The world's power grids are a patchwork of incompatible data, a fact that is both a massive climate problem and a very specific business opportunity. Tapestry, a project inside Alphabet's X Moonshot Factory, is betting that the first step to fixing the grid is simply to see it all at once. Its stated mission is to build an AI-powered unified platform that makes the entire electric grid visible, a foundational layer for planners and operators who currently work with fragmented, incomplete models [X, Unknown].
It is a classic X moonshot: a long-term, capital-intensive bet on a foundational technology with no immediate path to revenue. The team is led by General Manager Page Crahan, a former co-CEO of energy analytics firm Clarus Power, and includes Andy Ott, the former President and CEO of PJM Interconnection, the largest grid operator in the U.S. [Public neutral summary]. This pairing of software and deep grid-operator experience is the project's most tangible asset, a signal that Tapestry is being built by people who know where the data silos are buried.
The Wedge of Visibility
Tapestry's approach is not to sell a flashy optimization dashboard, but to first solve the data problem. The grid is managed by thousands of utilities and system operators, each with their own legacy software, naming conventions, and physical models. Integrating new renewable energy sources or assessing system-wide stability often requires manual, error-prone reconciliation of these disparate datasets. Tapestry's proposed wedge is to build what it calls "a first unified model of the grid" [X, Unknown]. This AI model would ingest and harmonize data from across generation, transmission, and distribution networks, creating a single, constantly updated digital twin.
From that foundation, the practical applications could be significant. One early partnership is with PJM itself, exploring how AI could speed up the notoriously slow process of interconnection studies for new power projects [Utility Dive, Unknown]. Another is with Vector, a distribution network company in New Zealand, working on AI models to improve grid resilience [Tapestry, Unknown]. These are not yet commercial deployments, but rather proof-of-concept collaborations that allow Tapestry to refine its models with real-world data.
The Moonshot Economics
Operating as an Alphabet project comes with distinct advantages and constraints. The primary advantage is patient capital; Tapestry does not need to show quarterly growth or justify a burn rate to external investors. It can focus on the technically grueling work of data ingestion and model training without the pressure of a near-term product launch. The constraint is that its destiny is tied to Alphabet's internal priorities. There is no external funding round to signal market validation, and its path to becoming an independent company or a widely deployed Google Cloud service remains undefined.
The project's current phase is best understood as a high-stakes R&D partnership. Its public-facing website is more of a recruiting tool and partnership beacon than a sales page, highlighting open roles for a Chief of Staff and senior engineers [X, Unknown]. The goal appears to be to embed its technology deeply within a few key grid partners, prove the value of a unified model, and then scale from there.
The Incumbent in the Crosshairs
For all its ambition, Tapestry's success will be measured in a very concrete unit: the reduction in megawatt-hours of fossil fuel generation required for grid stability. If its unified model can help operators integrate renewables faster and manage loads more efficiently, the climate math becomes compelling.
Consider a back-of-envelope scenario. PJM's grid covers 13 states and supports about 65 million people. If Tapestry's tools could improve operational efficiency by just one percent across that system, it would save roughly 16 terawatt-hours of generation annually, mostly from avoiding the use of marginal natural gas plants. That's equivalent to the annual electricity consumption of 1.5 million homes.
To achieve that, Tapestry isn't competing with a single startup. Its true rival is the entrenched inertia of the grid's own complexity, and the incumbent vendors like Siemens, GE, and OSIsoft whose proprietary systems have defined the landscape for decades. Tapestry must convince utilities that a unified, AI-native model from a tech giant is a better bet than incremental upgrades to the fragmented legacy stack they already own. It is a bet on a new paradigm, and the meter is running.
Sources
- [X, Unknown] Tapestry - A Google X Moonshot | https://x.company/projects/tapestry/
- [Utility Dive, Unknown] PJM, Google partner to speed grid interconnection using AI | https://www.utilitydive.com/news/pjm-google-tapestry-grid-interconnection-ai/744982/
- [Tapestry, Unknown] How Vector and Tapestry are using AI to build a more resilient power grid for New Zealand | https://www.tapestryenergy.com/en/projects/tapestry-and-vector
- [Public neutral summary] Tapestry company summary
- [X, Unknown] Chief of Staff (Tapestry) | https://x.company/careers/8395275002/