Tag: technology

Queue the Jeffersons Theme Song

easycoloquote

The pleasure is all mine being born at the beginning of the 80’s, got to dub tapes, rip mp3’s, burn cd’s, and has managed multi hundred kW to multi MW builds within one half of a lifetime. The most exciting theme I saw in 2025 was companies ‘Moving on Up.’ Not the short list of bloated commodity deals. This goes beyond the neolithic grunts of ‘me need hundreds of MW who has?’ to ‘oh my I’m going to become a consumer of my competitions offering if I don’t get cracking.’ These kinds of companies had all together a few MW’s tucked away in corners of data halls across colo providers throughout the US, but had never bought a sizable piece of the infrastructure pie. With half of the top tier of the colo market you’re going to be buying a quarter, half, or whole new site – or picking up the leftovers at a legacy building. This is the eat or be eaten new enterprise business data center segment. They are building  the colo version of nice new subdivisions not the entire city scale like around Abeline, TX which has everyone’s attention. There have been just enough data cycles to begin level one framework for success.

Start with the idea that it takes a minimum of 3 legs to build a stable stool.

  1. Find a partner from the gear side – think about the tens of millions that will be spent on AI servers, rpdus, or busway. You’re looking to land power in a data center. You’ll need to have data to understand what’s drawing power / generating heat. These will be the variables that you’ll be solving for once you’ve signed up for a larger piece of data center infrastructure than you typically take down. These timelines are long, and you’ll want to have everything in place before your data center search.
  2. Find a data center provider who’ll work as a partner. Ask your AI assistant of choice for a rank list of AI data center providers. Then look within to think where your company ranks within the overall ecosystem. The first time at a steakhouse you’ll look to order the biggest most expensive tomahawk and go home miserable. Over time you’ll look for a nicer filet with a shared side and salad that makes sense. The top data centers expect their customers to sign their legal paperwork as is and to come fully equipped to self-engineer their solution in to the site. They’re going to look at you the customer as a credit asset or liability if you’re taking on a considerable portion of their expensive asset.
  3. Talk to the leaders of your own company. They went to a conference or read a story in HBR on AI now they want that sparkling jewel. Start with the simple question ‘Are we going to invest hundreds of millions of dollars in IT gear and leases and not add headcount?’ Then ask ‘How are we going to train these people?’ There are teams who were managing colo PoP’s now with their heads spinning going from 50kW deployments to 4MW single deployments. There is a direct relationship between what companies invest in to the success of their projects to how successful their projects are. This is a no S concept that needs to be said out loud to be brought to reality.

Next run through this Base Level of Installation Considerations.

By thinking about these items early your team is able to get all of the parties involved early.

 Buildout Considerations
Power 
Power (kW)How many cabinets?  What is density per cabinet?
WhipsInclude how many per cabinet (typically 2 per cab).  This is typically defined by density..
BuswaySee the timelines for the entire solution down to the tap cans
CAGE 
Cage WallHeight needed / solid or perf wall. 
Doors# of Doors (swing vs slider, egress only) 3ft vs. 4ft.
SECURITY 
Badge/BioInclude details, data hall, office, storage, etc.
Badge ONLY (egress or entry)Include details, data hall, office, storage, etc.
Any otherunder floor mesh, proprietery access requirements.
CABLING/CONNECTIVITY 
IFC CablingInclude from each MMR, # of strands (make sure armored is standard)
Cross ConnectsQuantity what carriers are needed
Basket TrayWho is providing.  Draw layout, details and quotes.
Conduit if there’s office or storageDoes inter facility cabling need be in conduit back to MMR or Office/Storage Spaces?  What size?  Full path or only in shared spaces? Demarc in customer cage/office/storage)
Connectivity SolutionsWhat does customer need? (bandwidth, dedicated circuits, etc)
 Setup a time to discuss with data center connectivty SME
CABINETSProvide cabinet dimensions for layout.
Cabinet Setup and GroundIs customer providing white glove service? If so, just quote grounding.
Hot/Cold Aisle ContainmentDoors or full containment.  If quoting doors, need 2 sets per row and make sure height is lined up to cabinet height and aisle width.
OTHER …

