{"version":7,"as_of":"2026-09-30","editorial_status":"Dated, source-backed research records; not a live telemetry feed","records":[{"id":"synopsys-amazon-ip-20260930","company":"Synopsys / Amazon","title":"Custom silicon IP agreement with volume-linked royalty model","date":"2026-09-30","categories":["custom silicon enablement","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"Synopsys announced a multi-year agreement worth more than $1B with Amazon, expanding application-optimized silicon IP, EDA and engineering software. The license-plus-royalty model links part of the economics to production volumes.","unknowns":"Royalty rates, chip volumes, revenue recognition, tape-out dates and production milestones are undisclosed. A design agreement does not establish manufactured chips.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["Semiconductor IP","EDA and verification","SerDes and interface IP","Packaging and simulation"],"source_url":"https://news.synopsys.com/2026-09-30-Synopsys-and-Amazon-Announce-Strategic,-Multi-year-IP-Agreement-for-Custom-Silicon-Collaboration-Also-Extends-to-Cloud-and-AI-Powered-Engineering","source_label":"Primary company disclosure","procurement":{"lane":"Custom Silicon Enablement","specifications":{"contract_usd_m_floor":1000,"commercial_model":"license plus royalty","programs":["Nitro","Graviton","Trainium"],"royalty_rate":null,"production_volume":null},"named_relationships":[{"source":"Synopsys","target":"Amazon","kind":"announced IP and engineering agreement"}]},"bottleneck":{"lane":"Custom Silicon Enablement","claim_kind":"company disclosure","screening_fields":{"contract_usd_m_floor":1000,"commercial_model":"license plus royalty","programs":["Nitro","Graviton","Trainium"],"royalty_rate":null,"production_volume":null},"evidence_gates":[{"label":"Named customer and agreement","disclosed":true},{"label":"Commercial model disclosed","disclosed":true},{"label":"Royalty rates and production volume","disclosed":false},{"label":"Recognized revenue on agreement","disclosed":false}],"observations":[]},"silicon":{"deal_type":"IP_License_Royalty","contract_usd_m_floor":1000,"customer_tier":"hyperscaler","nre_recognized_usd_m":null,"design_win_status":"commercial IP agreement announced; chip design wins not specified","tapeout_target":null,"tapeout_actual":null,"production_target":null,"production_actual":null,"production_revenue_usd_m":null,"process_node_nm":null,"foundry_slot_status":"not disclosed","die_area_mm2":null,"large_die_planned":null,"chiplets":null,"interfaces":[],"ip_license_events":["Amazon application-optimized IP agreement"],"hyperscaler_direct_customer_verified":true,"taxonomy":["Interconnect_And_IP","Verification_And_Test"],"milestone_status":"IP agreement announced"}},{"id":"aib-nebius-20260930","company":"AIB Data Centers / Nebius","title":"Secured utility capacity converts to a 12-year AI tenant agreement","date":"2026-09-30","categories":["powered land & ai offtake","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"AIB announced a binding 12-year agreement with Nebius for 50 MW of critical IT capacity in two data halls in the southeastern U.S., supported by a 15-year agreement for 65 MW of utility load.","unknowns":"AIB expects customer prepayments, project debt and preferred equity to fund construction. Receipt of cash, closed financing, commissioning and tenant acceptance are not established. Utility MW and IT MW are different measures.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["EPC and commissioning","Electrical distribution","Liquid cooling","Fiber and network","Project finance"],"source_url":"https://www.financialcontent.com/article/gnwcq-2026-9-30-aib-data-centers-signs-contract-with-nebius-for-ai-data-center-capacity","source_label":"AIB GlobeNewswire release (syndicated)","procurement":{"lane":"Powered Land & AI Offtake","specifications":{"signed_it_mw":50,"utility_load_mw":65,"lease_years":12,"electric_service_years":15,"prepayments_received_usd_m":null,"financing_closed":null,"energized_it_mw":null},"named_relationships":[{"source":"AIB Data Centers","target":"Nebius","kind":"announced binding 50 MW capacity agreement"}]},"bottleneck":{"lane":"Powered Land & AI Offtake","claim_kind":"company disclosure","screening_fields":{"signed_it_mw":50,"utility_load_mw":65,"lease_years":12,"electric_service_years":15,"prepayments_received_usd_m":null,"financing_closed":null,"energized_it_mw":null},"evidence_gates":[{"label":"Binding tenant agreement","disclosed":true},{"label":"Executed utility service agreement","disclosed":true},{"label":"Construction financing closed","disclosed":false},{"label":"Prepayments received","disclosed":false},{"label":"Commissioning and tenant acceptance","disclosed":false}],"observations":[]},"capacity":{"signed_agreement":true,"contract_usd_m":null,"take_or_pay":null,"contract_years":12,"prepayments_received_usd_m":null,"prepayment_scope":"anticipated; receipt not disclosed","repeat_customer":null,"gpu_count_floor":null,"gpu_model":null,"delivery_target_known":false,"delivered_new_agreements":false,"production_conversion_new_agreements":false,"dependencies":["Construction financing","Utility-to-IT conversion","Cooling and fiber","Commissioning","Tenant acceptance"],"clusters":[{"it_mw":50,"stage":"contracted; development pending"}],"supplier_link_status":"research classes only"}},{"id":"jacobs-nvidia-twin-20260930","company":"Jacobs / NVIDIA","title":"Three-year facility digital twin SaaS agreement","date":"2026-09-30","categories":["ai factory operations & optimization","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"Jacobs announced selection by NVIDIA for a three-year SaaS agreement to deploy a Data Center Digital Twin at a large U.S. R&D facility. The platform combines engineering models, operational technology and live facility data using NVIDIA Omniverse libraries.","unknowns":"Contract value, deployment completion, measured savings, uptime impact and site-level acceptance are undisclosed. Selection for deployment does not prove the system is already operational.