Editorial clarification · October 24, 2026: Source-ID matches, acquisition-year clustering and corporate-control attribution have been distinguished. Unreproduced statistical scores have been removed; sequential identifiers do not establish joint registration.
In 2021, Forbes published an investigation into Sun Guangxin — a Xinjiang real estate tycoon, former People's Liberation Army official with close ties to the Chinese Communist Party — and his acquisition of roughly 140,000 acres in Val Verde County, Texas. The story broke through on national security grounds, triggering congressional concern, state legislative action, and a wave of coverage about foreign land ownership near sensitive U.S. infrastructure.
Source: John Hyatt, "Why A Secretive Chinese Billionaire Bought 140,000 Acres Of Land In Texas," Forbes, August 9, 2021 (full text republished by the Devils River Conservancy).
It was excellent journalism. It named a real actor, documented a real risk, and moved the policy conversation.
It also arrived years after the data was already filed with the federal government.
That gap — between public disclosure and public awareness — is exactly what AFIDA Watch was built to close.
The headlines named the man. The filings name the entities. Forbes, quoting GH America's own spokesman, identifies Brazos Highland Properties LP as 'the entity Sun used to acquire land in Val Verde County' — and reports that through Brazos and another subsidiary, Harvest Texas LLC, Sun controls nearly 7% of all land in the county. NPR describes the two companies as the fronts through which Sun holds his Texas land. Those are the exact entities in the AFIDA filings: the China cohort's eight Val Verde parcels are held by Brazos Highland Properties, LP (Owner ID 34106) and Harvest Texas, LLC (Owner ID 34107). The cohort analysis didn't find a different pattern — it surfaced the same named holders. The corporate-control attribution comes from the cited reporting, not from source-ID matching alone. What follows is what the filings add that the reporting didn't: when the land moved, how the scale exploded, and what the land is classified as.
Foreign land ownership in Texas runs deep. AFIDA records show the first foreign agricultural land holding in Texas dates to 1911 — that is the start of all Texas foreign holdings in the dataset.
The adversary lens is a narrower cut.
When AFIDA Watch screens for adversary-linked ownership — entities and individuals associated with nations of concern under current U.S. national security frameworks — the Texas cohort begins in 1978. That starting point matters. It means adversary-linked agricultural land interest in Texas predates many of today's foreign investment disclosure mechanisms by decades.
The China cohort within that adversary-linked universe is more recent, more concentrated, and more statistically anomalous.
The descriptive cohort groups selected records by reported country code, county and acquisition year. Matching names or source IDs does not independently resolve corporate control. Statistical screening results require a published population, calculation and reproducible method before they can support a quantitative claim.
The Val Verde cluster is one such pattern.
The China cohort in the AFIDA Watch Texas dataset contains 12 parcels. Here is the exact yearly breakdown: Year Parcels Acres Transaction value
2016 1 52.18 $172,194
2017 6 60,497.86 $66,731,420
2018 2 26,711.59 $24,854,396
2019 3 29,704.95 $41,524,468
The burst wasn't just volume — it was scale. Average parcel size leapt from 52 acres in 2016 to over 10,000 in 2017, peaking near 13,400 in 2018. Fewer transactions, vastly larger footprints.
And the land type sharpens the question: 99.8% of the cohort's acreage sits in 'Other Agriculture' — a category that by the government's own definition includes farmsteads, farm roads, idle land, and aquaculture. All eight Val Verde parcels fall entirely in that category; only the four small Hunt and Navarro parcels show any crop or pasture. The 2024 filings don't say which of those uses it is — the form's 'describe the use' field didn't exist until 2025.
2017 is the burst year. Six parcels. Over 60,000 acres. More than $66 million in a single calendar year.
The reported acquisition-year concentration is descriptive. The previously quoted z-scores have been removed because this article does not publish the comparison population, formula and reproducible calculation. Concentration alone is not a validated security-risk measure.
Eight of the 12 parcels in the China cohort are located in Val Verde County. The remaining four are concentrated in Hunt and Navarro Counties.
