> daily_signal(2026_07_17)

New Jersey signed a law making data centers fund home energy upgrades for ordinary ratepayers — Governor Sherrill's Data Center Fair Share Act, part of a package the state says will save New Jerseyans over $1 billion a year.

PickBits Daily Signal · Friday, July 17, 2026

1. New Jersey signed a law making data centers fund home energy upgrades for ordinary ratepayers — Governor Sherrill's Data Center Fair Share Act, part of a package the state says will save New Jerseyans over $1 billion a year.

This is the freshest, highest-composite AI-adjacent story in today's pool with a clean fetchable primary, and it is the constructive mirror image of the data-center-fights-your-town story: instead of a moratorium, New Jersey is making the AI-driven data-center boom pay residents directly. Governor Mikie Sherrill signed a package of three energy bills on 2026-07-07, the centerpiece being the Data Center Fair Share Act (S731/A796, sponsored by Sen. Burzichelli and Assemblyman Bailey), which creates a separate utility rate class for large data centers and establishes a retail program letting a data center offset its grid-capacity obligation by paying to fund customer-sited distributed energy resources — efficiency upgrades, demand response, behind-the-meter battery storage, and electrification — elsewhere on the system. Trade press covering the mechanism in more detail (Canary Media, 2026-07-16) frames this plainly: data centers can pay for households' inefficient electric-resistance heaters to be swapped for heat pumps, or fund rooftop solar and batteries, in exchange for priority in the interconnection queue. The administration's own numbers are the load-bearing fact: the full three-bill package is projected to save New Jersey ratepayers more than $1 billion annually, including a flat $25 credit to all 3.6 million ratepayers and a $150 credit for lower-to-moderate-income families, arriving as the state saw electricity rates climb more than 16% and 128,828 shutoffs in 2024. The honest caveat: the enabling statute itself does not spell out 'heat pumps' by name — it authorizes the Board of Public Utilities to design the demand-flexibility program, so the household-upgrade mechanism trade press is describing is the intended use of a fund whose exact rules are still being written.

Key fact: IF YOU OWN OR RENT A HOME IN NEW JERSEY WITH AN OLD ELECTRIC-RESISTANCE HEATER, WATCH FOR THE BOARD OF PUBLIC UTILITIES' PROGRAM RULES AND APPLY WHEN THEY OPEN. The Data Center Fair Share Act authorizes the BPU to design the demand-flexibility program that funds household heat-pump swaps and solar-plus-battery installs using data-center payments, but the specific application process and rules were not yet published as of this signing. Bookmark the BPU's site (nj.gov/bpu) and check back over the next few months for the program's launch; in the meantime, ask your utility whether you already qualify for the flat $25 ratepayer credit or the $150 lower-to-moderate-income credit from the broader package, since those apply regardless of the demand-flexibility program's rollout timeline.

nj.gov · rewiringamerica.org · primary source

2. Sunrun is testing AI compute nodes the size of a desktop computer inside its solar-and-battery customers' homes — paying homeowners to host the hardware powering the AI boom, with 1.1 million existing customers as its rollout base.

This is the constructive counterpoint to story 1, and it belongs in the slate because it answers the same power-grid strain with a completely different mechanism: instead of a state paying to retrofit homes for efficiency (New Jersey's approach in story 1), put tiny data centers inside houses that already generate and store their own power. Sunrun, the residential solar-and-battery company, is piloting a 'distributed data center' program that places small computing nodes — about the size of a small desktop computer — inside the homes of customers who already have Sunrun solar panels and battery storage installed. The nodes are managed to work around each household's own energy use, electricity rates, and grid-service programs, and Sunrun compensates homeowners for hosting the equipment, though it has not disclosed specific payment amounts. The company says it is 'actively in discussions with enterprise compute offtakers, homebuilders, and utility partners' but has not named a specific tech company partner yet. The number that matters for scale: Sunrun already has more than 1.1 million existing solar-and-battery customers, which is the addressable base it could tap if the pilot — expected to run over the coming months — proves out. For an IT-pro or policy-aware reader, the structural read is that this is a live test of whether residential distributed energy resources — the same rooftop solar and home batteries story 1's New Jersey law is trying to fund — can double as edge compute capacity for the AI industry, turning millions of individual homes into a genuine alternative to the hyperscale data-center model altogether.

Key fact: IF YOU OWN OR ARE CONSIDERING RESIDENTIAL SOLAR AND BATTERY STORAGE (SUNRUN OR OTHERWISE), ASK YOUR PROVIDER DIRECTLY WHETHER A DISTRIBUTED-COMPUTE PROGRAM LIKE THIS APPLIES TO YOU. Sunrun's pilot is aimed at its existing 1.1 million+ solar-and-battery customers, and the company says it's actively lining up partners now — meaning enrollment terms and compensation details are likely to firm up over the next few months. If you're a Sunrun customer, contact them directly (sunrun.com) to ask whether your home qualifies and what hosting a node would involve (space, power draw, contract length, compensation); if you use a different provider, ask whether they're evaluating a comparable program, since this is likely to become a competitive feature among residential solar companies.

utilitydive.com · sunrun.com · primary source

3. Spotify started rolling out an in-app AI chatbot Tuesday that Premium subscribers can talk or type to for playlists and recommendations — Spotify's own announcement calls it 'a Gemini/ChatGPT-esque chatbot inside of Spotify's homepage.'