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["DCIM and controls","Thermal telemetry","Leak detection","Electrical monitoring","Predictive maintenance"],"source_url":"https://www.financialcontent.com/article/bizwire-2026-9-30-jacobs-to-deploy-data-center-digital-twin-for-nvidias-r-and-d-facility","source_label":"Jacobs Business Wire release (syndicated)","procurement":{"lane":"AI Factory Operations & Optimization","specifications":{"saas_years":3,"platform":"NVIDIA Omniverse","contract_value":null,"measured_energy_savings":null,"operational_acceptance":null},"named_relationships":[{"source":"Jacobs","target":"NVIDIA","kind":"announced three-year digital twin SaaS agreement"}]},"bottleneck":{"lane":"AI Factory Operations & Optimization","claim_kind":"company disclosure","screening_fields":{"saas_years":3,"platform":"NVIDIA Omniverse","contract_value":null,"measured_energy_savings":null,"operational_acceptance":null},"evidence_gates":[{"label":"Named customer and SaaS term","disclosed":true},{"label":"Facility data integration described","disclosed":true},{"label":"Completed operational acceptance","disclosed":false},{"label":"Measured savings and uptime","disclosed":false}],"observations":[]}},{"id":"swi-powered-conversion-20260930","company":"SWI Group / SWI Digital","title":"Secured grid connections and crypto-to-AI conversion assets","date":"2026-09-30","categories":["power rights & conversion assets","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"SWI reports approximately 1.2 GW of secured grid connections at SWI Digital, predominantly in the U.S., and conversion of suitable mining sites toward AI/HPC. AiOnX has about 2.3 GW of planned European capacity, including one hyperscale-leased campus.","unknowns":"Aggregate power rights do not prove AI-ready IT capacity. Prospective offtake discussions are not signed backlog. Site cooling, fiber, conversion capex and commissioning require evidence.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["Industrial site conversion","Grid interconnection","Cooling retrofit","Fiber access","Conversion engineering"],"source_url":"https://swi.com/wp-content/uploads/SWICH-Half-Year-Results-2026.pdf","source_label":"Primary company disclosure","procurement":{"lane":"Power Rights & Conversion Assets","specifications":{"secured_connections_mw_approx":1200,"europe_planned_mw_approx":2300,"ai_ready_mw":null,"conversion_capex":null,"signed_us_ai_offtake":null},"named_relationships":[]},"bottleneck":{"lane":"Power Rights & Conversion Assets","claim_kind":"company disclosure","screening_fields":{"secured_connections_mw_approx":1200,"europe_planned_mw_approx":2300,"ai_ready_mw":null,"conversion_capex":null,"signed_us_ai_offtake":null},"evidence_gates":[{"label":"Secured grid connection aggregate disclosed","disclosed":true},{"label":"Conversion strategy disclosed","disclosed":true},{"label":"Site-level AI-ready MW","disclosed":false},{"label":"Signed U.S. AI offtake evidence","disclosed":false},{"label":"Accepted converted compute","disclosed":false}],"observations":[]}},{"id":"solidultra-sst-20260930","company":"SolidUltra","title":"Medium-voltage to 800 VDC modular power conversion","date":"2026-09-30","categories":["800 vdc / ai power conversion","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"SolidUltra describes an in-development solid-state transformer using modular 100 kW silicon-carbide cells to connect medium voltage directly to 800 VDC, with a claimed grid-to-DC efficiency above 98%.","unknowns":"This is a website observation dated September 30, not a dated launch release. Efficiency and compactness are vendor claims. Utility approval, independently tested output and first accepted deployments are missing.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["SiC power devices","SST modules","DC protection","Busway","Microgrid controls"],"source_url":"https://solidultra.com/","source_label":"Primary company disclosure","procurement":{"lane":"800 VDC / AI Power Conversion","specifications":{"cell_kw":100,"output_vdc":800,"claimed_efficiency_pct_floor":98,"product_status":"in development","measured_efficiency_pct":null,"utility_approval":null},"named_relationships":[]},"bottleneck":{"lane":"800 VDC / AI Power Conversion","claim_kind":"company disclosure","screening_fields":{"cell_kw":100,"output_vdc":800,"claimed_efficiency_pct_floor":98,"product_status":"in development","measured_efficiency_pct":null,"utility_approval":null},"evidence_gates":[{"label":"Modular architecture described","disclosed":true},{"label":"Development status explicit","disclosed":true},{"label":"Independent efficiency testing","disclosed":false},{"label":"Utility approval","disclosed":false},{"label":"Accepted first deployment","disclosed":false}],"observations":[]}},{"id":"netapp-novus-20260929","company":"NetApp / Supermicro","title":"Disaggregated AI storage and metadata architecture","date":"2026-09-29","categories":["ai data feed & storage bottleneck","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"NetApp announced orderable Novus, separating metadata and data movement under one namespace. Its initial design combines Data Director on qualified Supermicro infrastructure with ONTAP on AFF A90. The company targets throughput above 100 TB/s.","unknowns":"The throughput target is a vendor architecture claim; cited analyst modeling is not a full-scale independently reproduced benchmark. Installed throughput, metadata concurrency and GPU stall reduction need workload-specific measurement.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["Parallel file systems","Metadata engines","NVMe fabrics","Storage networking","Flash controllers"],"source_url":"https://www.businesswire.com/news/home/20260929517190/en/NetApp-Removes-Storage-Bottleneck-for-AI-Factories","source_label":"Primary company disclosure","procurement":{"lane":"AI Data Feed & Storage Bottleneck","specifications":{"target_throughput_tb_s":100,"target_kind":"vendor design target; not measured deployed throughput","orderable":true,"measured_throughput_tb_s":null,"gpu_stall_pct":null,"metadata_ops_second":null},"named_relationships":[{"source":"NetApp Novus","target":"Supermicro","kind":"qualified infrastructure in disclosed initial architecture"}]},"bottleneck":{"lane":"AI Data Feed & Storage Bottleneck","claim_kind":"company disclosure","screening_fields":{"target_throughput_tb_s":100,"target_kind":"vendor design target; not measured deployed throughput","orderable":true,"measured_throughput_tb_s":null,"gpu_stall_pct":null,"metadata_ops_second":null},"evidence_gates":[{"label":"Product and initial architecture disclosed","disclosed":true},{"label":"Orderable status announced","disclosed":true},{"label":"Full-scale throughput reproduced","disclosed":false},{"label":"Measured customer GPU utilization","disclosed":false}],"observations":[]}},{"id":"hbm-wafer-crowd-out-2026","company":"Samsung / industry DRAM","title":"HBM wafer-allocation outlook","date":"2026-09-29","categories":["hbm crowd-out","ai infrastructure"],"status":"reported executive estimate","fact":"Reuters reports Samsung’s estimate that HBM uses about 20% of industry DRAM wafer capacity currently and could approach 30% in 2027.","unknowns":"Approximate estimates are not a measured time series, bit-output share or proven supply shortfall. Yield, stack height and supplier-level allocations remain unknown here.","verify":"Review the missing evidence gates before qualifying a supplier or marking capacity operational.","supplier_classes":["Wafer test","TSV and metrology","Advanced packaging","Burn-in","Thermal interface materials"],"source_url":"https://srnnews.com/samsung-electronics-says-hbm-to-account-for-nearly-30-of-industry-dram-capacity-next-year/","source_label":"Reuters reporting of Samsung executive remarks","procurement":{"lane":"HBM crowd-out","specifications":{"current_wafer_share_pct_approx":20,"forecast_2027_share_pct_approx":30,"yield_pct":null,"stack_height":null,"conventional_dram_price":null},"named_relationships":[]},"bottleneck":{"lane":"HBM crowd-out","claim_kind":"reported executive estimate","screening_fields":{"current_wafer_share_pct_approx":20,"forecast_2027_share_pct_approx":30,"yield_pct":null,"stack_height":null,"conventional_dram_price":null,"allocation_change_percentage_points_approx":10,"forecast_allocation_index_current_100_approx":150,"index_scope":"30 / 20 × 100 from approximate executive estimates; not measured shortage or pricing"},"evidence_gates":[{"label":"Attributed current estimate","disclosed":true},{"label":"Attributed 2027 forecast","disclosed":true},{"label":"Realized 2027 allocation","disclosed":false},{"label":"Yield and conventional DRAM pricing evidence","disclosed":false}],"observations":[{"metric":"HBM wafer allocation","value":20,"unit":"percent","period":"September 2026","kind":"approximate executive estimate"},{"metric":"HBM wafer allocation","value":30,"unit":"percent","period":"2027","kind":"approximate executive forecast"}]}},{"id":"hbm-asp-outlook-2027","company":"TrendForce","title":"HBM cost and configuration pressure","date":"2026-09-29","categories":["hbm crowd-out","ai infrastructure"],"status":"analyst forecast","fact":"TrendForce forecasts 2027 blended HBM selling prices rising 121% year over year, with a richer HBM4 mix and constrained supply. It reports consideration of lower-capacity 8-Hi configurations.","unknowns":"A blended selling-price forecast is not a signed supplier quote or an increase for every memory product, GPU or server. No realized 2027 price is available.","verify":"Review the missing evidence gates before qualifying a supplier or marking capacity operational.","supplier_classes":["Memory test","Probe cards","Packaging","Memory interface boards"],"source_url":"https://www.trendforce.com/presscenter/news/20260929-13255.html","source_label":"TrendForce research forecast","procurement":{"lane":"HBM crowd-out","specifications":{"forecast_blended_asp_yoy_pct":121,"forecast_period":"2027","generation_mix":"HBM4 / HBM4e ramp","supplier_quote_usd":null},"named_relationships":[]},"bottleneck":{"lane":"HBM crowd-out","claim_kind":"analyst forecast","screening_fields":{"forecast_blended_asp_yoy_pct":121,"forecast_period":"2027","generation_mix":"HBM4 / HBM4e ramp","supplier_quote_usd":null},"evidence_gates":[{"label":"Dated analyst forecast","disclosed":true},{"label":"Product-specific customer quote","disclosed":false},{"label":"Realized 2027 pricing","disclosed":false}],"observations":[{"metric":"Blended HBM ASP change","value":121,"unit":"percent YoY","period":"2027","kind":"analyst forecast"}]}},{"id":"teradyne-magnum-e2","company":"Teradyne","title":"Next-generation memory testing","date":"2026-09-29","categories":["semiconductor test intensity","ai infrastructure"],"status":"company disclosure","fact":"Teradyne launched Magnum E2 for LPDDR6, DDR6, GDDR7 and NAND, with NRZ, PAM3 and PAM4 signaling. Its product page lists up to 64 test sites and 24,500 high-speed I/O pins.","unknowns":"Maximum configuration is not measured throughput, yield improvement or customer adoption. Customer qualifications, order values and revenue contribution are unverified. It is not identified here as an HBM-specific tester.","verify":"Review the missing evidence gates before qualifying a supplier or marking capacity operational.","supplier_classes":["Probe cards","Handlers","Sockets","Interface boards","Signal integrity","Metrology"],"source_url":"https://investors.teradyne.com/news-events/press-releases/detail/451/teradyne-announces-magnum-e2-to-address-next-generation-memory-test-demands-driven-by-ai-and-high-performance-computing","source_label":"Teradyne primary release","procurement":{"lane":"Semiconductor test intensity","specifications":{"test_sites_max":64,"high_speed_io_pins_max":24500,"standards":["LPDDR6","DDR6","GDDR7","NAND"],"signaling":["NRZ","PAM3","PAM4"],"test_seconds_per_device":null,"customer_qualification":null,"product_spec_source":"https://www.teradyne.com/products/magnum-E2/"},"named_relationships":[]},"bottleneck":{"lane":"Semiconductor test intensity","claim_kind":"company disclosure","screening_fields":{"test_sites_max":64,"high_speed_io_pins_max":24500,"standards":["LPDDR6","DDR6","GDDR7","NAND"],"signaling":["NRZ","PAM3","PAM4"],"test_seconds_per_device":null,"customer_qualification":null,"product_spec_source":"https://www.teradyne.com/products/magnum-E2/"},"evidence_gates":[{"label":"Named product and protocols","disclosed":true},{"label":"Customer qualification","disclosed":false},{"label":"Volume orders","disclosed":false},{"label":"Product revenue","disclosed":false}],"observations":[]}},{"id":"magpro-aerolev-1850","company":"MagPro","title":"Cooling without evaporative towers","date":"2026-09-29","categories":["waterless / high-ambient cooling","ai infrastructure"],"status":"company disclosure","fact":"MagPro announced a 1.85 MW-class air-cooled magnetic-bearing chiller designed for 45°C ambient conditions and chilled-water supply up to 45°C, without evaporative cooling towers.","unknowns":"Rated capacity at a specific hot-day operating point, site water use, refrigerant constraints and independent performance are unverified. Vendor efficiency comparisons are not benchmark facts. A tower-free design does not establish zero total campus water consumption.","verify":"Review the missing evidence gates