Build your three legged stool, run the traps of the buildout considerations, and you’ll be walking as confidently as George Jefferson himself. Having an understanding of the entire scope will set the expectations for a successful project. It’s key to leave you with the understanding that a true AI deployment isn’t just taking down power, but assuming a block of infrastructure. The more plans your team can bring to the table the longer it’ll take for your data center provider to begin billing. If you have whip and IFC locations, and power striping models baked into the power lease they can’t start billing till the physical pieces are in place. Find a way to get your planning put into a simple flexible order that can be amended. That way the framework is built and changes can be made accordingly. Power and cooling are not exact and can be upgraded as necessary. The top standard 22/17kW whip is essentially the building block that is being doubled and tripled within cabinets. Cooling can be dumbed down to 17kW cabs need containment, 70kW cabs are in the rear door heat exchanger range, 100kW and above start talking about liquid to chip. You can get more exact and reach for the top early, but that will limit your flexibility. The result is you potentially future goof not future proof your project. Have general costs ready for the top brass’s next great conference takeaway. “Oh I see we invested $15k so you could get great insight at a conference – did you know that plan will take $150M to kickoff, then tens of millions yearly, and take additional support?” Gold beats brass, or costs are the bosses boss. You’ll have all that research time back when they come to the monetary realization of the great idea they heard. It’s an exciting time in the data center industry. It’s important to remember that every company is living a dual reality of decisions being made at a point in time within a fluid environment. That makes for a high-stakes business landscape –

Last thoughts

– require all vendors to provide useful recommendations or move on. Don’t allow people to just take orders, take your money, and move on. Keep the markets competitive.

– know that AI will become a solid service so you can decide to go the OPEX route. Think on prem data center v the cloud evolution/revolution.

– if you can’t think big picture, get caught up in the details, you’ll choke on gnats, and die. 

okay so nobody will die, but enormous amounts of time and money will be wasted.

Okay We’re Releasing the Files

easycoloquote

It has been over a year since you’ve gotten new content here. I completed an MIT. In the spirit of Sam Altman I’m opening up my computer – This is assignment 1

Question:

Consider a working environment. It could be your current organization, an organization where you have previously worked, or a fictional organization. Think about the role of technology in the organization and how it assists managers and other employees with efficiency and productivity.

Response:

My organization relies on focus. We are not the cheapest, and often times the general public cannot tell our offering from other data centers. We are focused on a particular client who is able to utilize their data center as an asset for their company. We are focused on meeting the power requirements of what is being called data factories where businesses are finding value for their shareholders by having superior data. We also have a focus on interconnection that provides direct transmission of data for advantages in finance and other high speed interconnected processes.

We can leverage AI in the future at the data hall, campus, and global levels of our organization. In data halls we can preventively see where the supporting power infrastructure is about to go out and have solutions ready. At one of our large campuses our monthly report takes a few minutes longer every month because the data points are expanding by that much. We could use AI to toss any useless data to optimize this process and improve the overall data set. At a global level we can assure that investments that are in the hundreds of millions of dollars are made in the correct markets at the best times so that when customers come up with large requirements they can be filled on time and with a good profit. It is interesting to think that a company where AI is being stood up has so much potential to improve through the use of AI.

My company can gain differentiation and cost leadership through focus. Every property we build out is a gamble, every piece of machinery, every customer we sign. They are all opportunity costs of potentially better places we could have built, machinery that meets the needs of more clients in a more profitable way, and passing on a customer who isn’t going to grow for another one who ends up expanding and bringing tons of revenue. By being focused on optimizing decision making our company can save a lot of useless cycles of work that goes no where and focus on profits. With that we’d be able to pass some cost savings on to customers. The time savings would be a differentiator in speed to market and availability. Focus on who to serve, where, and how would be a clear strategic win in the data center industry.

