Strategic PartnersIssue 034 records
Four first-time sources for Oct. 31, 2026. Infrastructure dominated: Hermantown, Minn. gave final local approval to Google's data center after protests, with concessions but a 500 MW power agreement and three lawsuits still pending, and Michigan's Attorney General sought rehearing of the MPSC approval of DTE's contracts for Google's Van Buren Township data center over ratepayer cost-shifting. In higher education, NTU Singapore signed an exploratory MOU with Google covering Gemini Enterprise in teaching and a governed Google Cloud research sandbox (vendor-partnered, no outcomes). Ahead of the Nov. 3 midterms, an independent commercial study found Gemini contradicted official data in 6.2% of answers on local issues and read the fewest government pages. No new K-12, public safety or HHS evidence met the bar.
- Evidence records
- 4
- Cross-source patterns
- 1
- Evidence classes
- 1 independent reporting1 government evaluation1 vendor claim1 independent research
- Outcomes
- 2 cautionary1 mixed1 emerging
- Source freshness
- 3 new this fortnight1 recent
- Research completed
- 2026-11-02
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
State utility approval of power contracts is becoming the contested gate for Google AI data centers
In Minnesota, Hermantown's final local votes leave the 500 MW Google-Minnesota Power electric service agreement awaiting Public Utilities Commission approval, alongside lawsuits. In Michigan, the Attorney General is challenging the commission's already-granted approval of DTE's 20-year Google contracts over cost aggregation, a liability collateral cap and transmission costs. Interpretation: local zoning approval does not settle whether Google AI capacity is built; long-term utility contract terms and their ratepayer protections are where scrutiny is concentrating.
Operating questionBefore relying on a planned Google AI data center for regional capacity or local revenue, has the state utility regulator approved its power contracts, and how are costs reconciled to protect other ratepayers?
Supporting evidenceCity of Hermantown, Minnesota (Google LLC project; Minnesota Power electric service)Michigan Department of Attorney General (case U-22058 before the Michigan Public Service Commission; DTE Electric; Google)
Evidence ledger
Full record · every source keeps its link and limitations
Hermantown, Minn. council unanimously approves Google data center development, permit and decommissioning agreements; state utility approval and three lawsuits still pending
Late on Friday, Oct. 30, 2026, after two postponements caused by protests, the Hermantown City Council unanimously approved a development agreement, a special use permit and a decommissioning plan for Google's proposed 1.7 million-square-foot data center on more than 200 acres outside Duluth. The votes followed four hours of public comment, mostly opposed, with attendance capped at 85. MPR reports Google dropped a request for up to $80 million in property tax breaks (now an estimated $3 million a year in property tax), agreed to monitor nearby private wells, and cut the project from four buildings to three with a 33-foot height cap. A supporter cited a $210 million decommissioning agreement. Three lawsuits (environmental review, zoning process, land ownership) remain pending, and the Minnesota Public Utilities Commission must still approve a 500 MW electric service agreement with Minnesota Power. This follows the archive's Oct. 28 MPR item on the postponed vote; it is a new article reporting the actual decision.
Why it matters, evidence and limitations
- Why it matters
- Directly relevant to local governments, counties and state utility regulators asked to host Google AI infrastructure: it shows which concessions (tax abatement withdrawn, well monitoring, decommissioning security, scale reduction) a city obtained and that local approval is not the final gate. Limited relevance to agencies buying Google AI services; it concerns hosting infrastructure, not AI use.
- Evidence and measured results
- Single news report of a public meeting, with named council members, residents, the city attorney and a labor-group director quoted. Figures ($80 million abatement withdrawn, $3 million annual tax, 500 MW, 1.7 million sq ft) are reported by MPR; the $210 million decommissioning figure is attributed to a resident supporter, not confirmed from the agreement text. The 500 MW household comparison is attributed to the Citizens Utility Board of Minnesota. No agreement documents were opened.
- Limitations and uncertainty
- Single outlet; terms of the development and decommissioning agreements were not inspected; lawsuits and PUC review could still alter or stop the project. A correction note on the page concerns only the city attorney's name. Interpretation in role guidance is ours.
Michigan Attorney General petitions for rehearing of MPSC approval of DTE's 20-year contracts to serve Google's Van Buren Township data center, citing ratepayer cost-shifting
On Oct. 30, 2026, Michigan Attorney General Dana Nessel filed a petition for rehearing before the Michigan Public Service Commission in case U-22058, in which the MPSC approved DTE's special contracts to serve a Google hyperscale data center in Van Buren Township. The AG argues the MPSC applied the wrong burden of proof and standard of review and accepted a 20-year aggregation approach to comparing ratepayer benefits and costs, under which, per the AG, the MPSC itself concedes other customers will subsidize the data center at least from 2035 to 2038. She also objects to a collateral cap limiting Google and Alphabet liability for costs of service and to unaccounted transmission costs, and asks for annual cost reconciliation or rejection of the contracts. Michigan Public (corroborating) reports the MPSC had said the contracts ensure no costs pass to other customers, that the MPSC declined comment, and that Google did not respond.