before qualifying a supplier or marking capacity operational.","supplier_classes":["Magnetic-bearing compressors","EC fans","Heat exchangers","Refrigerant components","Controls","Dry coolers"],"source_url":"https://en.prnasia.com/releases/apac/magpro-launches-aerolev-1850-to-advance-water-smart-cooling-for-ai-data-centers-549868.shtml","source_label":"MagPro primary release","procurement":{"lane":"Waterless / high-ambient cooling","specifications":{"rated_thermal_mw":1.85,"ambient_design_c":45,"water_supply_max_c":45,"evaporative_tower_required":false,"capacity_at_45c_verified":null,"measured_wue":null,"independent_efficiency_result":null},"named_relationships":[]},"bottleneck":{"lane":"Waterless / high-ambient cooling","claim_kind":"company disclosure","screening_fields":{"rated_thermal_mw":1.85,"ambient_design_c":45,"water_supply_max_c":45,"evaporative_tower_required":false,"capacity_at_45c_verified":null,"measured_wue":null,"independent_efficiency_result":null},"evidence_gates":[{"label":"Product specifications announced","disclosed":true},{"label":"Customer qualification","disclosed":false},{"label":"Site deployment","disclosed":false},{"label":"Independent operating benchmark","disclosed":false}],"observations":[]}},{"id":"efficient-fabric-datacenter","company":"Efficient Computer","title":"From shipping edge chips to data-center roadmap","date":"2026-09-29","categories":["compute per watt / non-gpu ai","ai infrastructure"],"status":"company disclosure","fact":"Efficient announced agreements for more than $97 million in Series B financing at a $650 million valuation. Electron E1 is shipping for physical AI; funding supports scaling Fabric to data-center workloads.","unknowns":"Data-center design wins, OEM systems, foundry partner and comparable energy benchmarks are not disclosed here. Forward-looking energy gains cannot qualify a data-center system. Shipping E1 evidence applies to physical AI.","verify":"Review the missing evidence gates before qualifying a supplier or marking capacity operational.","supplier_classes":["Compiler tooling","Server OEM integration","Packaging","Foundry interfaces","Workload benchmarking"],"source_url":"https://www.prnewswire.com/news-releases/efficient-computer-taking-on-ais-energy-problem-announces-97m-to-scale-its-processors-from-physical-ai-to-the-datacenter-302892552.html","source_label":"Efficient Computer primary release","procurement":{"lane":"Compute per watt / non-GPU AI","specifications":{"series_b_usd_m_floor":97,"valuation_usd_m":650,"total_funding_usd_m":173,"edge_shipping":true,"datacenter_shipping_verified":false,"datacenter_energy_per_task_wh":null,"foundry_partner":null},"named_relationships":[]},"bottleneck":{"lane":"Compute per watt / non-GPU AI","claim_kind":"company disclosure","screening_fields":{"series_b_usd_m_floor":97,"valuation_usd_m":650,"total_funding_usd_m":173,"edge_shipping":true,"datacenter_shipping_verified":false,"datacenter_energy_per_task_wh":null,"foundry_partner":null},"evidence_gates":[{"label":"Data-center development funding announced","disclosed":true},{"label":"Data-center design win","disclosed":false},{"label":"Data-center system shipment","disclosed":false},{"label":"Comparable data-center energy benchmark","disclosed":false}],"observations":[]}},{"id":"qumulus-b300-contracts","date":"2026-09-29","company":"QumulusAI","title":"Two new inference capacity agreements covering more than 2,000 B300 GPUs","categories":["contract-backed ai capacity","capacity reservation","compute","backlog","cooling","connectivity","ai infrastructure finance"],"status":"contracted; turnover targeted","fact":"QumulusAI announced two three-year take-or-pay agreements totaling $240.9 million across more than 2,000 B300 GPUs. Both are repeat customers. The $211.9 million cluster targets early 2027 turnover; the $29.1 million cluster targets December 2026.","unknowns":"The newly announced clusters have not been shown as delivered or revenue producing. Extension options are outside contracted value; no prepayments were disclosed for these two awards. The earlier completed $18 million Philadelphia deployment is separate.","verify":"Track hardware receipt, powered capacity, network/storage/cooling readiness, customer acceptance, turnover, invoicing and reported revenue separately for each cluster.","supplier_classes":["B300 server integrators and GPU procurement","High-density power and cooling","Network optics and switching","Storage and CPU orchestration","Commissioning and operations"],"source_url":"https://www.businesswire.com/news/home/20260929733963/en/","source_label":"QumulusAI Business Wire release","capacity":{"contract_usd_m":240.9,"take_or_pay":true,"prepayments_received_usd_m":null,"prepayment_scope":"not disclosed for these awards","repeat_customer":true,"gpu_count_floor":2000,"gpu_model":"B300","clusters":[{"contract_usd_m":211.9,"turnover_target":"early 2027","stage":"contracted"},{"contract_usd_m":29.1,"turnover_target":"December 2026","stage":"contracted"}],"delivery_target_known":true,"delivered_new_agreements":false,"production_conversion_new_agreements":false,"dependencies":["GPU receipt","facility power and rack density","liquid cooling","network and storage","commissioning and customer acceptance"],"supplier_link_status":"research classes only"},"procurement":{"lane":"Contracted capacity","specifications":{"contracted_usd_m":240.9,"new_clusters_delivered":false,"earliest_target":"December 2026"},"named_relationships":[]}},{"id":"samsung-helix-investment","date":"2026-09-29","company":"Samsung affiliates / Helix","title":"$1 billion announced for integrated AI infrastructure","categories":["integrated infrastructure platform","power certainty","distributed power","connectivity","construction","capital","data center readiness"],"status":"investment announced","fact":"Samsung Electronics and five other Samsung affiliates announced a combined $1 billion investment in KKR-established Helix; Samsung Electronics' portion is $500 million. Helix describes a mandate spanning data centers, power generation and distribution, and fiber.","unknowns":"The announcement does not establish completed deployments, energized capacity, individual supplier awards or investment return.","verify":"Track closing and funding, sites, power rights, fiber routes, EPC awards, commissioning and operating capacity by project.","supplier_classes":["Power developers and interconnection specialists","Powered-land and data-center