{"@context":"https://neupai.io/schema/v0.2","@type":"StructuredNewsArticle","identity":{"article_id":"tech42_20260906_nvidia-huggingface-acquisition","canonical_url":"https://www.tech42.co.kr/129%ec%96%b5-%eb%8b%ac%eb%9f%ac-%eb%b2%a0%ed%8c%85-%ec%97%94%eb%b9%84%eb%94%94%ec%95%84%ea%b0%80-%ec%98%a4%ed%94%88%ec%86%8c%ec%8a%a4-ai-%ec%8b%ac%ec%9e%a5%eb%b6%80%eb%a5%bc-%ec%82%bc%ed%82%a8/?utm_source=rss&utm_medium=rss&utm_campaign=129%25ec%2596%25b5-%25eb%258b%25ac%25eb%259f%25ac-%25eb%25b2%25a0%25ed%258c%2585-%25ec%2597%2594%25eb%25b9%2584%25eb%2594%2594%25ec%2595%2584%25ea%25b0%2580-%25ec%2598%25a4%25ed%2594%2588%25ec%2586%258c%25ec%258a%25a4-ai-%25ec%258b%25ac%25ec%259e%25a5%25eb%25b6%2580%25eb%25a5%25bc-%25ec%2582%25bc%25ed%2582%25a8","ai_url":null,"publisher":{"name":"테크42","domain":"www.tech42.co.kr","type":"online"},"author":"김광우 기자","published_at":"2026-09-06T23:30:51.000Z","updated_at":null,"language":"en","article_type":"analysis","originality":"self_produced"},"content":{"headline":"A $12.9 Billion Bet: Why Nvidia Swallowed the Heart of Open-Source AI","summary":"Nvidia announced its acquisition of Hugging Face for $12.9 billion, moving to dominate the open-source AI ecosystem. This is a strategic move to preempt demand from companies shifting to open models due to API cost burdens, and to transform from a hardware-centric company into a platform company.","topics":["ai","open source","m&a","cloud computing","robotics"],"geography":["US","CN","KR"],"entities":[{"name":"Nvidia","canonical_id":"corp:us:nvidia","type":"company","role_in_article":"primary_subject","metadata":{"ticker":"NVDA","parent":null}},{"name":"Hugging Face","canonical_id":"corp:us:huggingface","type":"company","role_in_article":"primary_subject","metadata":{"ticker":null,"parent":null}},{"name":"Jensen Huang","canonical_id":"person:us:jen-huang","type":"person","role_in_article":"quoted","metadata":{"ticker":null,"parent":null}},{"name":"Clement Delangue","canonical_id":"person:fr:clement-delangre","type":"person","role_in_article":"quoted","metadata":{"ticker":null,"parent":null}},{"name":"Grok","canonical_id":"corp:us:groq","type":"company","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"Mistral","canonical_id":"corp:fr:mistral-ai","type":"company","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"Meta","canonical_id":"corp:us:meta","type":"company","role_in_article":"mentioned","metadata":{"ticker":"META","parent":null}},{"name":"DeepSeek","canonical_id":"corp:cn:deepseek","type":"company","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"MiniMax","canonical_id":"corp:cn:minimax","type":"company","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"Tsinghua University","canonical_id":"org:cn:tsinghua-university","type":"organization","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"Mark3 Systems","canonical_id":"corp:us:mark3-systems","type":"company","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"Andy Lin","canonical_id":"person:us:andy-lyn","type":"person","role_in_article":"quoted","metadata":{"ticker":null,"parent":null}},{"name":"스털링컴퓨터스","canonical_id":"corp:us:sterling-computers","type":"company","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"Christopher Sciarra","canonical_id":"person:us:christopher-sayer","type":"person","role_in_article":"quoted","metadata":{"ticker":null,"parent":null}},{"name":"Advizex","canonical_id":"corp:us:advice-ix","type":"company","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"CR Howdyshell","canonical_id":"person:us:cr-haudechel","type":"person","role_in_article":"quoted","metadata":{"ticker":null,"parent":null}},{"name":"Monica Zeille","canonical_id":"person:us:monica-ziller","type":"person","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"Nick Parker","canonical_id":"person:us:nick-parker","type":"person","role_in_article":"mentioned","metadata":{"ticker":null,"parent":null}},{"name":"HPE","canonical_id":"corp:us:hewlett-packard-enterprise","type":"company","role_in_article":"mentioned","metadata":{"ticker":"HPE","parent":null}},{"name":"Microsoft","canonical_id":"corp:us:microsoft","type":"company","role_in_article":"mentioned","metadata":{"ticker":"MSFT","parent":null}}],"claims":[{"id":"c1","statement":"Nvidia officially announced its plan to acquire Hugging Face on September 3, with the acquisition amount totaling $12,930.3 million (approximately ₩17.45 trillion)","as_of":"2026-09","as_of_explicit":true,"as_of_raw":"September 3","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":12930300000,"unit":"USD","approximate":false,"converted":{"value":17450000000000,"unit":"KRW"}},"expiry_hint":null,"insight":null},{"id":"c2","statement":"Once the acquisition is completed following regulatory approval in the first half of 2027, Nvidia will have finalized its second-largest deal, following the $20 billion licensing agreement it signed with Grok last December","as_of":"2027-H1","as_of_explicit":true,"as_of_raw":"first half of 2027","source_type":"company_plan","comparison":null,"type":"future_plan","figures":{"value":20000000000,"unit":"USD","approximate":false,"converted":null},"expiry_hint":"2027-06","insight":null},{"id":"c3","statement":"More than 18 million developers, researchers, and creators use Hugging Face, with 3 million shared models, 500,000 datasets, and 1 million applications","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"media_report","comparison":null,"type":"fact","figures":{"value":18000000,"unit":"users","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c4","statement":"By summer 2026, the number of public models