@Bitdeer has executed a 16-year #AIDC# colocation lease for Tydal, Norway 🇳🇴 Campus with Volta, an @nvidia Cloud Partner. 🤝 ~$4.7B in contracted base-term revenue, with the potential to reach $8B over 24 years. ⚡️ 121 IT MW configured to run #NVIDIA# GPUs for a leading AI lab, with @Dell as the #technology# provider. ♻️ 100% renewable energy with ~1.1 PUE, TDC expected to be among Norway's most efficient #AIInfrastructure# upon completion. 🛡️ Anticipated credit backstop totaling ~$1.3B, arranged via Letter of Credit by affiliates of @jpmorgan and another top-tier global financial institution. Read the full details here: $BTDR #Bitdeer# #datacenter# #gpu#

Crusoe just raised $4B at a $30B valuation It has 1 GW operational, $140B of contracted backlog, and 6 GW of contracted capacity They do mostly colocation today, but now they want to go vertical There may be a company with very similar potential at half the price Just saying

The year is 20XX. Everyone trades Hyperliquid perps at API levels of perfection. Because of this, the winner of a trade depends solely on RPC ping. The server colocation metagame has evolved to ridiculous levels due to it being the only remaining factor to decide fills. DeFi has reached its pinnacle. The manual UI clickers are living in poverty. It seems nothing can stop the great leader of 20XX, Jeff, and his army, the Quant monks who live in great server farms where they levitate while optimizing Rust with one hand, and farming points with the other. The orderbook metagame has gotten to this point where every arbitrage is played out to theoretical perfection, so market makers play Rock Paper Scissors for validator proximity, and that’s the market.

This chart is why secured power is becoming one of the most valuable assets in AI. A MW produces ~$1.5M in bitcoin mining, ~$4M in colocation and ~$10M in full-stack AI compute which explains why you're seeing $CIFR, $WULF, $CORZ and $RIOT all racing to convert capacity as fast as possible while $IREN already shows what that transition can look like. $CRWV and $NBIS are already clustering ~$10M per active MW while $SPCX is targeting closer to ~$40M as it scales toward ~10GW putting SpaceX in a monetization class of its own.

ROTHSCHILD REDBURN LAUNCHES COVERAGE ACROSS AI DATA CENTER NAMES $NBIS: SELL, $84 PT Says Nebius has a demanding valuation and questions the sustainability of its unit economics. Upside could come from newer-generation GPUs, faster growth of its Token Factory inference business and stronger capacity execution. Risks include customers bringing compute in-house and higher funding costs. $CRWV: SELL, $54 PT Redburn questions CoreWeave's unit economics and ability to convert its pipeline into attractive returns. Key risks include hyperscalers or AI labs bringing capacity in-house, falling GPU pricing and higher financing costs. $DLR: BUY, $227 PT Highlights Digital Realty's global wholesale and colocation footprint, long-term tenant contracts and expansion into larger, higher-density facilities built for AI workloads and hyperscale customers. $EQIX: BUY, $1,261 PT Points to Equinix's global interconnection ecosystem as a key advantage, with AI, HPC, enterprise and cloud customers increasingly needing high-density compute close to networks and other infrastructure. $IRM: BUY, $132 PT Iron Mountain has expanded beyond its legacy records-storage business into data centers, information management and IT asset lifecycle services, while developing more power-dense capacity for AI workloads. The miner-to-AI names were treated much more cautiously: $APLD: NEUTRAL, $22 PT Says unit economics and pipeline-conversion risks are already well priced in. Upside comes from lower build/operating costs and additional large tenant signings. Risks include local opposition, construction delays and tenant insolvency. $IREN: NEUTRAL, $40 PT Sees upside from additional large-scale AI tenants and better execution at existing sites. Risks include difficulty securing tenants, higher funding costs and delays or cancellations. $WULF: NEUTRAL, $15 PT Potential upside comes from securing additional U.S. capacity and signing more major tenants. Permitting issues and state-level opposition are key risks. $CIFR: NEUTRAL, $18 PT Sees better or faster grid-capacity allocation and more major tenant deals as upside. Tenant pullouts and buildout delays are the main downside risks. $HUT: NEUTRAL, $96 PT Upside depends on faster data-center execution and securing additional grid capacity. Delays or cancellations in the buildout remain the key risk. $CORZ: NEUTRAL, $16 PT Core Scientific has been shifting capacity from Bitcoin mining toward AI/HPC hosting after emerging from bankruptcy, while continuing its mining business. Its proposed acquisition by CoreWeave collapsed last year after shareholders rejected the deal.

Digital Infrastructure World (DIW) will take place February 3–4, 2027, in San Jose, California. As an OCP co-located event, DIW will unite semiconductor and systems companies, networking, power and cooling innovators, hyperscalers, cloud and enterprise leaders, colocation providers, EPCs, utilities, architects, engineers, and data center operators in one collaborative forum designed to accelerate the future of AI infrastructure. More than another industry event, DIW is an opportunity to extend the collaboration that defines OCP by bringing together every organization shaping AI infrastructure—from the chip to the grid. We encourage every OCP member organization to get involved: -Speak – Share your technical expertise, lessons learned, and vision through conference sessions and panels. -Sponsor or Exhibit – Showcase your innovations while connecting with senior decision-makers from across the digital infrastructure ecosystem. *OCP Member discount available* -Attend – Collaborate with peers spanning silicon, systems, networking, power, cooling, cloud, and data centers. *OCP Member discount available* Your participation is more than supporting an event—it's extending the impact of the Open Compute Project and ensuring the principles of openness, collaboration, and innovation continue to guide the future of AI infrastructure. Call for presentations ends today, August 21, 2026! Learn more, register, explore sponsorships and submit here:

