AI-054 · Source-linked project profile. Read the scope, reported result and what remains unknown.
Open the interactive projectHow to read this evidenceA Quantum portfolio company in the Piceance Basin uses an engineering-reviewed planning tool. Azure Databricks with Apache Sedona processes terrain; an integer-linear-programming solver selects pads and wells against more than 25 constraints. A Copilot Studio agent explains results. Azure Blob Storage, Dataverse, App Service and Azure OpenAI are named in the stack.
Microsoft reports a 20,000-acre scenario took 20 minutes versus three weeks; leasehold utilization rose from 50% to 85%. Production runs evaluated 2.3m pad candidates, 3.5m well possibilities and over 50m allocations; the accumulation period is unstated. Updated-version comparison: 50–65% shorter solve times in an updated version, said to compound over hundreds of runs. The source does not specify absolute times, comparable hardware or the evaluation period. Do not multiply this with the separate three-weeks-to-20-minutes comparison.
Microsoft comparisons remain vendor-reported. Quantum names QB Energy and reports 12 areas over nine months. The ~$10bn development program informed is not savings or value created. Engineers review plans; no autonomous capital approval is established. Solver attribution: Vendor/customer reporting with no full-cost return or independent performance validation. The version comparison is a different baseline from the manual-workflow comparison. Engineers retain plan review; no capital approval is delegated. The reported solver improvement is not attributed to the language model.
Optimization explores feasible development choices under assumptions. It should not be interpreted as delegated investment authority, proven investment alpha or a substitute for reserves and land diligence. The public description is undated.
Quantum Energy Cloud; Quantum IQ; AI at Work
From the field to the back office and beyond — land-title research paragraphs
Field development planning solution; Improved speed, scale, and precision; updated algorithm paragraph
Pages 11–15; page 68 disclaimer
Models support single-well economics, scenarios, acreage planning and well/pad combinations.
Optimization explores feasible development choices under assumptions. It should not be interpreted as delegated investment authority, proven investment alpha or a substitute for reserves and land diligence. The public description is undated.
A Quantum portfolio company in the Piceance Basin uses an engineering-reviewed planning tool. Azure Databricks with Apache Sedona processes terrain; an integer-linear-programming solver selects pads and wells against more than 25 constraints. A Copilot Studio agent explains results. Azure Blob Storage, Dataverse, App Service and Azure OpenAI are named in the stack.
Microsoft comparisons remain vendor-reported. Quantum names QB Energy and reports 12 areas over nine months. The ~$10bn development program informed is not savings or value created. Engineers review plans; no autonomous capital approval is established. Solver attribution: Vendor/customer reporting with no full-cost return or independent performance validation. The version comparison is a different baseline from the manual-workflow comparison. Engineers retain plan review; no capital approval is delegated. The reported solver improvement is not attributed to the language model.
The 2026 Quantum report names QB Energy behind the Piceance planning case. Earlier Microsoft reporting left the operator unnamed.
Microsoft comparisons remain vendor-reported. Quantum names QB Energy and reports 12 areas over nine months. The ~$10bn development program informed is not savings or value created. Engineers review plans; no autonomous capital approval is established. Solver attribution: Vendor/customer reporting with no full-cost return or independent performance validation. The version comparison is a different baseline from the manual-workflow comparison. Engineers retain plan review; no capital approval is delegated. The reported solver improvement is not attributed to the language model.
Microsoft says algorithm changes in an updated version of Quantum’s field-planning tool reduced solve time by 50–65%. The integer-linear-programming optimizer selects feasible pads and wells. A Copilot Studio agent explains outputs. The numerical improvement belongs to the optimization algorithm; the source does not attribute it to the language model.
Vendor/customer reporting with no full-cost return or independent performance validation. The version comparison is a different baseline from the manual-workflow comparison. Engineers retain plan review; no capital approval is delegated. The reported solver improvement is not attributed to the language model.
A source grade describes provenance, not independent verification. This is a research profile, not a measure of net return. Report a correction.