agreements with certification bodies. Um, they are. One thing that in the certification world you'll realize is that in some cases certifications are done so that organizations can do contracting work for the organizations that they're. They're certifying. So they can say, hey, this is a problem. I can fix it, um, if you want me to. Um, in our program, we don't allow the certification bodies to do consulting for the data centers that they certify. So they are solely certification bodies. Um, we are considered a neutral third party. So we have eight certification bodies right now all around the world. Um, and we audit them, uh, yearly. Um, we go through approving them to become a certification body. Do they have knowledgeable auditors on staff? Do they have documentation that allows them to do a sufficient certification? Um, to. I should mention that the certification is to the four levels of, um, 942. So, um, as I mentioned earlier, the data centers want to be certified three or four. Um, so, um, they, we rely on the matrix in the, in the, uh, standard, um, to do that certification. Um, data centers want this to be able to tell their customers that they've been certified by a third party like tia, somebody who doesn't have an other interest in what they're doing at the data center, um, to run a certain level of availability. So we will put the certificate up on our website. The data center operators point to our, um, website and say, see, I'm certified. Um, so it's, it's, it's of course not just for the data center operators, but data center customers. They really want to go to their customers. And what we find is that in some cases, regulators want to see data centers with, um, certifications. Um, India's well known for this. They, they require. If you do data center business really with the government, you need to have your data center certified by a third party. Um, TIA 942 is one of those certifications. Uh, so we have a lot of certifications in India. Um, by the way, I should mention that we have, we've done over 1,000 certification activities, um, in over 800 unique data centers, uh, in more than 60 countries. Um, every Continent, but Antarctica. Um, so, uh, it's very, it's been around for a little over five years. It's, it's very successful. Um, and uh, we continue to do more every year. Um, so um, in a nutshell, that's the certification, uh, process. Great.
Speaker A: And with AI architecture's rapidly changing data center design, what updates or new considerations are being explored for future versions of the standard?
Speaker B: So back in February they just approved doing an addendum to the standard, um, which is sort of a new document that will be paired with the standard 942C. C is the current version M. Eventually the addendum will be integrated into a new version of the document when they do D, for example. Um, but um, they wanted to get to work on it pretty quickly. So they wanted to do this addendum that addresses really what is the high density compute environment now that is being driven by AI. Um, in this environment there are uh, GPUs as opposed to CPUs. Um, and the, there are much more um, GPUs than there normally is. And the racks themselves need much more power. They're talking about your typical rack today would have 10, maybe 15 megawatts of power. They're talking about 100 megawatt rack these days. Um, which is a very big change. Um, of course all of that power needs cooling. Um, and you almost need um, to cool this type of equipment. What's called liquid cooling. So um, today most cooling is done with air, air conditioning. Liquid cooling would either immerse a rack or um, um, uh, in liquid itself or provide liquid directly to the chips. So um, it's changing pretty dramatically. And there's also, um, the cabling. There are so many GPUs that the cabling is just getting exponentially larger. Um, now the important part of 942 is how to maintain redundancy. So with all of this extra power, with all of this extra um, cooling, with all of the extra cabling, um, they actually, quite frankly, what our CDs have been seeing is they haven't really been installing them in a way that would make them fully redundant and highly available. Um, so that's what they approve doing. The addendum for is to how do we address um, this new environment. I should mention also that the racks are so heavy that they potentially could impact the weight of the floor. Um, so um, that could be a very significant uh, retrofit for data centers if they have to actually strengthen their floors. Um, you know, we've sort of seen this coming for a Little while that it's such a significant change that it has to be addressed in and of itself. So we have this new addendum, they are calling it the AI Addendum, um, that will talk about this new environment and how to protect it.
Speaker A: Okay. And Tom, I can imagine with um, all of the rapid advancements in technology and demand for this data, the standard will continue to grow and absorb these things. So what Support is the TR942 standard body looking for to, to enable this advancement? And how can interested individuals become a part of this?
Speaker B: Sure. So, um, because it's so sort of focused on this um, particular issue and because power and cooling is so important in liquid cooling, certainly organizations that are involved in power and liquid cooling um, would benefit from being part of it. We would benefit from them being part of it. Um, data center operators of um who are interested in maintaining their availability um, would be important to have um, and they can reach out to tia, um, come to our website. There's plenty of contact there for the membership group. Uh, you can reach out to me. Um, you can find TR42 by the way I should mention is the committee, um, at TIA where they developed the 942 standard. So you can also look at the TR42 committee website. Um, and uh, I encourage all to, to get involved and um, let me add one more thing. We, we have another effort that started in our quality group. I oversee our technical standards. I have a peer that runs quality standards that we've had for many years. Um, and the hyperscalers, uh, have come to them and realized that with this rapid growth, um, the quality processes and quality certifications, if you're folks who are familiar with ISO9001, of course um, we do standards based on that that are focused on telecom and now data centers. So um, you can also come and look at that. That's a brand new effort that just started in the beginning of the year. Um, and you know look forward to uh, working hand in hand with those folks on the quality side of it as well as our technical standards side. So thank you Michelle. I really appreciate the time to talk to your folks and um, thanks for asking.
Speaker A: Well Tom, thank you for being on our show and sharing your expertise on the TIA942 standard as well as TIA certification program. And for those of you that are interested in looking for the standard, I Recommend going to TIA's website. Thank you for listening to Telecommunications Industry Therapy presented by the Telecommunications Industry Foundation. The views, information and opinions expressed herein do not necessarily reflect the views of tiff, its board, or the hosts. These podcasts are not meant to supersede regulations, standards, or AES requirements that govern the reference subject matter. Rather, they are offered to help educate listeners so that they may gain information that will lead them in the search for additional information and opinions. To view more TIF podcasts and other educational content, to provide feedback on an existing podcast, or to submit a potential topic for a future podcast, please Visit our website tifonline.org.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
AI in the Network, AI on the Network - with Iain Gillott, WIA85 / 100Ground Truth: Corrosion, Anchors & Asset IntegrityWhy Mount to Tower Interaction Matters More Than EverMicrowave Backhaul Site Selection: Expanding Rural Connectivity with Scott Vance & Lynn BurnettInside NATE: Collaboration, Safety, and What’s Next
