{"id":"https://openalex.org/W7164892782","doi":"https://doi.org/10.48550/arxiv.2606.15870","title":"Google's Training Supercomputers from TPU v2 to Ironwood: Architectural Stability, Scale, Resilience, Power Efficiency, and Sustainability Across Five Generations","display_name":"Google's Training Supercomputers from TPU v2 to Ironwood: Architectural Stability, Scale, Resilience, Power Efficiency, and Sustainability Across Five Generations","publication_year":2026,"publication_date":"2026-06-14","ids":{"openalex":"https://openalex.org/W7164892782","doi":"https://doi.org/10.48550/arxiv.2606.15870"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.15870","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.15870","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.15870","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137766562","display_name":"Norman P. Jouppi","orcid":"https://orcid.org/0000-0003-1765-1929"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jouppi, Norman P.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138738031","display_name":"Sridhar Lakshmanamurthy","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lakshmanamurthy, Sridhar","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5035427716","display_name":"Cliff Young","orcid":"https://orcid.org/0000-0003-2172-1651"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Young, Cliff","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5035204872","display_name":"David Patterson","orcid":"https://orcid.org/0000-0003-1025-7515"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Patterson, David","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.42309999465942383,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.42309999465942383,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10232","display_name":"Optical Network Technologies","score":0.04699999839067459,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10558","display_name":"Advancements in Semiconductor Devices and Circuit Design","score":0.03579999879002571,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/supercomputer","display_name":"Supercomputer","score":0.6431999802589417},{"id":"https://openalex.org/keywords/sustainability","display_name":"Sustainability","score":0.5906000137329102},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5828999876976013},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.557200014591217},{"id":"https://openalex.org/keywords/power","display_name":"Power (physics)","score":0.4438999891281128},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.430400013923645},{"id":"https://openalex.org/keywords/node","display_name":"Node (physics)","score":0.3901999890804291}],"concepts":[{"id":"https://openalex.org/C83283714","wikidata":"https://www.wikidata.org/wiki/Q121117","display_name":"Supercomputer","level":2,"score":0.6431999802589417},{"id":"https://openalex.org/C66204764","wikidata":"https://www.wikidata.org/wiki/Q219416","display_name":"Sustainability","level":2,"score":0.5906000137329102},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5828999876976013},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.557200014591217},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4936999976634979},{"id":"https://openalex.org/C163258240","wikidata":"https://www.wikidata.org/wiki/Q25342","display_name":"Power (physics)","level":2,"score":0.4438999891281128},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.430400013923645},{"id":"https://openalex.org/C62611344","wikidata":"https://www.wikidata.org/wiki/Q1062658","display_name":"Node (physics)","level":2,"score":0.3901999890804291},{"id":"https://openalex.org/C2776257435","wikidata":"https://www.wikidata.org/wiki/Q1576430","display_name":"Bandwidth (computing)","level":2,"score":0.3783000111579895},{"id":"https://openalex.org/C2779585090","wikidata":"https://www.wikidata.org/wiki/Q3457762","display_name":"Resilience (materials science)","level":2,"score":0.37400001287460327},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.34940001368522644},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.32850000262260437},{"id":"https://openalex.org/C552854447","wikidata":"https://www.wikidata.org/wiki/Q131201","display_name":"Sustainable development","level":2,"score":0.32330000400543213},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.32100000977516174},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.3206000030040741},{"id":"https://openalex.org/C75027835","wikidata":"https://www.wikidata.org/wiki/Q1064746","display_name":"Green computing","level":3,"score":0.3116999864578247},{"id":"https://openalex.org/C81860439","wikidata":"https://www.wikidata.org/wiki/Q251212","display_name":"Internet of Things","level":2,"score":0.30090001225471497},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.2800999879837036},{"id":"https://openalex.org/C423512","wikidata":"https://www.wikidata.org/wiki/Q383973","display_name":"Electricity generation","level":3,"score":0.2777999937534332},{"id":"https://openalex.org/C110354214","wikidata":"https://www.wikidata.org/wiki/Q6314146","display_name":"Engineering management","level":1,"score":0.2709999978542328},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.2612000107765198},{"id":"https://openalex.org/C2992444157","wikidata":"https://www.wikidata.org/wiki/Q775079","display_name":"Single point","level":3,"score":0.2522999942302704}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.15870","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.15870","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.15870","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.15870","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/12","score":0.4110665023326874,"display_name":"Responsible consumption and production"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"This":[0],"paper":[1,92],"(to":[2],"appear":[3],"in":[4,70,80,88,107,120,125],"the":[5,39,49,58,95],"July/August":[6],"2026":[7],"issue":[8],"of":[9,16,60,97,145],"IEEE":[10],"Micro":[11],"magazine)":[12],"summarizes":[13],"five":[14],"generations":[15],"Google":[17],"s":[18,41],"TPUs,":[19],"from":[20],"TPU":[21,40],"v2":[22],"to":[23],"Ironwood,":[24],"highlighting":[25],"their":[26],"evolution":[27],"as":[28,57],"scalable,":[29],"resilient,":[30],"power-efficient,":[31],"sustainable":[32],"supercomputers":[33],"for":[34],"AI":[35],"training.":[36],"It":[37,132],"details":[38],"stable":[42],"architecture,":[43],"which":[44],"has":[45],"surprisingly":[46],"easily":[47],"accommodated":[48],"rapidly":[50],"changing":[51],"deep":[52],"neural":[53],"network":[54],"workloads,":[55],"such":[56],"rise":[59],"Transformers.":[61],"Key":[62],"advancements":[63],"over":[64],"eight":[65],"years":[66],"include":[67],"10x":[68],"increase":[69,79,87],"HBM":[71],"capacity":[72],"and":[73,84,104,110,124],"bandwidth":[74],"per":[75,122,128],"node,":[76],"a":[77,85],"100x":[78],"peak":[81],"node":[82],"performance,":[83],"3600x":[86],"supercomputer":[89],"performance.":[90],"The":[91],"also":[93],"discusses":[94],"role":[96],"optical":[98],"circuit":[99],"switches,":[100],"built-in":[101],"self":[102],"test,":[103],"hardware":[105],"replay":[106],"enhancing":[108],"resilience":[109],"how":[111],"TPU's":[112],"environmental":[113],"impact":[114],"is":[115],"reduced":[116],"with":[117],"substantial":[118],"improvements":[119],"performance":[121],"Watt":[123],"carbon":[126],"emissions":[127],"floating":[129],"point":[130],"operation.":[131],"concludes":[133],"by":[134],"identifying":[135],"six":[136],"features":[137],"that":[138],"may":[139],"well":[140],"characterize":[141],"successful":[142],"training":[143],"accelerators":[144],"this":[146],"decade.":[147]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-17T00:00:00"}