Virginia DEQ's critique of the Vantage data-center health study | Magica
Emails show Virginia DEQ prepared a pre-release critique of Vantage data-center health study
Editorial Team
••📖6 min read
Emails show Virginia environmental officials began developing a critique of a health-impact study of Vantage Data Centers VA2 before it was published. The disagreement is less a single factual contradiction than a clash between a full-permit screening scenario and DEQ's estimate of the facility's reported 2025 operations.
Virginia DEQ was preparing a "defensible critique" of a Vantage Data Centers VA2 health-impact analysis before the study was released publicly.
The commissioned analysis models the facility at its permitted annual emissions ceiling; DEQ models estimated 2025 emissions from reported turbine use, test data and permit rates.
Both sides leave the key future question open: how long VA2 will rely on on-site turbines before a grid connection is available, and what emissions it will produce in the meantime.
Internal emails show that Virginia's air regulator was preparing its response to a health-impact analysis of Vantage Data Centers VA2 before the analysis was published. The timing matters because VA2, the Vantage-owned Sterling facility with eight natural-gas turbines, is the only Virginia data center described in the reporting as powering itself.
Michael Rolband, director of the Virginia Department of Environmental Quality, the state agency that issued VA2's air permit, forwarded the Piedmont Environmental Council's forthcoming study to air-permitting staff two days before its March 4 release. In the account based on the internal emails, Rolband later wrote that the agency needed a critique that could withstand public and critical review. Some staff objected that evaluating health effects was outside their expertise.
That record does not determine whether the health estimates are right. It does show that the permit issuer became an early participant in a public argument over the consequences of the emissions it authorized.
VA2 began operating its combustion turbines in April 2024, according to DEQ's April 27 assessment. DEQ's permit inventory lists eight 16.5-megawatt simple-cycle gas turbines, 40 emergency diesel generator sets and two black-start diesel units. The Piedmont report describes 49 emergency sets and two black-start units; neither retained source reconciles that equipment-count discrepancy.
Alphabet’s proposed $80 billion equity package is partly tied to AI infrastructure, but its $40 billion at-the-market program is primarily intended to cover employee-equity tax obligations. Together with Microsoft’s $190 billion 2026 capital-expenditure forecast, the disclosures show a more complicated shift in how hyperscalers fund constrained compute capacity.
Editorial Team
The larger difference is the emissions case each analysis asks readers to consider.
Analysis
Emissions basis
What its result can show
Piedmont-commissioned analysis
Facility-wide annual permit limits: 56.51 tons of primary PM2.5, 95 tons of nitrogen oxides, 19.21 tons of sulfur dioxide and 42.87 tons of volatile organic compounds
A bounding, full-permit screening scenario—not a forecast of normal dispatch, individual illness or regulatory compliance
DEQ assessment
Reported 2025 turbine use: 47,131 combined normal-operation hours and 637 uncontrolled hours, combined with 2024 test results where available and permit rates where they were not
DEQ's estimate of one year's facility emissions, not a guarantee of future utilization
The commissioned analysis explicitly says actual impacts would scale with actual annual emissions. Its full-permit scenario uses InMAP to estimate annual-average fine-particle exposure, including particles formed downwind from pollution precursors, and EPA's COBRA tool to estimate population-level health outcomes. It puts annual damages at $53 million to $99 million in 2023 dollars and estimates 3.4 to 6.5 additional premature deaths a year. Those are model outputs for a population under that specified emissions scenario, not observed deaths or a finding about the risk to a particular resident.
DEQ says the screening scenario substantially overstates current emissions. Its facility-wide 2025 estimate is 38.9 tons of PM2.5, 44.7 tons of nitrogen oxides, 13.3 tons of sulfur dioxide, 20.3 tons of carbon monoxide and 12.1 tons of volatile organic compounds—28% to 69% of the applicable permit limits. The agency calls the year representative, while acknowledging that future annual emissions could change with turbine availability.
The full-permit and 2025 cases are therefore not competing measurements of the same thing. One frames authorized capacity; the other reconstructs an operating year from utilization, testing and permit assumptions.