developers","Fiber route and network operators","Electrical EPC and commissioning firms"],"source_url":"https://news.samsung.com/global/samsung-to-invest-usd-1-billion-in-ai-infrastructure-company-helix","source_label":"Samsung release","theme":{"key":"integrated-platform","label":"Integrated AI infrastructure","geography":"Global","stage":"investment announced","evidence":{"primary_source":true,"named_parties":true,"commercial_commitment":true,"funded_or_ordered":false,"site_identified":false,"equipment_installed":false,"operating_measurement":false,"paying_customer":false},"screening":{"power_rights":"verify by project","grid_and_transmission":"verify by project","fiber":"verify by route","construction":"EPC awards unverified","compute":"operating capacity unverified"},"next_milestone":"Funded closing and first named project with power, fiber and EPC evidence"},"bottleneck":{"lane":"Integrated AI infrastructure","claim_kind":"company capital commitment","screening_fields":{"capital_commitment_usd_b":1,"project_power_secured":null,"project_fiber_secured":null,"project_compute_partner":null,"project_construction_capability":null},"observations":[],"evidence_gates":[{"label":"Named capital commitment","disclosed":true},{"label":"Project power secured","disclosed":false},{"label":"Project fiber secured","disclosed":false},{"label":"Project compute and construction qualification","disclosed":false}]},"procurement":{"lane":"Integrated AI infrastructure","specifications":{"capital_commitment_usd_b":1,"project_power_secured":null,"project_fiber_secured":null,"project_compute_partner":null,"project_construction_capability":null},"named_relationships":[]}},{"id":"eaton-infineon-sst","date":"2026-09-29","company":"Eaton / Infineon","title":"SiC devices for medium-voltage solid-state transformer 2.0","categories":["solid-state power chain","800vdc","transformers","ai power architecture","long lead power equipment","qualified infrastructure"],"status":"supply collaboration announced","fact":"Infineon says it will supply SiC power devices for Eaton's MVSST 2.0 platform designed for APAC deployment and emerging 800 VDC data-center architectures. The companies are exploring future 2.3 kV and 3.3 kV module platforms.","unknowns":"The release does not disclose volume shipments, customer deployments, commercial revenue or measured facility efficiency. Future module platforms remain exploratory.","verify":"Track device and module qualification, IEC certification scope, production orders, installation, conversion efficiency and field uptime.","supplier_classes":["SiC power modules and packaging","Gate drivers and control electronics","High-frequency magnetics","DC protection and breakers","Busway and connectors","Thermal components"],"source_url":"https://www.infineon.com/press-release/2026/infpr202609-146","source_label":"Infineon release","procurement":{"lane":"800 VDC power","specifications":{"architecture":"MV SST 2.0 for 800 VDC","delivered_volume":null},"named_relationships":[{"supplier":"Infineon","customer":"Eaton","component":"SiC power devices","stage":"Announced supply relationship"}]}},{"id":"liquidstack-cdu-2x","date":"2026-09-29","company":"Trane Technologies / LiquidStack","title":"LiquidStack CDU 2.X released for pre-order","categories":["cooling","cooling orchestration","cooling flexibility","cdus","qualified infrastructure"],"status":"pre-order; shipments planned","fact":"Trane announced the configurable CDU 2.X platform. Pre-orders are open and shipments are planned to begin in Q2 2027.","unknowns":"Field installations, measured performance and delivery timing remain unverified.","verify":"Track rated flow and thermal boundary, actual shipment, commissioning, redundancy and site-measured performance.","source_url":"https://liquidstack.com/coolant-distribution-unit-2x","source_label":"Manufacturer product page","procurement":{"lane":"Cooling","specifications":{"thermal_capacity_mw":2.5,"capacity_condition":"4°C approach temperature difference","flow_lpm":3750,"pressure_bar":3.5,"deployment_verified":false},"named_relationships":[]}},{"id":"delta-800vdc-inrow","date":"2026-09-29","company":"Delta","title":"800 VDC InRow power for AI modular data centers","categories":["800 vdc power","ai infrastructure"],"status":"Company disclosure; verification pending","fact":"Delta describes 660 kW output and up to 98% AC/DC efficiency for its 800 VDC InRow power system, alongside a 480 kW battery backup unit.","unknowns":"Manufacturer specifications do not establish installed output, commissioning, battery duration or efficiency at every load.","verify":"Request dated delivery, commissioning and acceptance evidence for the specific project or product.","supplier_classes":["Power conversion","Busway and protection","Liquid cooling integration"],"source_url":"https://en.prnasia.com/releases/apac/delta-electronics-unveils-ai-modular-data-center-with-800-vdc-in-row-power-for-nvidia-vera-rubin-and-microgrid-solutions-at-data-center-world-asia-2026-549945.shtml","source_label":"Delta primary disclosure","procurement":{"lane":"800 VDC power","specifications":{"output_kw":660,"maximum_efficiency_pct":98,"bbu_power_kw":480,"bbu_energy_kwh":null},"named_relationships":[]}},{"id":"compal-800vdc-showcase","date":"2026-09-29","company":"Compal","title":"AI factory infrastructure showcase","categories":["800 vdc power","ai infrastructure"],"status":"Company disclosure; verification pending","fact":"Compal announced an OCP Global Summit showcase including 800 VDC megawatt power architecture and liquid cooling.","unknowns":"Showcase claims do not establish orders, exact product output, shipped units or operating performance.","verify":"Request dated delivery, commissioning and acceptance evidence for the specific project or product.","supplier_classes":["Rack integration","Power delivery","Liquid cooling"],"source_url":"https://www.prnewswire.com/news-releases/compal-brings-nvidia-ai-factory-infrastructure-to-ocp-global-summit-2026-302892744.html","source_label":"Compal primary disclosure","procurement":{"lane":"800 VDC power","specifications":{},"named_relationships":[]}},{"id":"phaidra-aligned-prism","date":"2026-09-29","company":"Phaidra / Aligned","title":"Read-only facility operations intelligence","categories":["facility operations","ai infrastructure"],"status":"Company disclosure; verification pending","fact":"Phaidra and Aligned announced a partnership to deploy Prism across Aligned’s portfolio. Prism