Editorial — Today’s theme: the physical and privacy costs of scaling AI. Two stories: one about where hyperscale infrastructure meets local rules and habitats, the other about how adtech is quietly replacing cookies with harder‑to‑block identifiers.
In Brief
Google is allegedly cutting forest to build data centers
Why this matters now: Finnish regulators halted preparatory earthworks after finding hundreds of hectares cleared for Google’s planned data‑center sites, potentially delaying a roughly €13 billion AI infrastructure push.
Finnish authorities told the local project company behind Google’s Muhos and Kajaani sites to stop major preparatory measures because the work “alter[s] the environment” before a mandatory environmental impact assessment, according to reporting by The Guardian. Google says it complied with forestry law and conducted ecological surveys, while environmental groups and locals report extensive logging in areas they believe should have been preserved. The pause is a clear reminder that global tech buildouts run into local permitting, habitat protection, and community scrutiny.
"immediately suspend, and no later than 23 October 2026, all preparatory measures that would significantly alter the environment."
Fingerprints Are the New Cookies
Why this matters now: As browsers and regulation choke off third‑party cookies, adtech is increasingly turning to browser and device fingerprinting to persistently identify users without consented cookies.
The landscape sketch in Roy Fish’s post tracks a predictable market shift: with cookies losing ground, trackers stitch together tiny signals — canvas quirks, fonts, timezones, hardware details — into persistent IDs. That’s important because fingerprinting is stealthier and harder for users to control, and it complicates compliance and fraud prevention. Privacy browsers and regulators are trying to adapt (the UK ICO treats fingerprinting like cookies in ads enforcement), but the technology arms race continues.
"Fingerprints Are the New Cookies"
Deep Dive
Google data centers in Finland — halted earthworks and what comes next
Why this matters now: Google’s planned Finnish data‑center campuses are part of a multi‑billion‑euro AI buildout, and the suspension of preparatory work could set a precedent for how regulators and communities push back on hyperscale siting.
The core dispute is straightforward: Finnish regulators say preparatory construction — tree felling, topsoil stripping, roads and drainage — happened before a required environmental impact assessment, which undermines the purpose of assessing cumulative ecological impacts. Google’s project company counters that the work complied with forestry law and that ecological surveys were performed. That legal nuance (what counts as premature preparatory work vs. lawful forestry activity) is where these fights usually hinge.
Practically, this matters for three reasons. First, the scale: the sites are part of roughly a €13 billion buildout to support Google’s AI expansion, so delays could ripple through timelines and power contracts. Second, infrastructure is not just servers — it’s land, water, and grid capacity. Local communities and ecosystems can be affected long before a rack is racked. Third, precedent: if Finnish regulators press this case and win stricter enforcement, other jurisdictions might follow, forcing hyperscalers to tighten permitting discipline and community engagement.
There are safety‑and‑planning takeaways for operators and policymakers. Companies should anticipate that large, concentrated land changes — even for preparatory works — will draw scrutiny and that doing “surveys” after clearing can look like backfilling compliance. Regulators should be clear about thresholds for preparatory activities and coordinate faster so environmental reviews can keep pace with commercial timelines. For local activists and municipalities, the Finnish pause is a reminder that permitting is a lever: sustained local attention can make a global project slow down long enough for new terms to be set.
Beyond the immediate legal quarrel, watch how utilities respond. Hyperscale campuses negotiate long‑term power deals and often require grid upgrades; if siting becomes more contentious, utilities and regulators will need more transparent processes around capacity planning, environmental tradeoffs, and community compensation. For listeners tracking AI infrastructure: the physical footprint of compute matters now — not just for emissions or power, but for land rights and biodiversity.
Fingerprinting — the quiet adtech fallback that’s getting harder to stop
Why this matters now: Advertisers and fraudsters increasingly rely on browser and device fingerprinting to replace cookies, making cross‑site tracking more persistent and consent harder to enforce.
Cookie deprecation was supposed to nudge the market toward privacy‑friendly solutions; instead, many actors optimized for evasion. Fingerprinting aggregates dozens of small, often public signals to create probabilistic identifiers that survive cookie clearing and private browsing. That makes it effective, and also ethically and legally messy: users don’t easily see or opt out of the signal collection, and blockers that target cookies don’t catch many fingerprinting schemes.
There are technical and policy ripples to watch. On the technical side, browsers can and are adding defenses: randomizing or suppressing high‑entropy signals, or reducing API surface area that leaks device characteristics. On the policy side, regulators are moving toward treating fingerprinting like cookies for ad purposes — which could force disclosure and consent flows to evolve, but also push more tracking into brittle workarounds like on‑device attribution or server‑side heuristics.
For product and privacy teams, the practical move is twofold. First, adopt or support browser mitigations and defensive defaults that reduce identifiable signal leakage. Second, prepare for regulatory scrutiny: document what signals you collect, why, and how you obtain consent. For consumers, the simplest protection remains choosing browsers and extensions that block or randomize fingerprinting vectors; for the industry, meaningful change will require both technical defaults from major browser vendors and clearer legal lines about what counts as permissible measurement.
Closing Thought
Big tech’s infrastructure and tracking strategies are converging on the same problem: scaling capability while pushing friction outward — into ecosystems, communities, and user privacy. Expect more local fights over land and harder legal debates over invisible identifiers; both are shaping how the next wave of AI gets built and measured.