The modeling dispute has limits on both sides
Piedmont's consultants used InMAP because the question they posed was long-run, regional exposure to both direct and secondary PM2.5. The report says it is a screening-level, population-exposure exercise and does not replace permitting models. It represents the facility's turbines and generators as one aggregated point source rather than separate stacks, and it could not derive facility-specific ammonia emissions from permit documents—an omission it says could understate secondary-particle formation.
DEQ says InMAP is not an EPA-approved model for an individual-source regulatory application and argues that EPA's preferred near-field approach, AERMOD, is the appropriate starting point. It also says a comparison in a different DEQ permitting action found InMAP at least five times higher than an AERMOD-based result at one downwind location. DEQ further contends that COBRA is a regional and national screening tool, not a tool intended to establish the effects of a single stationary source. Those are substantive reasons to treat the report's dollar and mortality figures as scenario estimates, not as an observed health ledger.
But DEQ's assessment also says it does not conduct or evaluate the merits of health-impact calculations because it lacks expertise in that field. Its conclusion that impacts even under Piedmont's modeled concentration increases are unlikely relies partly on federal particulate-matter standards; former EPA air-quality official Michael Koerber told the email investigation that soot can harm health below those standards. The source record thus does not support treating permit compliance, regional monitor readings, or either model as a direct answer to VA2's actual health effects.
After DEQ disclosed its 2025 inputs, Michael Cork, the researcher who conducted the analysis for Piedmont, told the email investigation that he reran his models with those inputs. He said the resulting estimate was $33 million to $63 million in annual health damages and two to four premature deaths. That is an updated, source-attributed model result; the retained materials do not provide an independent validation of it.
The permit is also an infrastructure choice
Vantage told DEQ it plans to shut down the combustion turbines and use grid power as soon as it can. DEQ says the timing depends on Dominion's schedule for transmission upgrades in and around Loudoun County. Vantage's public-policy manager separately characterized the turbines as a temporary response to grid constraints in communications described in the emails.
That makes the permit more than a compliance question. It authorizes on-site generation while the company waits for a different power arrangement, yet it does not establish when that transition will occur. The health study measures the authorized upper boundary; DEQ's analysis supplies one lower operating reference point. Neither establishes how close future operations will come to the ceiling.
What evidence can resolve the dispute
DEQ started a Northern Virginia air-monitoring study on March 2. Its first phase uses temporary sensors for nitrogen dioxide, carbon monoxide and PM2.5 to identify possible hotspots; the agency says those sensor readings cannot determine compliance with national ambient-air standards. Certified mobile instruments could follow in areas identified in that first phase, before DEQ decides on potential permanent sites.
The next useful evidence is more concrete than another exchange over headline estimates: public facility-level emissions records across multiple years, a disclosed timetable for VA2's grid connection, and monitoring designed to distinguish local contributions from regional background pollution. Until those records exist, the fairest reading is that the permit-limit study and DEQ's 2025 estimate describe different, relevant bounds—not a settled measure of harm or safety.
Intel is putting €5 billion into equipment and connections at its operating Leixlip campus to increase output of Intel 3-based Xeon processors. The investment strengthens an existing European manufacturing base, but Intel has not disclosed its added wafer capacity, the split between its own products and foundry work, or an outside customer.
Sundar Pichai said Google supports an industry letter on open models, aligning the company with developers and technology firms resisting broad limits on Chinese AI systems while leaving allegations of IP theft and national-security risks unresolved.
Brazilian courts and bar associations have responded to hidden prompts in legal filings with fines, referrals, suspensions and an investigation. The known cases were detected, however, and do not show that an AI system changed a judicial decision; they instead test how courts secure AI-assisted review of untrusted case documents.
Salamanca City Central School District has put its proposed Realbotix humanoid robot and Optio AI assistant pilot on hold while it works on student-data privacy agreements. The pause leaves the district with a more basic task: establish what the digital tutor and the physical robot each collect, do and improve before either reaches students.