reads building and electrical monitoring information and documentation to assist operations teams.","unknowns":"No facility telemetry is connected here. The announcement does not quantify independently measured reliability improvement or completed site count. Human operators retain control.","verify":"Request dated delivery, commissioning and acceptance evidence for the specific project or product.","supplier_classes":[],"source_url":"https://www.prnewswire.com/news-releases/phaidra-and-aligned-data-centers-partner-to-empower-operations-teams-and-strengthen-data-center-reliability-with-ai-302892292.html","source_label":"Phaidra / Aligned primary disclosure","procurement":{"lane":"Facility operations","specifications":{"control_mode":"Read-only","verified_site_count":null},"named_relationships":[{"supplier":"Phaidra","customer":"Aligned Data Centers","component":"Prism operations intelligence","stage":"Announced deployment partnership"}]}},{"id":"runpod-2026-compute-scarcity","date":"2026-09-29","company":"Runpod","title":"GPU pricing and scarcity observations","categories":["compute economics","ai infrastructure"],"status":"Company disclosure; verification pending","fact":"Runpod reports a 31% B200 list-price increase from January to August 2026 on its platform. Its report discusses constrained HBM supply and delayed power availability.","unknowns":"This is a platform-specific observation, not a market-wide price series. No current absolute quote, availability or workload performance is supplied.","verify":"Request dated delivery, commissioning and acceptance evidence for the specific project or product.","supplier_classes":[],"source_url":"https://www.prnewswire.com/news-releases/runpod-fall-2026-state-of-ai-compute-report-finds-ai-developers-adapting-to-gpu-crunch-302892458.html","source_label":"Runpod primary disclosure","procurement":{"lane":"Compute economics","specifications":{"platform":"Runpod","b200_january_price_index":100,"b200_august_price_index":131,"index_note":"Normalized from reported 31% change; not actual dollar quotes"},"named_relationships":[]}},{"id":"teco-contracted-capacity-september","date":"2026-09-29","company":"TECO","title":"Contracted infrastructure and busway capacity","categories":["contracted capacity","ai infrastructure"],"status":"Company disclosure; verification pending","fact":"TECO reports nearly 1 GW in cumulative contracted data center capacity and an ongoing 10 MW modular project. Its TECOBAR Penang facility has designed annual capacity of 400,000 meters of busway.","unknowns":"Contracted capacity is not energized capacity. Designed factory throughput is not actual production. Named hyperscalers, shipment volumes and commissioning remain unverified.","verify":"Request dated delivery, commissioning and acceptance evidence for the specific project or product.","supplier_classes":["Busway","Power equipment","Cooling","Energy management"],"source_url":"https://kyodonewsprwire.jp/release/202609296836","source_label":"TECO primary disclosure","procurement":{"lane":"Contracted capacity","specifications":{"cumulative_contracted_gw_approx":1,"ongoing_project_mw":10,"designed_busway_meters_year":400000},"named_relationships":[]}},{"id":"corning-att-fiber-2026","date":"2026-09-29","company":"Corning / AT&T","title":"Multi-year domestic fiber supply agreement","categories":["optical backbone","ai infrastructure"],"status":"Company disclosure; verification pending","fact":"AT&T and Corning announced a multi-year fiber and cable agreement exceeding $3 billion. It supports AT&T broadband expansion.","unknowns":"The full agreement is not exclusively AI infrastructure. Contract value is not recognized revenue; shipment schedules and product mix are not provided here.","verify":"Request dated delivery, commissioning and acceptance evidence for the specific project or product.","supplier_classes":["Optical fiber","Cable manufacturing"],"source_url":"https://www.corning.com/worldwide/en/about-us/news-events/news-releases/2026/09/att-and-corning-team-up-to-engineer-the-connected-world-with-fiber.html","source_label":"Corning / AT&T primary disclosure","procurement":{"lane":"Optical backbone","specifications":{"contract_usd_b_floor":3,"scope":"Broadband expansion"},"named_relationships":[{"supplier":"Corning","customer":"AT&T","component":"Fiber and cable","stage":"Announced multi-year agreement"}]}},{"id":"corenix-modular-ai-factory","company":"Corenix / Submer","title":"Factory integration through site acceptance","date":"2026-09-28","categories":["modular ai factories","ai infrastructure"],"status":"company disclosure","fact":"Submer launched Corenix to factory-integrate and test IT, power, cooling and networking modules before delivery, site acceptance and commissioning, using NVIDIA reference designs.","unknowns":"Submer’s reported 500 MW-plus liquid-cooled deployments and 8 GW-plus powered land are group-level claims, not Corenix operating capacity. Factory acceptance, energization and operational compute dates must be tracked separately.","verify":"Review the missing evidence gates before qualifying a supplier or marking capacity operational.","supplier_classes":["Rack integration","Prefab electrical","Cooling skids","Busway","Optics and connectors","Controls","Logistics","Commissioning"],"source_url":"https://submer.com/blog/press/submer-group-launches-corenix-to-deliver-modular-data-center-platforms/","source_label":"Corenix / Submer primary release","procurement":{"lane":"Modular AI factories","specifications":{"group_liquid_cooled_mw_floor_claim":500,"group_powered_land_gw_floor_claim":8,"project_order_date":null,"project_power_date":null,"project_operational_compute_date":null,"components_per_mw":null},"named_relationships":[]},"bottleneck":{"lane":"Modular AI factories","claim_kind":"company disclosure","screening_fields":{"group_liquid_cooled_mw_floor_claim":500,"group_powered_land_gw_floor_claim":8,"project_order_date":null,"project_power_date":null,"project_operational_compute_date":null,"components_per_mw":null},"evidence_gates":[{"label":"Integrated factory delivery model announced","disclosed":true},{"label":"Named project factory acceptance","disclosed":false},{"label":"Named project energization","disclosed":false},{"label":"Named project operational compute","disclosed":false}],"observations":[]}},{"id":"terraflow-dgmatrix-power-conditioning","date":"2026-09-28","company":"TerraFlow Energy / DG Matrix","title":"Flow-battery UPS and