Excited about (i) building datacenters alongside carbon-efficient energy sources, (ii) developing specialized AI models that deliver capability with fewer calculations, (iii) managing computational jobs to match energy cost. #EnergyEfficiency #AI #DataCenters #Sustainability.

Thrilled to co-host a recent workshop with Arthur van Benthem at @Wharton. We convened an exceptional, multi-disciplinary group to tackle these issues. While challenges are significant, opportunities for innovation are immense.

Excited to contribute to @engadget's coverage of @OpenAI's open-weight LLMs, their first in six years. Sharing learned weights enables developers to use and fine-tune models without needing hyperscale infrastructure, lowering barriers to innovation.

Great to see David Irwin's insights too! We agree on opportunities and challenges. #AI #Datacenters #CarbonAwareComputing #EnergyEfficiency #GreenAI #ComputerScience.

* Multi-agent game theory, in which users and their jobs, make decentralized decisions about power while anticipating competitive dynamics is an interesting part of the solution.

* Datacenters allocate shared power to users within large organization, creating a coordination problem as each user tries to avoid performance losses and optimize energy use.

* Batch workloads, such as model training and data analytics, are ideal for carbon-aware scheduling because they are flexible and not time-sensitive.

I shared thoughts with CACM on how smarter job scheduling can reduce AI datacenter carbon.

Thanks to WHYY Studio 2 for hosting this important conversation! . #AIDataCenters #SustainableAI #EnergyEfficiency #ComputerArchitecture #WHYY #AI #Infrastructure.

* Social Impacts — Local communities face complex trade-offs. While construction creates jobs, long-term employment may be modest because datacenter operations are automated. Impacts on electricity rates depend on how grids and datacenters build energy infrastructure.

* Energy Constraints — Securing and efficiently using electricity will define the state of play for AI infrastructure for the foreseeable future.

* Infrastructure for Exploration — Datacenter builds are driven not just by commercial demand for today’s uses of AI but also by exploratory research in future capabilities and applications.

It was a pleasure joining @cherrigregg and @Avi_WA, hosts of @whyy Studio 2, and @CadeMetz of the New York Times on a podcast "Should we welcome AI data centers?" We discussed unprecedented growth in AI infrastructure. A few key insights.

Thanks to WHYY Studio 2 for hosting this important conversation! . #AIDataCenters #SustainableAI #EnergyEfficiency #ComputerArchitecture #WHYY #AI #Infrastructure.

* Social Impacts — Local communities face complex trade-offs. While construction creates jobs, long-term employment may be modest because datacenter operations are automated. Impacts on electricity rates depend on how grids and datacenters build energy infrastructure.

* Energy Constraints — Securing and efficiently using electricity will define the state of play for AI infrastructure for the foreseeable future.

* Infrastructure for Exploration — Datacenter builds are driven not just by commercial demand for today’s uses of AI but also by exploratory research in future capabilities and applications.

Quoted in @ITBrew’s piece on PA datacenters. While construction are tech jobs are positive, datacenter operations are highly automated and the jobs picture is complex. #datacenters #AI #jobcreation #economicdevelopment #techjobs #Pennsylvania.

Two strategies merit further study. Deploy many smaller datacenters for inference tasks. Colocate datacenters with natural gas and nuclear plants. We need to coordinate both compute and energy infrastructure. #AI #Datacenters #Energy #GPUs #Sustainability #TechPolicy.

I spoke with CBS News about energy for AI and why Pennsylvania is at the center of it. AI datacenters are scaling with AI models, drawing gigawatts of power and putting pressure on energy infrastructure.