Samsung Electronics and Broadcom have signed an MOU covering memory, foundry technologies and advanced packaging for next-generation AI infrastructure. The companies expect more than $200 billion of collaboration over five years, but have not disclosed how that figure translates into purchases, production or capacity commitments.
OpenAI says GPT-5.6 Sol reached a 24.9% ExploitGym pass rate under a two-hour cap, up from GPT-5.5’s 15.1%. The benchmark measures a bounded step from a known vulnerability to an exploit; a separate assessment performed with OpenAI finds continued limits on hardened targets and long operations.
Chey Tae-won, chairman of SK Group, said Anthropic had sought SK Hynix component-supply cooperation for chips it hopes to make itself. But Anthropic has disclosed no chip plan, and the same summit also produced a separate letter of intent between Anthropic and SK Telecom on a large AI data center.
South Korea announced $950 billion in planned five-year semiconductor cooperation involving Samsung, SK Group and global technology companies, alongside prospective AI-data-center projects. The figure aggregates arrangements with different terms and disclosed commitments, leaving delivery, power and the overlap between the SK and NVIDIA figures unresolved.
OpenAI said it applied a fix after a July 25 disruption affected ChatGPT, its Codex coding service and parts of its API. The available reporting points to recovery in about 50 minutes, but the retained incident records do not explain the cause, the mitigation or whether the services failed through a common component.
Anthropic CEO Dario Amodei said the Claude developer has signed supply deals with Samsung Electronics and SK hynix. The available disclosures show a useful infrastructure relationship, but not an HBM purchase, a dedicated allocation or a claim on either supplier’s output.
Moody's Ratings expects six US hyperscalers to spend $785 billion in 2026 and close to $1 trillion in 2027. The forecast signals continued demand for AI capacity, but a separate Moody's analysis shows why capex alone understates the financing question: five of the companies already reported $969 billion in undiscounted future lease payments, much of it not yet on their balance sheets.
Anthropic has released Claude Opus 5 at Opus 4.8's API price and says it approaches Claude Fable 5 on coding and knowledge work. An Epoch AI listing places it at 159 ECI, below Fable 5 and GPT-5.6 Sol, leaving customers to test whether the cheaper model delivers better economics in their own workloads.
Nvidia plans to invest $1 billion in Naver as Brookfield signs a nonbinding term sheet for up to $9 billion toward a South Korean AI-factory project. The companies target a 200-megawatt expansion by 2028, but financing, definitive terms and construction remain ahead.
Nvidia and SK Group signed letters of intent for an AI-infrastructure initiative they value at more than $500 billion, centered on an up-to-2-gigawatt Korean AI factory and a long-term SK hynix memory partnership. The disclosed projects give the plan a technical shape, but not a capital breakdown, time horizon or basis for treating the headline figure as committed spending.
A transmission-line outage in Northern Virginia prompted data centers to transfer to backup power, removing about 3.1 GW of load in roughly 30 seconds. PJM reported no reliability impact, but the event put the operational behavior of concentrated computing campuses—not just their growing electricity demand—at the center of the grid-planning debate.
Imperial Valley Computer Manufacturing is asking a California court to overturn Imperial Irrigation District’s denial of water for a proposed 330-megawatt data center. The developer says it can offset cooling demand by fallowing nearby farmland, but IID policy places conserved water under the district’s control and requires a separate review of industrial service.
DeepSeek has reportedly told prospective investors that it will not sign second-round agreements in the coming days. The account describes a potentially temporary pause, leaving the size, valuation and timing of the proposed financing unconfirmed.
U.S. technology employers announced 139,156 job cuts through June 2026, 83% more than a year earlier. The figures coincide with huge planned data-center outlays, but they do not show how many jobs AI directly replaced—or whether the announced reductions become a lasting employment decline.
No AI Vancouver’s July 26 rally follows City Council’s withdrawal of a public-hearing referral for Telus and Westbank’s Mount Pleasant data centre. The pause leaves land use unresolved, but TELUS says it has secured the cluster’s initial 85 MW; the key test is now what conditions and evidence the next municipal process can require.