solid-state transformer for live HPC","categories":["power conditioning","solid-state power chain","long-duration ups","ai power architecture"],"status":"commercial deployment agreement; operating results pending","fact":"TerraFlow and DG Matrix announced an agreement for an initial deployment combining TerraFlow LDUPS vanadium-redox-flow storage, DG Matrix Interport solid-state transformer technology and Dell PowerEdge servers in an HPC cluster. DG Matrix separately announced a 400 kW Interport module using STMicroelectronics SiC devices.","unknowns":"The first-U.S. characterization is the companies’ claim. The agreement does not yet show commissioned operation, measured grid-load buffering, uptime, customer revenue or repeat orders. The 400 kW module announcement is separate from proof of the integrated deployment.","verify":"Track named site, commissioning, metered transient response, grid interaction, UPS duration, 400 kW qualification, repeat purchase orders and operating uptime.","supplier_classes":["SiC power devices and packaging","Vanadium electrolyte and flow-battery components","Solid-state transformer modules","Power controls and harmonic filtering","Grid commissioning and monitoring"],"source_url":"https://www.dgmatrix.com/news-and-media/press-release/terraflow-energy-and-dg-matrix-to-deploy-first-u-s-vrfb-and-solid-state-transformer-architecture-to-power-live-high-performance-compute/1235","source_label":"DG Matrix company release","theme":{"key":"power-conditioning","label":"AI power conditioning","geography":"United States","stage":"deployment agreed; performance pending","evidence":{"primary_source":true,"named_parties":true,"commercial_commitment":true,"funded_or_ordered":false,"site_identified":false,"equipment_installed":false,"operating_measurement":false,"paying_customer":false},"screening":{"grid_to_rack_architecture":"LDUPS plus Interport proposed","transient_response":"metered result pending","storage_duration":"claimed multi-hour; site measurement pending","sic_device":"ST collaboration announced separately","commissioning":"pending","repeat_orders":"unverified"},"next_milestone":"Commissioned HPC cluster and third-party metered performance"},"procurement":{"lane":"800 VDC power","specifications":{},"named_relationships":[]}},{"id":"netapp-peak-aio-20260925","company":"NetApp / PEAK:AIO","title":"Planned acquisition of specialist metadata and parallel file technology","date":"2026-09-25","categories":["compute efficiency infrastructure","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"NetApp announced its intent to acquire PEAK:AIO, adding specialist metadata and parallel file-system technology to its AI infrastructure roadmap.","unknowns":"Closing and regulatory conditions remain outstanding. Transaction price, realized integration benefits and deployed hyperscale performance are not established.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["Metadata architecture","Parallel NFS","Namespace software","HPC storage IP"],"source_url":"https://investors.netapp.com/news/news-details/2026/NetApp-Announces-Intent-to-Acquire-PEAKAIO-to-Advance-Scalable-AI-Infrastructure-Architecture/default.aspx","source_label":"Primary company disclosure","procurement":{"lane":"Compute Efficiency Infrastructure","specifications":{"transaction_status":"announced intent; subject to closing","transaction_value":null,"closing_completed":false},"named_relationships":[{"source":"NetApp","target":"PEAK:AIO","kind":"planned acquisition; closing not confirmed"}]},"bottleneck":{"lane":"Compute Efficiency Infrastructure","claim_kind":"company disclosure","screening_fields":{"transaction_status":"announced intent; subject to closing","transaction_value":null,"closing_completed":false},"evidence_gates":[{"label":"Acquisition intent disclosed","disclosed":true},{"label":"Technology category disclosed","disclosed":true},{"label":"Transaction closed","disclosed":false},{"label":"Integrated customer deployment","disclosed":false}],"observations":[]}},{"id":"atlas-energy-ai-power-20260925","company":"Atlas Energy Solutions","title":"AI-customer equipment reimbursement agreements","date":"2026-09-25","categories":["ai power equipment & financing","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"Atlas announced equipment cost-reimbursement agreements with an unnamed frontier AI lab. One supports an incremental 283 MW of Caterpillar generation commitments; a separate 328 MW purchase is for 2027 delivery under its framework agreement.","unknowns":"Equipment orders do not establish energized capacity. The two purchase scopes must remain separate; cash reimbursement received and completed long-term power purchase agreements are not established.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["Gas generation","Emissions controls","Electrical distribution","Battery storage","Balance-of-plant"],"source_url":"https://ir.atlas.energy/news-events/press-releases/detail/76/atlas-energy-solutions-announces-equipment-purchase","source_label":"Primary company disclosure","procurement":{"lane":"AI Power Equipment & Financing","specifications":{"incremental_customer_project_mw":283,"separate_2027_framework_purchase_mw":328,"reimbursement_cash_received":null,"power_purchase_agreement_executed":null},"named_relationships":[{"source":"Atlas Energy Solutions","target":"Caterpillar","kind":"disclosed generation equipment purchase commitments"}]},"bottleneck":{"lane":"AI Power Equipment & Financing","claim_kind":"company disclosure","screening_fields":{"incremental_customer_project_mw":283,"separate_2027_framework_purchase_mw":328,"reimbursement_cash_received":null,"power_purchase_agreement_executed":null},"evidence_gates":[{"label":"Cost reimbursement agreements","disclosed":true},{"label":"Equipment commitments","disclosed":true},{"label":"Reimbursement cash received","disclosed":false},{"label":"Completed long-term PPA","disclosed":false},{"label":"Energized generation","disclosed":false}],"observations":[]}},{"id":"akamai-anthropic-cpu-20260924","company":"Akamai / Anthropic","title":"Seven-year CPU workload infrastructure commitment","date":"2026-09-24","categories":["secondary ai bottlenecks","ai infrastructure"],"status":"Company disclosure; operating proof tracked separately","fact":"Akamai announced $11.6B of Anthropic contractual commitments over seven years for CPU workloads, with up to $9B of potential expansion. It estimates $5.5B of related capex and an additional $1.7B in 2026 to