Why it matters, evidence and limitations
- Why it matters
- Directly relevant to state utility commissions, attorneys general and consumer advocates evaluating power contracts for Google AI data centers, and to local governments whose approvals depend on those contracts. It is an adversarial filing, not an adjudicated finding; the outcome of rehearing is unknown.
- Evidence and measured results
- Primary state government press release describing the AG's own petition; the petition PDF itself was not opened. Key quantitative element (subsidy period 2035-2038) is the AG's characterization of the MPSC order. Michigan Public article (opened) corroborates the filing and adds MPSC and Google non-comment.
- Limitations and uncertainty
- Advocacy position of one party; MPSC and DTE positions are summarized only via Michigan Public. The petition PDF and MPSC order were not inspected. The michigan.gov page returned HTTP 403 to curl and was read through a web-fetch tool. Interpretation in role guidance is ours.
NTU Singapore and Google sign MOU expanding Gemini Enterprise use in teaching, a secure Google Cloud research sandbox, and Southeast Asian language models with AI Singapore
NTU Singapore and Google signed a memorandum of understanding on Oct. 30, 2026 to explore initiatives in four areas: campus-wide AI integration, regional language and sovereign model development with AI Singapore (AISG), frontier scientific research, and AI safety. NTU states its undergraduates have had access since August 2026 to Gemini Enterprise, Google AI Studio and Vertex AI. Under the MOU NTU will explore integrating Gemini Enterprise with its NTU Learn platform, supports a goal of embedding AI into 40% of undergraduate courses by 2030, and will design and deploy a secure research sandbox on Google Cloud Platform with pre-approved environments to reduce provisioning bottlenecks. The partners plan Southeast Asian language datasets and models using Gemma and MedGemma, with AISG continuing independent evaluation frameworks, and joint research on post-quantum cryptography, deepfake detection and model privacy.
Why it matters, evidence and limitations
- Why it matters
- Relevant by analogy to U.S. research universities and state systems considering Google Cloud research sandboxes and Gemini Enterprise integration with learning management systems. It is an MOU with "explore" language and no outcomes; Singapore governance, funding and national AI programs limit transferability.
- Evidence and measured results
- Institutional press release with quotes from the NTU provost and a Google vice president. No budget, contract value, user counts, adoption metrics or evaluation results are given. The only quantitative target is 40% of undergraduate courses embedding AI by 2030.
- Limitations and uncertainty
- MOU uses exploratory language; nothing is reported as delivered beyond pre-existing undergraduate tool access. Vendor-partnered announcement. International source with limited U.S. transferability. Interpretation in role guidance is ours.
Searchable midterm study: 1 in 12 AI chatbot answers on local political issues contradicted official data; Gemini false in 6.2%, Google AI Mode 2.9%
Searchable asked ChatGPT, Gemini, Google AI Mode and Perplexity 106 questions about local political issues in five U.S. metro areas and graded 7,449 answers against official government data. Overall 8.2% (610) of answers were false. Perplexity was false 22.6% of the time, Gemini 6.2%, ChatGPT 4.4% and Google AI Mode 2.9%. Gemini was most often wrong on cost of living and housing (13.0%) and schools (8.7%). Gemini read an average of 2 pages from 2 websites per answer, the fewest of the four, and drew under a third (32.2%) of its pages from government websites, the lowest share, with 15.1% from company websites. Why now: the page was updated Oct. 7, before the last google run, but the Nov. 3, 2026 midterms are three days after this edition, making its findings on Google tools newly relevant to election offices and civic-information teams.
Why it matters, evidence and limitations
- Why it matters
- Relevant to state and local election offices, communications teams and agencies that publish official data residents may access through Google AI products: it quantifies how often Gemini and Google AI Mode misstate local official figures and how little Gemini relies on government pages. It evaluates consumer chatbots, not Google Public Sector enterprise deployments grounded on agency data.
- Evidence and measured results
- Structured test: 106 questions (59 single-answer factual, 47 built from fact-checked false claims), multiple prompt forms, each asked five times in fresh sessions via anonymized logged-in accounts (not APIs); an answer counted false only when two graders agreed, with correct answers confirmed by two researchers against official sources (Census, BLS, FBI, EIA, CDC, NAEP). 43,079 pages read were classified by type. Not peer reviewed; model versions and test dates are not stated on the page.
- Limitations and uncertainty
- Produced by a company that sells AI-visibility products, a commercial interest in showing chatbot errors; five metros only; model versions, dates of testing and confidence intervals not disclosed; consumer interfaces only. Published before the previous google run; included for election timing. Interpretation in role guidance is ours.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.
- Government evaluation
- A public body’s measured evaluation or documented pilot.
- Vendor claim
- A supplier-provided assertion that has not been upgraded to independent evidence.
- Independent research
- Research conducted outside the implementing organization.