{"schema":"sb.node.v1","id":"sb:e/5058898917ec55cc4e0b8c27","kind":"entity","label":{"text":"tensor processing units","locale":"en"},"identity":{"anchor":"label:en:tensor processing units","strength":"label"},"asOf":{"knowledgeCutoff":"2026-10-08T20:34:57.810068+00:00","snapshot":"deploy 04403dd1b6ff6ffe00b03195c3da75dd204003f6"},"lens":{"kind":"public"},"build":{"plane":"projected","builtFrom":["entity_graph.slim.json","entity_receipts (deploy asset)","event_graph.json","truth_records (deploy asset)","corroborated_record.json"],"rule":"node-projection.entity@v1"},"standing":{"evidence":{"supporting":42,"separatePublishers":24,"basis":"documents that mention the subject and the outlets that carried them; a mention is evidence the referent is in use, not evidence about any claim"},"admission":{"state":"provisional","reason":"entity graph provenance 'observed'; the node carries no promotion receipt in v1","policy":"EQ-PROMOTION (receipt not carried)"},"disposition":{"state":"served","reason":"public record"},"validity":{"from":"2026-09-03","to":"2026-10-07","basis":"first and last day we saw it mentioned; not the subject's own lifetime"},"measurement":{"state":"partial","blind":["assertion coverage 0.143: about 86% of mentions carry no typed assertion"]}},"dimensions":{"domains":["auto"],"activeDays":15,"assertions":{}},"up":[],"upNote":"An entity is a root: it is contained by nothing. Pan out by following R.","children":{"count":12,"byKind":{"observation":12},"items":[{"id":"sb:o/32d2fa83bf2d9df4306ac2d5","kind":"observation","label":"Network World · 2026-10-07 · Arista steps up its Ethernet challenge to Nvidia’s AI scale-up networking fabric","url":"https://www.networkworld.com/article/4232036/arista-steps-up-its-ethernet-challenge-to-nvidias-ai-scale-up-networking-fabric.html"},{"id":"sb:o/d2fbd9385718f762b78f32a9","kind":"observation","label":"Fast Company · 2026-10-04 · Why tech companies are racing to put AI data centers in space","url":"https://www.fastcompany.com/91617037/why-tech-companies-are-racing-to-put-ai-data-centers-in-space"},{"id":"sb:o/de64b091c354a9e8ac11f11d","kind":"observation","label":"Tech Times · 2026-10-03 · Kyrgyzstan Launched First Satellite for $130,000: SpaceX Rideshare Rewrites National Space Economics","url":"https://www.techtimes.com/articles/328489/20261002/kyrgyzstan-launched-first-satellite-130000-spacex-rideshare-rewrites-national-space-economics.htm"},{"id":"sb:o/b2a9ff7e8f26d6a4a1b483bb","kind":"observation","label":"analyticsinsight · 2026-10-02 · Google’s Project Suncatcher Tests AI Chips in Space","url":"https://www.analyticsinsight.net/news/googles-project-suncatcher-tests-ai-chips-in-space"},{"id":"sb:o/e8e46c9c5d07f5275d3fe8f6","kind":"observation","label":"Denkstrom · 2026-10-02 · Anthropic sichert sich rund eine Million TPU-Chips","url":"https://denkstrom.org/artikel/anthropic-sichert-sich-rund-eine-million-tpu-chips/"},{"id":"sb:o/02b3ffb8f165c078302fbc29","kind":"observation","label":"Tekedia · 2026-10-02 · Alphabet to Test AI Chips in Orbit Via SpaceX as Google Pushes Toward Space-Based Data Centers","url":"https://www.tekedia.com/alphabet-to-test-ai-chips-in-orbit-via-spacex-as-google-pushes-toward-space-based-data-centers/"},{"id":"sb:o/ed895b019a8b9ad89a440e98","kind":"observation","label":"NASASpaceflight · 2026-10-01 · Starfish Otter, other payloads launch on Transporter 18","url":"https://www.nasaspaceflight.com/2026/10/transporter-18/"},{"id":"sb:o/6a48a328a220a06690b74ca5","kind":"observation","label":"Crypto Briefing · 2026-10-01 · Broadcom to backstop up to $42B for Anthropic infrastructure leases","url":"https://cryptobriefing.com/broadcom-anthropic-42b-infrastructure-lease-backstop/"},{"id":"sb:o/936c774555c8348aef4ff313","kind":"observation","label":"CSET (Georgetown) · 2026-09-29 · Tracking AI Chips: What Does It Cost?","url":"https://cset.georgetown.edu/article/tracking-ai-chips-what-does-it-cost/"},{"id":"sb:o/acbc7a37e92968a482cb10cd","kind":"observation","label":"Latest from Tom's Hardware · 2026-09-29 · Anthropic CEO described Jensen Huang as 'kind of Trump-like' during 2022 meeting","url":"https://www.tomshardware.com/tech-industry/anthropic-ceo-described-jensen-huang-as-kind-of-trump-like-during-2022-meeting-nvidia-ceo-reportedly-called-anthropic-executive-a-bean-counter-after-google-tpu-comparison"}],"cursor":"c1.eyJpZCI6InNiOmUvNTA1ODg5ODkxN2VjNTVjYzRlMGI4YzI3IiwibCI6ImNoaWxkcmVuIiwibyI6MTAsImsiOm51bGwsImEiOiIyMDI2LTEwLTA4VDIwOjM0OjU3LjgxMDA2OCswMDowMCJ9","order":"newest first; we keep the 12 newest receipts for this