Start with structured Events, connect a company to its filings, and hand your application records it can use. The same data is available through REST and MCP.
Retrieve a focused feed for a known company. 3 records in this saved page.
01 / EVENT
AWS commits to more than 3 million Nvidia GPUs by 2029
Occurrence date inferred from publication. Verify the original reporting before treating it as the date the event happened.
On 2026-09-21, AWS reportedly committed to purchase 2 million additional Nvidia GPUs, adding to more than 1 million committed in March 2026. The combined total exceeds 3 million chips due before 2029, while Nvidia deliveries reportedly begin in the current quarter. Neither company disclosed the deal's cost or the precise definition of a GPU in the commitment.
Occurrence date inferred from publication. Verify the original reporting before treating it as the date the event happened.
On 2026-09-21, Nvidia reportedly reached a market valuation of $1 trillion. The title-only report identifies this as a major valuation milestone but provides no details on the trading move, operational drivers, or broader market effects.
Zayo secures Corning fiber for AI network expansion
Occurrence date inferred from publication. Verify the original reporting before treating it as the date the event happened.
On 2026-09-18, Zayo announced an expanded strategic supply agreement with Corning to secure fiber optic cable for its planned network expansion. Zayo plans to add 15,000 route miles by 2030, including more than 8,000 long-haul miles in collaboration with Nvidia. The agreement responds to reported tightening fiber supply driven by broadband, data-center and AI demand.
What happened. Who’s involved. Why it deserves attention.
GDELT Cloud adds structured classification, actor roles, geographic precision and explainable Event metrics to the raw article stream. Here is what the API returned for one Event in this workflow.
CAMEOPLUS · CO05
AWS commits to more than 3 million Nvidia GPUs by 2029
Capital Allocation Event · 2026-09-21
Reported location
Seattle, United Statesnearby area
Occurrence date basis
publication day inference
Evidence returned
1 articles · 1 evidence sources · 1 supporting StoriesArticle volume is not a count of independent corroborations.
Model judgments for prioritization, not measured losses or calibrated probabilities. Compare magnitude within its event domain. Missing scores stay unknown.
Open the reasoning returned with these scores
Magnitude
AWS committed to more than 3 million Nvidia GPUs due before 2029, a major strategic AI infrastructure deployment.
Systemic importance · interconnectedness
The chips serve hyperscale cloud and AI data-center operations, although the article does not quantify downstream customers.
Propagation · transmission channel
The commitment directly links AWS cloud infrastructure demand with Nvidia's GPU production and deliveries.
Market sensitivity · economic effect
A major purchase commitment has been made, but delivery and total deal cost remain unspecified and much of the deployment is future-dated.
MEET THE RECORDS
Four building blocks. Real examples.
A Story collects the reporting. An Event describes what happened. Entities identify who; facilities locate physical assets.
Story
Related reporting, grouped together.
Zelensky says US is preparing new investments in Ukraine's defense and energy
3 articles · linked to the Event beside it
Event
What happened, classified and located.
Zelensky says US is preparing investments in Ukraine
Make Public Statement · confidence 0.58
Entity
A person, organization or place with a reusable identity.
Nvidia
organization · 7 Stories on 2026-09-24
Facility
A physical asset with a location and registry details.
Amercoeur power station
oil gas plant · BEL · Global Energy Monitor
The Story and Event shown here are linked. NVIDIA and the facility are separate examples; their IDs let you connect the records returned by your own queries.
THREE STEPS, A WORKING WORKFLOW
Choose your scope. Inspect the evidence. Use the records.
01
Retrieve a focused feed for a known company.
Dates, topics, places and classification are explicit request parameters.
02
Inspect the event type, identity and linked evidence.
Keep original sources, model scores and coverage beside the result.
03
Add filing context and copy the integration.
Copy a request into your application. Start with one bounded page.
DATA BEHIND THIS WORKFLOW
Connect the entity to SEC EDGAR
We combine the GDELT Project article stream with additional regional news sources we ingest directly. GDELT Cloud groups the reporting into Stories, classifies Events and links people, organizations and places.
Find NVIDIA in the entities endpoint, reuse its identity across the Event feed and SEC EDGAR, and inspect disclosed business relationships.
The Event query and SEC filer share the same resolved company identity. Filing evidence has its own reporting date.
Build a company watch that connects reporting, a reusable entity identity and filing evidence.
Redistributing GDELT Cloud-generated Events, Stories, entity matches, model extractions, scores, and other generated intelligence requires a separate written license for the agreed product and delivery scope. Source-only records and customer-authored analysis based solely on them are governed by the original source terms and carry no additional GDELT Cloud redistribution fee. Review licensing scope →
WORK WITH WHAT YOU HAVE
Fill a gap. Or build your first version.
Alongside your existing workflow
Give an existing agent or application structured external context alongside your own retrieval and business data.
From a blank page
Begin with one working company watch, then adapt the request and response schema for a client application.
YOUR WORKLOAD, MADE EXPLICIT
Small start. Room to build.
Each list page, detail lookup and context request counts as an API call. Set your schedule and see exactly what it adds up to.
1 QU per list page · 1 QU per detail lookup. Your own agent’s direct MCP calls use the same rates as REST.
List requests
4 runs × 1 pages × 1 QU = 4 QU/day
Detail lookups
10 × 1 QU = 10 QU/day
Additional context
0 QU/day
Monthly budget
14 QU/day × 30 days
420 QU / month420 requests including the retry allowance
Builder · $59/month
Lowest current public monthly plan covering these QU and required REST data access. 1,500 QU included.
Annual option: $590 billed yearly.
Hosted Monitors are separate: 3 capacity · daily minimum cadence. This calculator models API polling.
This estimate covers subscription access and API usage. It includes no separate GDELT Cloud redistribution fee for source-only records or customer-authored analysis based solely on them; upstream source terms still apply. A separate written license and any related fee may apply when redistributing GDELT Cloud-generated output. Review licensing scope →
Estimate assumptions
30 days; each context workflow runs once per scheduled run, not once per result. Detail lookups are additional daily calls. Each saved list page requests up to 3 records. Retry allowance budgets additional charged calls; it is not a claim that every failed request is billed. Other account usage and Monitors are excluded.
Describe the task. Let your agent run the workflow.
Connect your agent to GDELT Cloud through MCP. The MCP server wraps our API endpoints, so your agent can discover the right tools, apply filters and chain the same data requests shown here.