secure components including memory.","unknowns":"Expansion is optional. Contract commitments are not recognized revenue; spending estimates are not realized costs. CPU and memory supplier allocation, funded expansion and delivery economics need verification.","verify":"Check the missing disclosure gates against dated project evidence before qualifying capacity or suppliers.","supplier_classes":["Server CPUs","DDR memory","NICs and optics","Distributed cloud","Storage"],"source_url":"https://www.ir.akamai.com/news-releases/news-release-details/akamai-announces-116-billion-multi-year-agreement-anthropic","source_label":"Primary company disclosure","procurement":{"lane":"Secondary AI Bottlenecks","specifications":{"contract_usd_m":11600,"term_years":7,"optional_expansion_usd_m":9000,"estimated_capex_usd_m":5500,"additional_2026_estimated_capex_usd_m":1700},"named_relationships":[{"source":"Akamai","target":"Anthropic","kind":"announced seven-year CPU workload commitment"}]},"bottleneck":{"lane":"Secondary AI Bottlenecks","claim_kind":"company disclosure","screening_fields":{"contract_usd_m":11600,"term_years":7,"optional_expansion_usd_m":9000,"estimated_capex_usd_m":5500,"additional_2026_estimated_capex_usd_m":1700},"evidence_gates":[{"label":"Named customer and contractual commitment","disclosed":true},{"label":"Term and capex estimate disclosed","disclosed":true},{"label":"Supplier-specific CPU/memory allocation","disclosed":false},{"label":"Measured delivered contract economics","disclosed":false}],"observations":[]}},{"id":"corning-verizon-fiber-2026","date":"2026-09-08","company":"Corning / Verizon","title":"Fiber for broadband and AI infrastructure","categories":["optical backbone","ai infrastructure"],"status":"Company disclosure; verification pending","fact":"Verizon and Corning announced a multi-year, multibillion-dollar agreement for more than 80 million fiber miles over 2027–2032, including AI infrastructure connectivity.","unknowns":"Fiber miles are not route miles. Exact contract dollars, installed routes, delivery and recognized revenue remain distinct.","verify":"Request dated delivery, commissioning and acceptance evidence for the specific project or product.","supplier_classes":["Optical fiber","High-density connectivity"],"source_url":"https://www.corning.com/worldwide/en/about-us/news-events/news-releases/2026/09/verizon-and-corning-announce-multi-year-multi-billion-dollar-supply-agreement-for-broadband-expansion-and-next-gen-ai-infrastructure.html","source_label":"Corning / Verizon primary disclosure","procurement":{"lane":"Optical backbone","specifications":{"fiber_miles_floor":80000000,"delivery_window":"2027–2032"},"named_relationships":[{"supplier":"Corning","customer":"Verizon","component":"Fiber and connectivity","stage":"Announced supply agreement"}]}}],"named_relationships":[{"source":"Synopsys","target":"Amazon","kind":"announced IP and engineering agreement","source_url":"https://news.synopsys.com/2026-09-30-Synopsys-and-Amazon-Announce-Strategic,-Multi-year-IP-Agreement-for-Custom-Silicon-Collaboration-Also-Extends-to-Cloud-and-AI-Powered-Engineering","signal_id":"synopsys-amazon-ip-20260930"},{"source":"AIB Data Centers","target":"Nebius","kind":"announced binding 50 MW capacity agreement","source_url":"https://www.financialcontent.com/article/gnwcq-2026-9-30-aib-data-centers-signs-contract-with-nebius-for-ai-data-center-capacity","signal_id":"aib-nebius-20260930"},{"source":"Jacobs","target":"NVIDIA","kind":"announced three-year digital twin SaaS agreement","source_url":"https://www.financialcontent.com/article/bizwire-2026-9-30-jacobs-to-deploy-data-center-digital-twin-for-nvidias-r-and-d-facility","signal_id":"jacobs-nvidia-twin-20260930"},{"source":"NetApp Novus","target":"Supermicro","kind":"qualified infrastructure in disclosed initial architecture","source_url":"https://www.businesswire.com/news/home/20260929517190/en/NetApp-Removes-Storage-Bottleneck-for-AI-Factories","signal_id":"netapp-novus-20260929"},{"supplier":"Infineon","customer":"Eaton","component":"SiC power devices","stage":"Announced supply relationship","source_url":"https://www.infineon.com/press-release/2026/infpr202609-146","signal_id":"eaton-infineon-sst"},{"supplier":"Phaidra","customer":"Aligned Data Centers","component":"Prism operations intelligence","stage":"Announced deployment partnership","source_url":"https://www.prnewswire.com/news-releases/phaidra-and-aligned-data-centers-partner-to-empower-operations-teams-and-strengthen-data-center-reliability-with-ai-302892292.html","signal_id":"phaidra-aligned-prism"},{"supplier":"Corning","customer":"AT&T","component":"Fiber and cable","stage":"Announced multi-year agreement","source_url":"https://www.corning.com/worldwide/en/about-us/news-events/news-releases/2026/09/att-and-corning-team-up-to-engineer-the-connected-world-with-fiber.html","signal_id":"corning-att-fiber-2026"},{"source":"NetApp","target":"PEAK:AIO","kind":"planned acquisition; closing not confirmed","source_url":"https://investors.netapp.com/news/news-details/2026/NetApp-Announces-Intent-to-Acquire-PEAKAIO-to-Advance-Scalable-AI-Infrastructure-Architecture/default.aspx","signal_id":"netapp-peak-aio-20260925"},{"source":"Atlas Energy Solutions","target":"Caterpillar","kind":"disclosed generation equipment purchase commitments","source_url":"https://ir.atlas.energy/news-events/press-releases/detail/76/atlas-energy-solutions-announces-equipment-purchase","signal_id":"atlas-energy-ai-power-20260925"},{"source":"Akamai","target":"Anthropic","kind":"announced seven-year CPU workload commitment","source_url":"https://www.ir.akamai.com/news-releases/news-release-details/akamai-announces-116-billion-multi-year-agreement-anthropic","signal_id":"akamai-anthropic-cpu-20260924"},{"supplier":"Corning","customer":"Verizon","component":"Fiber and connectivity","stage":"Announced supply agreement","source_url":"https://www.corning.com/worldwide/en/about-us/news-events/news-releases/2026/09/verizon-and-corning-announce-multi-year-multi-billion-dollar-supply-agreement-for-broadband-expansion-and-next-gen-ai-infrastructure.html","signal_id":"corning-verizon-fiber-2026"}],"readiness_methodology":{"signed_demand":20,"power_secured":15,"colocation_secured":10,"financing_secured":10,"equipment_received":15,"commissioned":15,"customer_accepted":10,"revenue_recognized":5},"scope":"Curated disclosures and manufacturer specifications; no live quotes or facility telemetry connected"}