entity. Its other observations are counted in standing.evidence.supporting but are not addressable in v1"},"relations":{"count":10,"byKind":{"co-mentioned":10},"items":[{"kind":"relation","label":"co-mentioned with google (42 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/39e3931746310b6a0b088529"},{"kind":"relation","label":"co-mentioned with ai (41 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/69213f484faa5fb2019fce18"},{"kind":"relation","label":"co-mentioned with tpus (33 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/c0ccc10e3d82424ce7917142"},{"kind":"relation","label":"co-mentioned with project suncatcher (24 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/fd974a8c7bb26fc44d547afe"},{"kind":"relation","label":"co-mentioned with spacex (24 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/556301ed26ed5219f95f18a9"},{"kind":"relation","label":"co-mentioned with earth (20 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/2f930679a011ca4aadcc9467"},{"kind":"relation","label":"co-mentioned with nvidia (19 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/c9300b35f443b5ecbe80f009"},{"kind":"relation","label":"co-mentioned with transporter-18 (18 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/7e991314d42b59aeec9ef6de"},{"kind":"relation","label":"co-mentioned with gemini (16 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/6f86897b9158d8b88e806e4b"},{"kind":"relation","label":"co-mentioned with falcon (15 documents)","role":"co-mentioned","relationKind":"co-mentioned","members":2,"id":"sb:r/5f0bca201e36320224a31a61"}],"cursor":null,"order":"situations first, then claims (the subject's own maintained record, then site-wide corroborated clusters; each label ends with its standing), then co-mentions by shared documents, descending. Co-mentions are the ten strongest the entity graph keeps per entity"},"edge":{"state":"continues"},"links":{"rest":"/api/v1/node/e/5058898917ec55cc4e0b8c27","mcp":"traverse {\"id\":\"sb:e/5058898917ec55cc4e0b8c27\"}","page":"/entity/tensor-processing-units/"},"limits":{"levelsPerCall":1,"pageSize":10,"maxPageSize":50,"membersMax":50,"statement":"One level per call: depth is 1 and only 1, and there is no whole-graph dump. pageSize 1-50 (default 10); a page limit is a cursor, never an edge. Members are listed up to 50. At most 60 traverse calls a minute per caller (runaway-loop protection, not a meter). One knowledge cutoff is served: the deploy's."},"actions":[{"verb":"unfold","lit":true,"effect":"read","layer":"cantina","call":"traverse {\"id\":\"sb:e/5058898917ec55cc4e0b8c27\",\"level\":\"children\"}"},{"verb":"fold","lit":true,"effect":"read","layer":"cantina","call":"traverse {\"id\":\"sb:e/5058898917ec55cc4e0b8c27\"}"},{"verb":"panOut","lit":false,"effect":"read","layer":"cantina","reason":"An entity is a root: it is contained by nothing. Pan out by following R."},{"verb":"follow","lit":true,"effect":"read","layer":"cantina","call":"traverse {\"id\":\"sb:e/5058898917ec55cc4e0b8c27\",\"level\":\"relations\"}"},{"verb":"receipts","lit":false,"effect":"read","layer":"cantina","reason":"receipts are the observations below this node; unfold to reach them"},{"verb":"edge","lit":true,"effect":"read","layer":"cantina","call":"traverse {\"id\":\"sb:e/5058898917ec55cc4e0b8c27\"}"},{"verb":"asOf","lit":false,"effect":"read","layer":"cantina","reason":"v1 serves one knowledge cutoff, this deploy's; walking an earlier cutoff needs retained history (EQ-REPLAY), which v1 does not keep"},{"verb":"switchObserver","lit":false,"effect":"read","layer":"cantina","reason":"the public record is the same for every observer; traversal through a Watch's lens is not served in v1"},{"verb":"readOriginal","lit":false,"effect":"read","layer":"cantina","reason":"no translation is applied on this node; every label is the source's own text"},{"verb":"save","lit":false,"effect":"watch-declaration","layer":"workshop","reason":"saving applies to a configured Watch; configure one around this node first (free), then save it"},{"verb":"deliver","lit":false,"effect":"watch-declaration","layer":"workshop","reason":"delivery applies to a saved Watch"},{"verb":"format","lit":false,"effect":"watch-declaration","layer":"workshop","reason":"output format applies to a saved Watch"},{"verb":"deltaR","lit":false,"effect":"read","layer":"workshop","reason":"Delta R needs a saved Watch with a previous reading"},{"verb":"read","lit":false,"effect":"read","layer":"docking-bay","reason":"one-time Reads are parked until the Commander decides their shape (#938)"}]}