In a 2023 podcast, Dwarkesh asked me if HFT was zero-sum for society. I argued that extreme competition at the margins for nanosecond-level advantage might subsidize the creation of broadly beneficial new technology. How AI training/inference stacks have become a clear example of this Over the last ten years, quantitative trading firms built or invested in the fundamental components of distributed systems and high-performance computing. Many technologies now commonly associated with AI were created for HFT during this period or reached scale because of it. A non-exhaustive list of these technologies, from high-throughput networking to vectorized databases to precision timestamping: • Mellanox InfiniBand and ConnectX NICs: HFT colocation and cross-connect environments drove early large-scale deployment of RDMA-capable InfiniBand and the ConnectX series. These became the default high-bandwidth, low-latency fabric inside modern GPU pods for all-reduce, parameter servers, and expert parallelism. Mellanox was acquired by Nvidia in 2020. • Solarflare/OpenOnload (and later Xilinx/AMD equivalents): Kernel-bypass networking stacks and specialized low-latency Ethernet NICs let applications DMA packets straight into user space. These techniques were required for software-based trading systems to reach the microsecond range of tick-to-trade latency. Engineers for hyperscalers commonly use this technology today for distributed cloud training and inference networks. • RDMA and RoCE: Remote Direct Memory Access, refined under HFT firms’ nanosecond requirements, has become the transport that allows GPUs to exchange gradients and activations with microsecond-scale latency and minimal CPU involvement. • FPGA feed handlers and inline accelerators (Alveo, custom RTL on Xilinx/Intel): FPGAs are not the hardware solution of choice for AI, where GPUs and custom ASICs have dominated. However, HFT use of FPGAs for wire-speed market-data parsing, book building, and sub-microsecond decision logic created a pipeline of talent and verification processes that transferred into AI inference accelerators. • Vectorized/columnar analytics engines: Quant and HFT shops require extreme scan and aggregation performance over massive tick and order-book histories, using pioneering technologies like KDB/Q. Several firms I’ve worked at became large users and funders of ClickHouse, which has become essential to logging and offline data preparation for large model training. • Hardware timestamping and PTP: HFT firms’ need for exact packet ordering and latency measurement hardened the timing infrastructure that distributed AI training and multi-node inference rely on for correct collective operations and debugging. • Lock-free, zero-allocation, cache-aware software patterns: The performance culture and specific techniques developed for tick-to-trade loops became standard practice in high-performance CUDA kernels, inference servers, and communication libraries. These techniques include carefully tuned concurrent data structures, huge pages, CPU pinning, and avoidance of syscalls in the critical path. The natural followup question in 2023 was whether too much capital and talent had been or could be spent on HFT. The HFT industry is still relatively small, employing around 10,000 people and producing about as much yearly revenue as Nvidia’s and AMD’s R&D budgets. If an ancillary use case for HFT is a bounty program to fund the infrastructure of our next technological epoch, the costs still seem low.

$BTDR Q2 2026 Financial & Operational Highlights 📊 🔹Total revenue $228.8M (+47% Y/Y); Adjusted EBITDA $31.1M (+576% Y/Y) 🔹Self-mining #hashrate# 73 EH/s (+342% Y/Y) across 243,000 #miningrigs# (+113% Y/Y); 2,694 $BTC mined (+377% Y/Y). 🔹3 GW global power portfolio for #AIDC# and #BitcoinMining#; Executed a $4.7B, 16-year #AI# colocation lease in Tydal, Norway. #Bitdeer# #AIInfrastructure# #datacenter# #Investors#

I'll make this super clear for people wondering if $DGXX or $SLNH is more asymmetric: They serve two completely different purposes, in different layers of the same supercycle. Both genuinely asymmetric in their own way. Both sit in the Neocloud ecosystem. $DGXX as a GPU-as-a-Service operator and $SLNH as the renewable powered data center beneath it. Different theses, different risks, same tailwind. $DGXX (~$600M MC) - GPU-as-a-Service operator deploying $NVDA Blackwell GPUs directly to customers. Initially shared at ~$4 (up 105%+ now). > Similar model as $CRWV (~$60B MC), $NBIS (~$45B), $IREN (~$20B). First AI revenue contract signed. $1.1B $CBRS colocation deal. Hans Vestberg / $BLK connection. > 1.9% institutional ownership leaves massive room for re-rating. Earnings tomorrow, GPU rental starts on Friday. Risks: Early stage, $750M shelf filed (dilution capacity), negative margins, execution heavy. $SLNH (~$250M MC) - Renewable powered AI data centers. Wind farm acquisition closes vertical integration loop. Initially shared at ~$1 (up 65% so far). > Same renewable power thesis as $TLN (~$17B), $CEG (~$106B), $VST (~$50B). 4.3GW development pipeline. Difference between them is instead of wind farm → grid → data center, $SLNH does wind farm → data center. > Dorothy campus operational and expanding. Nasdaq compliance just regained. Earnings May 19. Risks: Overhang from active dilution. Cash burning. Execution risk on Dorothy 3 (300MW+ campus). Both are very early stage at this point. Both have execution risk. But both have real catalysts incoming. As for dilution, that's a risk with any early stage company. Again, bears were saying the same thing about $PLTR at ~$15. Now the same bears would full-port if it ever dips to $100. Valuation gap between current MC and what their competitors are trading at is what makes both asymmetric in their own layers.