That geographic clustering — two distinct concentration zones, Val Verde and the Hunt/Navarro corridor — is itself a data signal. County clustering alone does not establish related entities or a coordinated strategy; those conclusions require transaction and corporate records.
The Val Verde holdings represent the dominant mass. The Hunt/Navarro parcels are the overhang.
Two legal entities account for the bulk of the Val Verde cohort's disclosed transaction value:
Brazos Highland Properties LP (Owner ID 34106): $90,918,251.32
Harvest Texas LLC (Owner ID 34107): $41,524,469.79
In the AFIDA system itself, the two entities carry sequential Owner IDs — 34106 and 34107 — adjacent source identifiers. Sequential IDs alone do not establish simultaneous registration, common control or coordination.
Together, these two entities represent the structural spine of the Val Verde cluster.
The Hunt and Navarro County parcels are held by four named individual owners: Li Wang, Simiao Cheng, Zhan Shan Wang, and Jian Ping Chen. The geographic and structural split between the LP/LLC holdings in Val Verde and the individual-ownership parcels in Hunt/Navarro is itself a characteristic pattern in adversary-linked land cohort analysis — institutional vehicles for large-acreage assembly, individual registrations for smaller footprint acquisitions in secondary counties.
AFIDA Watch applies a four-step deterministic screening protocol to every entity in the dataset: Step 1 — Entity Resolution
Cross-reference ownership names against known entity networks, beneficial ownership registrations, and related-party filings. Resolve LLCs, LPs, and individual names to their underlying ownership structures where possible.
Step 2 — Sanctions Screening
Screen resolved entities against OFAC SDN, BIS Entity List, and relevant State Department designations. Flag direct matches and proximity matches within the entity graph.
Step 3 — Enhanced Due Diligence
For cohort-flagged entities, apply deeper review: corporate formation records, registered agent networks, historical acquisition patterns, and cross-parcel ownership overlap.
Step 4 — DOD Proximity Scoring
Score each flagged parcel against proximity to Department of Defense installations, critical infrastructure corridors, and sensitive airspace. Val Verde County parcels score in the elevated tier on this dimension due to proximity to Laughlin Air Force Base.
This workflow is systematic, repeatable, and entirely grounded in public records. It does not require investigative tips. It does not require leaks. It runs on disclosed data.
Forbes did the right thing. The 2021 investigation was well-reported, well-sourced, and consequential. Investigative journalism found something real and told the public about it.
The purpose of this article is not to critique that work. The purpose is to demonstrate a different capability: deterministic screening that surfaces the same finding from public records, without waiting for a tip.
The AFIDA filings that documented Sun Guangxin's Val Verde acquisitions were available at the time of acquisition. The z-score anomaly in 2017 was calculable the day those disclosures were filed. The entity structure — Brazos Highland Properties LP, Harvest Texas LLC — was public.
The data was always there. The system to read it at scale wasn't.
The conservationists and lawmakers who raised the alarm did so without this picture — they fought the project on environmental and security grounds. The filings validate their concerns and sharpen them: a concentrated 2017 burst, a 193-fold explosion in average parcel size, and a land-use classification that remains opaque — 99.8% of the acreage sits in a category the filings don't specify further.
That system now exists.
This article is the first in an ongoing AFIDA Watch series focused on Texas foreign land ownership.
The Val Verde cohort is the established example — the case where journalism and data screening converge on the same finding. Future reports will move beyond convergence and into new territory.
Next report: Beyond Val Verde — the 1,171 parcels under adversary-coded ownership most Texans don't know about.
Editor's Note: This is Report 1 of the AFIDA Watch Texas Series — an ongoing, sequentially published analysis of foreign agricultural land ownership in Texas, screened through an adversary-linked cohort methodology. Each report will advance the dataset, expand the geographic scope, or introduce a new analytical dimension. The series is designed to demonstrate what systematic public-records screening can surface — and what policymakers, journalists, and researchers can do with it.