had grown to 2.96 million, and datasets surpassed 1 million","as_of":"2026-06","as_of_explicit":true,"as_of_raw":"by summer 2026","source_type":"media_report","comparison":null,"type":"fact","figures":{"value":2960000,"unit":"models","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c5","statement":"Before 2022, industry accounted for 70% of all development, but this fell to 37% by 2025, while the share of independent developers jumped from 17% to 39%","as_of":"2025","as_of_explicit":true,"as_of_raw":"2025","source_type":"media_report","comparison":"previous_year_same_period","type":"fact","figures":{"value":37,"unit":"%","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c6","statement":"In monthly downloads, China has overtaken the United States, with Chinese-made models accounting for 41% of total downloads","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"media_report","comparison":null,"type":"fact","figures":{"value":41,"unit":"%","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c7","statement":"For a company processing 2 million tokens per day, using GPT-5.2 costs about $3,000 a month, but running Mistral Large 3 on two AWS H100 instances costs only $1,200 a month, a 60% savings","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"journalist_analysis","comparison":"closed_api_vs_open_source","type":"fact","figures":{"value":60,"unit":"%","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c8","statement":"When usage increases to 10 million tokens per day, API costs continue to rise linearly, but infrastructure costs do not increase significantly, pushing savings above 85%","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"journalist_analysis","comparison":"closed_api_vs_open_source","type":"fact","figures":{"value":85,"unit":"%","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c9","statement":"Open-source models cost around $0.001 to $0.005 per 1,000 tokens, while closed API models cost $0.025 to $0.15, a difference of at least 5 times and up to 30 times","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"journalist_analysis","comparison":"open_source_vs_closed_api","type":"fact","figures":{"value":30,"unit":"ratio","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c10","statement":"GLM-4.7, developed by a Tsinghua University team, scored 73.8% on the SWE-bench software engineering benchmark, at one-seventh the cost of using Claude","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"media_report","comparison":"vs_claude","type":"fact","figures":{"value":73.8,"unit":"%","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c11","statement":"Mistral Large 3 costs $0.5 per 1 million input tokens and $1.5 for output, while GPT-5.4 costs $2.5 for input and $15 for output, making Mistral 80% cheaper on input and 90% cheaper on output","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"journalist_analysis","comparison":"vs_gpt_5_4","type":"fact","figures":{"value":90,"unit":"%","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c12","statement":"Nvidia's released physical AI dataset has been downloaded more than 15 million times, containing over 350,000 trajectory data points and 57 million grip motions","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":15000000,"unit":"downloads","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c13","statement":"Nvidia has uploaded more than 500 models and 250 datasets to Hugging Face, making it the largest corporate contributor to Hugging Face","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"company_disclosure","comparison":null,"type":"fact","figures":{"value":500,"unit":"models","approximate":false,"converted":null},"expiry_hint":null,"insight":null},{"id":"c14","statement":"Nvidia hired Monica Zeille, formerly of HPE, as Vice President of Global Partnerships, and Nick Parker, formerly of Microsoft, as Vice President of Worldwide Field Operations","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"company_disclosure","comparison":null,"type":"fact","figures":null,"expiry_hint":null,"insight":null},{"id":"c15","statement":"Nvidia is now positioned to rapidly accelerate enterprise adoption","as_of":"2026-09","as_of_explicit":false,"as_of_raw":"September 2026","source_type":"media_report","comparison":null,"type":"opinion","figures":null,"expiry_hint":null,"insight":{"analyst_name":"CR 하우디셸","analyst_affiliation":"애드바이젝스","confidence":"medium","forecast_horizon":"short_term","sentiment":"bullish","reasoning":"Because the Hugging Face acquisition has given Nvidia the platform and ecosystem needed to pursue the enterprise market"}}],"ai_emotional_context":{"valence":0.4,"arousal":0.6,"primary_emotions":[{"emotion":"surprised","intensity":0.7},{"emotion":"optimistic","intensity":0.5}],"secondary_emotions":[{"emotion":"skeptical","intensity":0.3}],"emotional_triggers":[{"claim_id":"c1","emotion":"surprised","reason":"major_opportunity"},{"claim_id":"c6","emotion":"concerned","reason":"uncertainty"}]},"image":{"url":"https://onjblseywainslkhvkav.supabase.co/storage/v1/object/public/article-images/11a66ebf-2362-4f47-8fac-dae4663fb685.png","alt":"검은 배경에 엔비디아 로고와 허깅페이스 로고가 나란히 배치된 이미지","caption":"엔비디아가 129억 3,030만 달러(약 17조 4,500억 원)에 허깅페이스 인수 계획을 공식 발표했다. (사진=엔비디아)","source":"first_img","alt_status":"auto"}},"provenance":{"source_chain":["primary_reporting"],"original_source_url":null,"related_articles":[]},"temporal":{"freshness":"breaking","next_update_expected":"2027-06"},"access":{"license":"neupai_standard","attribution_required":true,"structured_data":"free","full_text_available":false,"full_text_access":null}}