edge compute
AI computation performed on distributed local devices (cars, robots) rather than concentrated in data centers, allowing power draw to be spread across the grid and charged at night.
As defined in Elon Musk - "In 36 months, the cheapest place to put AI will be space”
Where it comes up · 3
- Pioneering Digital Infrastructure4 claims
- Elon Musk - "In 36 months, the cheapest place to put AI will be space”2 claims
- Josh Wolfe and Tony Thomas - The Past, Present, and Future of Defense – [Invest Like the Best, EP. 216]2 claims
Most important claims · 8 of 8Search all claims for “edge compute” →
- Cloud availability zones form four concentric tiers defined by power density: large core zones above 5 megawatts, mid-range hyperscale at 1 to 5 megawatts, edge enterprise at 250 to 500 kilowatts, and micro edge sitting in tower bases or office basements, moving closer to end users.Pioneering Digital Infrastructure
- Edge compute is not power-constrained the way concentrated clusters are, because distributing AI over a large area (e.g. Tesla's AI5 in Optimus and cars) lets you charge at night, tapping the incremental ~500 gigawatts of US capacity that goes unused since average usage is 500GW while peak production exceeds 1,000GW.Elon Musk - "In 36 months, the cheapest place to put AI will be space”
- Electrical output outside of China is roughly flat while chip output grows exponentially, so the binding constraint for concentrated ground-based AI compute is power generation; towards the end of this year chip production will outpace the ability to turn chips on, and chips will pile up unused.Elon Musk - "In 36 months, the cheapest place to put AI will be space”
- Across maritime, land, air, and space, the default question should be why a platform cannot be unmanned, since unmanned systems avoid the pressure on individual pilots and the human element is increasingly a liability.Josh Wolfe and Tony Thomas - The Past, Present, and Future of Defense – [Invest Like the Best, EP. 216]
- Edge and core assets complement each other because of data gravity: cloud players need availability zones in the periphery, so wholesale multi-location edge leasing will grow as hyperscalers build coverage into suburbs and tertiary markets, just as mobile operators did in the 1990s.Pioneering Digital Infrastructure
- Autonomous systems require AI and compute at the edge so they can function when not tethered to elaborate command and control structures, which Tony believes will not be reliably available in future conflicts.Josh Wolfe and Tony Thomas - The Past, Present, and Future of Defense – [Invest Like the Best, EP. 216]
- Southeast Asia's four largest countries, Malaysia, Thailand, the Philippines, and Indonesia, have about 380 million people, and DigitalBridge's Edge Point investment grew to over 18,000 towers in about two and a half years, beating other funds by a year and a half.Pioneering Digital Infrastructure
- Rather than slowing after the COVID pull-forward, DigitalBridge is seeing acceleration in cloud player demand for network and compute, especially fiber connecting data centers and wholesale edge leasing.Pioneering Digital Infrastructure