OpenPrediction
AI will use 7% of US electricity in 2030
“That rises to 7% of all electricity consumption in 2030”
Compute & chips
21 claims
OpenPrediction
AI will use 4% of US electricity in 2027 (7% in 2030)
“The median expert predicts that 4% of U.S. electricity consumption will be used for training and deploying AI systems in 2027.”
OpenPrediction
Frontier training clusters will cost over $100B by 2030, enabling ~1e29 FLOP runs
“On current trends, the clusters used for training frontier AI would cost over $100B by 2030. Such clusters could support training runs of about 10^29 FLOP”
PendingPromise
Hyperion data centre: 2 GW online by 2030, scaling to 5 GW
“Gabriel says Meta plans to bring two gigawatts of data center capacity online by 2030 with Hyperion, but that it would scale to five gigawatts in several years.”
PendingPromise
First 200 MW of Stargate UAE expected to go live in 2026
“A 1GW Stargate UAE cluster in Abu Dhabi with 200MW expected to go live in 2026”
OpenPrediction
Data centre electricity demand will more than double to ~945 TWh by 2030
“It projects that electricity demand from data centres worldwide is set to more than double by 2030 to around 945 terawatt-hours (TWh), slightly more than the entire electricity con…”
PendingPromise
Rubin Ultra (NVL576) in the second half of 2027
“Huang also announced Rubin Ultra, which will follow in the second half of 2027.”
PendingPromise
Vera Rubin GPUs to ship in the second half of 2026
“The centerpiece announcement was Vera Rubin, first teased at Computex 2024 and now scheduled for release in the second half of 2026.”
PendingPromise
Stargate: $500B of US AI infrastructure over four years
“The Stargate Project is a new company which intends to invest $500 billion over the next four years building new AI infrastructure for OpenAI in the United States.”
PendingPromise
UK to expand its AI Research Resource compute "by at least 20x by 2030"
“Expand the capacity of the AI Research Resource ( AIRR ) by at least 20x by 2030 – starting within 6 months.”
PendingPromise
About $80B on AI datacenters in FY2025
“In FY 2025, Microsoft is on track to invest approximately $80 billion to build out AI-enabled datacenters”
KeptPromise
Instinct MI350 accelerators to launch in the second half of 2025
“AMD also shared new details on next-gen AMD Instinct MI350 series accelerators expected to launch in the second half of 2025”
OpenPrediction
Data-center AI accelerator market to reach $500 billion by 2028
“Looking ahead, we see the data center AI accelerator market growing to $500 billion by 2028.”
OpenPrediction
Individual training clusters costing hundreds of billions of dollars by 2028
“We’re on the path to individual training clusters costing $100s of billions by 2028—clusters requiring power equivalent to a small/medium US state and more expensive than the Inter…”
BrokenPromise
Over $500 million of Gaudi AI accelerator revenue in 2024
“We now expect over $500 million in accelerated revenue in second half of 2024 and with increasing momentum into 2025 based on Gaudi 3's vastly superior TCO as well as our own expan…”
KeptPromise
EO 14110: launch a National AI Research Resource pilot within 90 days
“Within 90 days of the date of this order, in coordination with the heads of agencies that the Director of NSF deems appropriate, launch a pilot program implementing the National AI…”
BrokenPromise
20% of secured compute to superalignment over four years
“We are dedicating 20% of the compute we’ve secured to date over the next four years to solving the problem of superintelligence alignment.”
WrongPrediction
Median forecast: largest ML model would have 150 trillion parameters by 2024 (actual ~10 trillion)
“49. Largest Number of Parameters in a Machine Learning Model 150 trillion 10 trillion 2.74 7”
WrongPrediction
Median forecast: largest ML model would have 100 trillion parameters by 2024 (actual ~10 trillion)
“49. Largest Number of Parameters in a Machine Learning Model 100 trillion 10 trillion 1.71 31”
WrongPrediction
Median forecast for the maximum compute used in an AI experiment by 2024 was about 6x too low
“45. Maximum Compute Used in an AI Experiment 100,000 578,703.7 1.92 33”
OpenPrediction
A new chip shortage in 2026
“We’re in a new chip shortage. Just when the fabs thought they had caught up to demand…”
See the leaderboards by topic.