07/27/2025 What I Read Last Week
#DigitalInfrastructure news: Crusoe, Aligned, Amazon, Armada, Cologix, Inference, resilient communications, Quantum Computing
Weekly Edition of curated news about Digital Infrastructure
[Link] Crusoe announced a strategic partnership with Tallgrass to develop a 1.8 GW AI data center campus located in southeast Wyoming. The company said it is designed to scale up to 10 GW - but I think it should go to 11. Crusoe says the campus will leverage a diverse and robust power strategy, integrating multiple energy sources fueled by natural gas and future renewable energy developments in the region.
[Link] Aligned Data Centers announced an expansion in Ohio with a new mega-scale AI campus. As its third site in Ohio, this is a new multi-building campus has already secured a foundational customer for its first data center, targeting initial capacity delivery mid 2026.
[Link] Amazon was named as the company behind 290-acre data center campus in Tuscon, Arizona. Amazon Web Services (AWS) has been named as the end customer for project Blue, a large campus with 3 buildings initially, and some reports suggest up to 10 buildings totaling 2 million sq ft and 600MW are planned.
[Link] Edge hyperscaler Armada announced a $131 million strategic funding round with new strategic investors including Pinegrove, Veriten, and Glade Brook, as well as participation from existing investors including Founders Fund, Lux Capital, Shield Capital, 8090 Industries, M12 (Microsoft's Venture Fund), Overmatch, Silent Ventures, Felicis, and Marlinspike.
[Link] Cologix announced that it has closed a $525 million asset-backed securitization (ABS) to support the company’s continued growth in digital infrastructure. Proceeds will support strategic investments across Cologix’s Scalelogix and Digital Edge portfolio, enabling the company to address increasing demand for inference AI and cloud workloads.
[Link] Canadian infrastructure company 5C announced it has successfully secured $835 million of capital to fuel its rapid growth. The funding includes equity financing led by Brookfield Asset Management and debt financing led by Deutsche Bank.
[Link] From the White House: Accelerating Federal Permitting of Data Center Infrastucture. President Trump has signed an executive order to rapidly accelerate the buildout of large-scale data center infrastructure in the U.S., with a special focus on supporting artificial intelligence development. The order aims to cut federal permitting delays, offer new financial incentives (like tax breaks, grants, and loans), and make federal lands—including Brownfield and Superfund sites—available for data center construction.
[Link] Oklo and Liberty Energy, an energy services provider, have formed a strategic alliance to deliver integrated power solutions for large-scale, high-demand sectors such as data centers, industrial facilities, and utility-scale sites. Their plan combines Liberty’s Forte natural gas generation and load management solutions to meet immediate power needs, with a pathway to integrating Oklo’s zero-carbon, advanced nuclear powerhouses as they come online.
[Link] Cerebras Partners with Hugging Face, DataRobot, Docker to bring World’s Fastest Inference to AI Developers and Agents. I think this story is important, in that it reveals what comes next after everyone has touted the importance of AI and we have built 50 more gigawatt-scale data center campuses to power everything. Inference as an AI accelerator is next, and I think this press release from Cerebras showcases an impactful way to show how powerful it can be. I also watch the company Groq. I’ll use my preferred AI (Perplexity) to compare the two companies:
Cerebras and Groq are both highly specialized AI hardware companies, but their architectures, performance profiles, and use cases differ significantly, making them comparable within the AI accelerator landscape yet optimized for distinct requirements.Groq focuses on a deterministic, ultra-low-latency architecture ideal for real-time AI applications. Its Tensor Streaming Processor (TSP) is designed for predictable latency and throughput, making it attractive for edge deployments and tasks needing tight response times. Groq tends to have lower upfront and operational costs due to its smaller hardware footprint and energy efficiency.
Cerebras employs a radically different approach with its Wafer-Scale Engine (WSE), the world’s largest AI chip. This design is geared for extreme parallelism and massive compute capability—ideal for large-scale model training and inference on massive neural networks needing high memory bandwidth and capacity. However, this scale comes with higher cost and power consumption, best suited to data centers or organizations with significant infrastructure budgets.
[Link] It’s funny how there is always a call to look closer at ‘resilient communications’…. after an outage. SpaceX’s Starlink constellation experienced a ~2.5hr network outage last Thursday. This outage was a day after the T-Mobile satellite service with Starlink entered commercial service. Meanwhile… The Economist has an article about How US Space Command is preparing for satellite on satellite combat.
[Link] In a recent Light Reading podcast Jeff Baumgartner talks with BSP, a technical due diligence company about the surge in the data center market, liquid cooling, and the big challenges faced by existing data centers amid rising demands for AI workloads, power, and cooling.
🤯 [Link] Paving the way for Integrated Photonic chips: quantum dot lasers achieve breakthrough on-chip integration and longevity. Researchers are advancing the development of integrated photonic chips by successfully monolithically integrating indium arsenide quantum dot lasers directly onto silicon chiplets, overcoming a key hurdle for processing light signals on a single platform. This innovation could lead to faster, more energy-efficient data processing and communications, paving the way for next-generation computing technologies.
[Link] Infleqtion announced it will build the first utility-scale, neutral atom quantum computer in Illinois. In collaboration with the Illinois Quantum and Microelectronics Park (IQMP) and the National Quantum Algorithms Center (NQAC), the initiative represents an expected $50M public-private partnership over the next four years and will culminate in one of the most advanced quantum platforms in the world.
[Link] Quantum Insider: Researchers at Harvard have developed a new approach to quantum computing using metasurfaces, which are ultra-thin, nanostructured layers that can control light. These metasurfaces are able to do the same complex quantum tasks—like creating and managing entangled photons—that traditionally require bulky and fragile setups full of lenses, mirrors, and other optical parts. The research was published in Science and funded by the Air Force Office of Scientific Research (AFOSR).
Follow me on X for more links and news.
Coinbase is the easiest (and best) place to buy and sell cryptocurrency. Use my referral link to sign up.