Happy Wednesday all, and welcome to Data Center Digest. 🎃
We’re covering the business and technology of data centers and digital infrastructure.

Here’s what we’re looking at today:

• $7 billion announced for new hydrogen hub projects, per DOE

• Synergy reports hyperscale capacity to triple over the next 6 years

• Whats More: QTS announces behemoth in New Albany, BrainBox and Trane team up, NVIDIA and Foxconn building AI factories everywhere

• Hot Links: Data Centre Club, great mission critical listen

Est. read time: 5 mins 22 secs

DOE announces selected projects for new $7 billion “H2Hubs” program

Part of a larger $8 billion hydrogen hub program, the H2Hubs will be the basis of a national clean hydrogen network.

Hydrogen - when and why?

Hydrogen was first used as fuel in 1839 by Sir William Grove who invented the first fuel cell.

No one did anything with it until the 1930s, which saw the rise of large-scale hydrogen production through the steam reforming process, which is still one of the primary methods of hydrogen production today.

The tech stayed mostly dormant until the mid-20th century when NASA used hydrogen fuel cells for its Apollo missions. (That did not go well.)

Because it’s really hard to store and transport, and historically has been too expensive to produce, the technology has never had a chance at mass adoption. 
As technology advances through developments such as the development of PEM (Proton Exchange Membrane) fuel cells, hydrogen has become a more promising alt-energy option.

Here’s a little something if you want to get smarter on this:

H2Hubs

The H2Hubs are individual projects aiming to establish interconnected systems where hydrogen is produced, used, and supported by local infrastructure.
Their objective is to facilitate the growth of clean hydrogen production in line with rising regional demands, acting as a blueprint for large-scale, economically sustainable hydrogen ecosystems.

The vision of the DOE is to use these hubs to transition away from carbon-heavy processes, innovating within the industrial sector and ultimately fulfilling the nation’s clean energy targets.

Their goal is to achieve a carbon-free electrical grid by 2035, and net-zero carbon by 2050.

The regions

Here’s a look at the individual projects that represent seven regions of the country:

• Appalachian Regional Clean Hydrogen Hub (ARCH2)

• (California): Alliance for Renewable Clean Hydrogen Energy Systems (ARCHES) 

• Gulf Coast: HyVelocity H2Hub

• Heartland Hydrogen Hub (HH2H)

• Mid-Atlantic Clean Hydrogen Hub (MACH2) (🥇 Winner: Best Name)

• Midwest Alliance for Clean Hydrogen (MachH2)

• PNW H2

To read more about each of these regions, how they're generating their hydrogen power, and what each project looks like, see DCF’s great article HERE.

Synergy says hyperscale capacity to triple in 6 years

Hyperscale data center giants like Google and Amazon are set to triple their capacities in just six years, and it's all thanks to the surging demand for generative AI, reveals a report by Synergy Research Group

AI’s huge footprint

The tech isn't just making data centers pop up more frequently; it's super-sizing them.

By 2028, expect the total capacity of all hyperscale data centers to grow nearly threefold.
By the start of 2024, global data giants will operate a whopping 926 major data centers, with at least 427 more in the pipeline.

Plus, with the rise of GPUs, driven largely by AI, the power density in data centers will need a significant boost. That's leading operators to re-evaluate their architectural plans.
(Earlier this year Facebook had to completely reimagine its data infrastructure plan due to the sudden rise in AI.)

Too costly for enterprise

Companies are increasingly interested in leveraging generative AI for business growth. However, the steep initial investment in specialized hardware, like Nvidia's DGX servers—which can cost as much as ten regular servers—makes it difficult to justify.

Additionally, the ongoing operational expenses can't be ignored. Nvidia GPUs, known for their high power consumption, are among the most energy-intensive components in data centers.
This combination of hefty upfront and running costs is prohibitive for most small and mid-size companies.

Synergy’s John Dinsdale remarks that enterprises theoretically could manage this independently, but prohibitive costs and expertise gaps exist.

Flywheel of growth

The emerging solution? Farm out AI training to big players and manage the lighter inference in-house. Hence, "AI as a Service" is becoming the new normal, with more expected from hyperscalars like AWS and Microsoft soon.

Hence: many more gigantic data centers run by the handful of operators that have the capacity and expertise.

- What’s More -

Big Deals 🤝

NVIDIA, Foxconn partner to build AI factories worldwide

These AI factories are specialized data centers designed to process vast amounts of data, converting it into AI models.
Nvidia's CEO, Jensen Huang, emphasized that these AI factories will use Nvidia's GH200 Grace Hopper Superchip and AI enterprise software.

Foxconn plans to produce systems using Nvidia's tech suite for global clients aiming to set up their own AI factories. These systems will aid in accelerating autonomous machine training, including industrial robots and autonomous vehicles.

Additionally, Foxconn is considering integrating Nvidia's technology into its own manufacturing plants, foreseeing significant operational efficiency enhancements, time savings, and cost reductions through simulation and automation.

“An AI factory with the Nvidia stack can give Foxconn the ability to accomplish AI training and inference, enhance factory workflows, and run simulations in the virtual world before deployment in the physical world,” the companies said.

 Read more here

QTS stamps a gigantic footprint into Ohio

Blackstone-owned QTS is set to develop data centers in New Albany, Ohio, spanning over 1 million sq ft with a capacity of 200MW.

The New Albany City Council approved agreements allowing QTS to invest approximately $1.5 billion in four data centers. These will be situated on a 57-acre property and a neighboring 37-acre site. Construction is anticipated to start in March 2024, concluding by March 2025.

Sara Zeigler, economic development manager for New Albany, “QTS has a great track record of being invested in the communities where they have built facilities. The company’s core values align closely with ours, and we’re excited that they are choosing to call New Albany home.”

 QTS recently acquired the combined 93-acre land for $11.2 million.

Read more here

BrainBox AI and Trane team up to superpower HVAC systems

BrainBox AI, a leader in autonomous decarbonization, is partnering with global commercial heating and air giant, Trane.

The collaboration combines Trane's cloud-based building automation system with BrainBox’s decarbonizing technology, aiming to reduce greenhouse gas emissions and save operational costs for Trane customers.

The primary focus of the partnership will be on HVAC systems, which account for 40% of a building’s energy use, 30% of which is wasted

The partnership will introduce technology to enhance energy efficiency, especially for multi-site retailers in North America.
Trane hopes to achieve AI-driven, real-time HVAC system adjustments, considering weather, occupancy patterns, and more.

Hot Links 🌭

Thanks a lot for reading!

Please let us know how we’re doing by replying directly to this email. And if you want to help grow this, share this with someone who loves data centers.

- Taylor