AI could bring all of the data points from the data hall to the global footprint to optimize our company effectiveness. The data could be taken from the many different disciplines like, accounting, finance, engineering, and sales to make everything make sense for all parties. We could replace automated tasks with automation, avoid potential break down issues, plan for the future, and there are ways to improve in our industry that haven’t even been imagined yet.

The real magic will be the solutions that haven’t yet been imagined. It’s pretty cool to think a fundamental platform using a revolutionary tool for clients also doing the same to mee their own goals.

Stay tuned there are a handful of these to come….

New Clear Power Path

easycoloquote

In early 2023 we were hearing about the new power demands that will come as a result of the processes needed to make AI/machine learning work. Not much changed until Spring when news of AI companies gobbling up entire data centers to meet their requirements. Some of the 100+ MW asks almost seemed like a fantasy.

It seems like only a few years ago onsite substations became common, and everyone had signed their promises to be carbon neutral by 2030. Providers were left with a new twist to the problem to not only go green but also go bigger than ever imagined.

Why aren’t large data centers just covered in solar panels to get a green source of energy? It’s because solar just isn’t a robust enough point at the moment. Simple math says that one can achieve about 100kW per 100,000 sqft of roof-space. Furthermore, AI data centers are going to be shrinking their physical space as every 3-5 cabinets pull up to a MW of power. We could be looking at a 50,000 sqft building pulling around 50MW!

Other great alternatives while we wait for solar and wind to evolve are SMR’s, onsite natural gas generation, and even hydrogen. These are all great choice with one immediate issue… they’re not immediate. Think for a second how long licensing for your own small nuclear reactor or gas/hydrogen plant might be if a simple substation is held up for 3-5years because of transformer shortages and permitting.

The industry is going to see data center providers connecting to already built gas plants and nuclear reactors behind the meter to meet the veracious power demands of AI companies desperately needing power. This will allow speed to market. Also, data center providers won’t have their scope of responsibility doubled by now running a power plant and a data center in tandem. Attaching to one or two existing plants will bring 300 MW data centers online by Q3 of this year if not early 2025. By this time we’ll see AI evolve from a novel co-pilot / assistant closer to the more mature product that it was intended to be.

back to easycoloquote.net home page

DIY PDU Power Ramp

easycoloquote

Okay so finding a data center isn’t the most difficult thing. You can just go to sites like https://baxtel.com/map or https://www.megaport.com/megaport-enabled-locations/ that have great global maps.

Then when it’s time to move in you could be faced with challenging ramp options. Power might not be available right off, or it might make financial sense to move in over time while as your migrating equipment. Rpdu’s come in block options of basically 5, 8.5, and 17kW. In many ways 17kW has become the new standard to support high powered servers and chips. It can be difficult in this form factor to not blow through a lot of distributed power in just a few cabinets.

A potential solution that we’ve found is leveraging extra deep 1200mm cabs and planning for a quad pdu setup. This way you can populate more cabinets broadly and dense them up over time. This is also a way to hedge your bets if it turns out that power requirements actually need to be tweaked.

In this case you could have options like this.

L21-30’s at 8.5kW starting – adding L6-30’s at 5kW for a total of 13.5kW. L21-30’s at 8.5kW starting – adding a 2nd set of L21-30’s 17kW (new standard) L21-30’s at 8.5kW starting – adding 460C9W’s at 17kW for 25.5kW (high density).

You get the picture. Rightsizing is the ultimate goal to not need to over commit to the facility while still having ample power. I know what you’re thinking, that doubles my rpdu budget, but that can be made up in a couple months of power savings. Whenever you walk through a datahall and see rack after rack lightly populated you’ll know their options were constrained and as much as they tried they not only got to buy another set of rpdu’s but another cab too.

It’s also a good idea to look into universal pdu options if you’re worried about managing rpdu plug types in a complex deployment.

We’re here to talk you through options in designing your cage power infrastructure.

back to easycoloquote.net home page