This is the clearest day-to-day, literally-everyone-with-a-Spotify-Premium-account story in today's pool: on 2026-07-14, Spotify began a gradual beta rollout of 'Talk to Spotify,' a conversational AI feature built into the Home screen and Now Playing view of its mobile app. Premium subscribers 18 and older in the US, Ireland, and Sweden can type or speak requests — 'play some artists I haven't heard before,' 'add some Bad Bunny, just his recent stuff' — and the chatbot builds and modifies playlists, adds tracks to the queue, saves songs, follows artists, and answers questions about release dates, the story behind a song, or a podcast guest's background, all by referencing the user's own listening history and preferences. Spotify's own framing, quoted directly, is unusually blunt for a corporate announcement: the company describes it as putting 'a Gemini/ChatGPT-esque chatbot inside of Spotify's homepage.' For a reader who has never touched an AI chatbot on purpose, this is the moment one shows up inside an app they already use daily, defaulting on for anyone who taps it rather than requiring an active decision to go try ChatGPT or Gemini. The rollout is limited (beta, three countries, English only, Premium-only) but the direction — conversational AI as a default interface layer inside consumer apps, not a separate destination — is the structural story worth tracking for anyone who builds or manages consumer software.

Key fact: IF YOU HAVE SPOTIFY PREMIUM, CHECK YOUR HOME SCREEN THIS WEEK FOR THE 'TALK TO SPOTIFY' PROMPT AND TRY ONE SPECIFIC REQUEST BEFORE FORMING AN OPINION ON IT. The beta is rolling out gradually to US/Ireland/Sweden Premium users 18+, so you may not see it immediately — if you do, try a concrete natural-language request ('build me a playlist for a rainy Sunday' or 'what's the story behind this song') rather than a vague one, since the feature is built to reference your actual listening history and works best with specific asks. If it doesn't appear yet, no action needed; it's a staged rollout, not a removed feature.

9to5google.com · androidauthority.com · primary source

4. The engineer who helped build Apple's FaceID has a new AI model that reads a 15-minute brain scan and flags PTSD, depression, traumatic brain injury, or Alzheimer's — trained on 250,000 hours of recordings from over 100,000 people, and now headed to the FDA.

This closes the slate on the constructive axis: the same category of AI-infrastructure fight running through stories 1-3 — safety failures, grid strain, power politics — pointed instead at something unambiguously aimed at helping people. Hemispheric, co-founded by computational neuroscientist Hagai Lalazar and Gidi Littwin (the deep-learning architect who co-built the facial-recognition technology Apple acquired and shipped as FaceID), emerged from six years of stealth development on 2026-07-15 with $52 million in early-stage funding led by Hanaco Ventures and OneMind/Awareness Capital. Its model, Descartes, is a 6-billion-parameter neural foundation model trained on 250,000 hours of multimodal EEG and behavioral recordings from more than 100,000 people, designed to decode brain activity from a non-invasive dry-electrode EEG headset worn for about 15 minutes while the patient completes simple tasks on a phone — no surgery, no dye injection, no MRI machine. The target conditions are ones where diagnosis today is slow, subjective, or requires specialist access most people don't have: PTSD, mild traumatic brain injury, depression, and Alzheimer's disease. Littwin's stated goal is to make a brain-health read as routine and affordable as a blood test. The honest caveat that belongs in this frame: Hemispheric has presented architectural data to the FDA's Center for Devices and Radiological Health and is pursuing what it calls 'the first objective vital sign for precision brain health,' but no product has been cleared yet and no approval timeline is confirmed — this is a well-funded, technically serious effort at an early regulatory stage, not a diagnostic available to patients today.

Key fact: IF YOU OR SOMEONE YOU KNOW HAS STRUGGLED TO GET A CLEAR DIAGNOSIS FOR PTSD, DEPRESSION, A BRAIN INJURY, OR EARLY ALZHEIMER'S, THIS IS WORTH BOOKMARKING FOR WHEN HEMISPHERIC FILES ITS FIRST FDA SUBMISSION. There is no product available yet — Hemispheric has only presented architectural data to the FDA so far — so there's nothing to sign up for today. Save this for the trigger that matters: watch for a formal FDA submission or clearance announcement from Hemispheric (search 'Hemispheric FDA' or check hemispheric.ai periodically), since that's the point a 15-minute EEG-based screen could become something a doctor can actually order. In the meantime, if diagnosis has been slow through traditional channels, ask your provider whether any existing quantitative-EEG or biomarker-based screening is available now for the specific condition in question.

hitconsultant.net · techbuzz.ai · primary source