AI Compute Deals Meet Harder Science-Model Benchmarks
Monday, August 17, 2026 · 10 min
Microsoft and Nebius are locking down fresh AI cloud capacity while Nature papers stress-test protein and materials models; the through-line is practical infrastructure, from megawatts and UK campuses to benchmarks that ask whether scientific ML holds up off the happy path.
Listen
Show notes
Microsoft and Nebius are locking down fresh AI cloud capacity while Nature papers stress-test protein and materials models; the through-line is practical infrastructure, from megawatts and UK campuses to benchmarks that ask whether scientific ML holds up off the happy path.
In this episode
- What Does Microsoft (MSFT) Gaining 50MW Of AI Cloud ... — Simply Wall St
What Does Microsoft (MSFT) Gaining 50MW Of AI Cloud Capacity Mean? # What Does Microsoft (MSFT) Gaining 50MW Of AI Cloud Capacity Mean? Bailey Pemberton Fri, August 14, 2026 at 11:11 AM MDT 2 min read - MSFT - -0.30% ## Trading disclosure - Microsoft (NasdaqGS:MSFT) has accepted the first 50MW of AI cloud capacity from IREN at the Childress campus under a multibillion dollar contract.…
- Nebius to deploy AI capacity at Vantage's Newport campus in first South Wales AI Growth Zone commitment - Capacity — Capacity
Nebius to deploy AI capacity at Vantage's Newport campus in first South Wales AI Growth Zone commitment - Capacity ## Nebius to deploy AI capacity at Vantage’s Newport campus in first South Wales AI Growth Zone commitment 14 August 2026 3 minutes Nebius has agreed to lease high-density, Nvidia-powered capacity at Vantage Data Centers' CWL1 campus in Newport. Senior Content and Insights…
- Aligning protein-generative models to experimental fitness with ProteinDPO | Nature Methods — Nature Methods
Aligning protein-generative models to experimental fitness with ProteinDPO | Nature Methods ## Abstract Biological generative models can predict biological functions without task-specific training data but often under-perform specialized models. This is due to a fundamental ‘alignment gap’, where the rules learned during unsupervised training are not related to the function of interest. Here we…
- MatUQ: a benchmark for uncertainty-aware out-of-distribution materials property prediction with graph neural networks | npj Computational Materials — npj Computational Materials
MatUQ: a benchmark for uncertainty-aware out-of-distribution materials property prediction with graph neural networks | npj Computational Materials ## Abstract Reliable uncertainty quantification (UQ) for graph neural networks (GNNs) under out-of-distribution (OOD) shifts remains insufficiently characterized in materials discovery. Existing benchmarks based on random splits can overestimate…
- Vibrational power spectra as a tool to benchmark universal machine-learning interatomic potentials for molecular systems: the OMOL-1k-MD data set | npj Computational Materials — npj Computational Materials
Vibrational power spectra as a tool to benchmark universal machine-learning interatomic potentials for molecular systems: the OMOL-1k-MD data set | npj Computational Materials ## Abstract We present OMOL-1k-MD, a new dataset for the benchmark and training of universal machine-learning interatomic potentials (uMLIPs). It contains three independent ab initio molecular dynamics (AIMD) trajectories…