{"id":189195,"date":"2026-10-20T23:15:18","date_gmt":"2026-10-20T23:15:18","guid":{"rendered":"https:\/\/www.europesays.com\/dk\/189195\/"},"modified":"2026-10-20T23:15:18","modified_gmt":"2026-10-20T23:15:18","slug":"ai-stocks-sold-off-ahead-of-nvidias-berlin-gtc-jensen-huangs-keynote-must-answer-wheres-the-return-biggo-finance","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/dk\/189195\/","title":{"rendered":"AI Stocks Sold Off Ahead of Nvidia&#8217;s Berlin GTC; Jensen Huang&#8217;s Keynote Must Answer &#8216;Where&#8217;s the Return?&#8217; \u2014 BigGo Finance"},"content":{"rendered":"<p>Nvidia&#8217;s Berlin GTC conference runs Oct. 20-22, with CEO Jensen Huang scheduled to deliver the keynote at 11 a.m. Berlin time on Oct. 21 \u2014 5 a.m. EDT on Oct. 21 (Wednesday). Against a backdrop of last week&#8217;s selloff in AI-related assets and rising concerns over the debt and financing risks of AI cloud providers, this developer conference \u2014 ordinarily a technical launch event \u2014 has been given the added significance by the market of testing whether AI capital spending can earn a return.<\/p>\n<p>The keynote will be held at the Tempodrom in Berlin, officially scheduled from 11 a.m. to 1 p.m. Berlin time. Converted to the time zones of key readers:<\/p>\n<p>RegionOct. 21 TimeBerlin11:00\u201313:00Taiwan17:00\u201319:00Japan18:00\u201320:00U.S. Eastern05:00\u201307:00<\/p>\n<p>The first day of the conference is mainly training workshops, with the main sessions and exhibitions scheduled for Oct. 21-22. More than 100 sessions cover accelerated computing, open models, robotics and physical AI. Notably, this year&#8217;s conference features as many as 142 external speakers from organizations including Nebius, Mercedes-Benz, Siemens, Schneider Electric and SAP, but core supply-chain players such as Micron, SK Hynix, Samsung, Broadcom, Marvell, Intel and AMD are absent from the main stage.<\/p>\n<p>AI Trade Under Pressure; Investors Now Ask for Commercialization Evidence<\/p>\n<p>On the eve of the conference, the market is putting the AI theme through a stress test. The debt and financing structures of AI cloud computing companies have drawn investor attention, and the market has begun reassessing whether highly leveraged expansion models can continue. At the same time, controversy over the IPO valuations of AI startups is heating up, and whether massive infrastructure investment can be supported by future commercial revenue has become a focus for Wall Street. Last week, AI-related assets broadly weakened, memory stocks fell notably, and capital rotated into defensive and value sectors.<\/p>\n<p>Against this backdrop, investor expectations for this year&#8217;s GTC go beyond new chips and specification parameters to whether Nvidia can more clearly articulate the growth runway for AI computing demand and how massive capital expenditure will translate into sustainable commercial returns.<\/p>\n<p>The strongest support for Nvidia&#8217;s fundamentals remains its data center business. In its fiscal 2027 second quarter, total revenue reached $96.2 billion (about NT$3.1 trillion), up 106% year over year; data center revenue was $89 billion (about NT$2.9 trillion), up 117% and accounting for more than 92% of total revenue.<\/p>\n<p>ItemAmountYear-over-Year GrowthTotal revenue$96.2 billion (about NT$3.1 trillion)106%Data center revenue$89 billion (about NT$2.9 trillion)117%Data center share of total revenueMore than 92%\u2014<\/p>\n<p>Note: Figures are for fiscal 2027 second quarter.<\/p>\n<p>The Blackwell platform is still in the midst of large-scale revenue realization, and Vera Rubin is next in line to take on the next round of platform upgrades.<\/p>\n<p>For Rubin, It&#8217;s the Timeline, Not the Specs<\/p>\n<p>Nvidia&#8217;s strategic focus is undergoing a structural shift \u2014 from emphasizing single-GPU performance in the past to integrating CPU, GPU, networking, storage, software and data center design into a complete &#8220;AI factory.&#8221; This change is closely tied to AI workloads shifting from training to inference: large-scale inference must simultaneously balance latency, throughput, energy efficiency and cost per token, and what customers ultimately buy is no longer just a GPU but a computing system that can lower long-term operating costs.<\/p>\n<p>Looking back at this year&#8217;s cadence, the San Jose GTC in March disclosed details of the Vera Rubin system, the Kyber rack and the next-generation Feynman architecture; the Taipei GTC in June further announced that the Vera Rubin system had entered full-scale production and unveiled the standalone CPU Vera and its companion rack. As a result, the market&#8217;s core expectation for Rubin at this conference is not another round of specification upgrades but the actual procurement expansion timeline from major cloud providers and enterprise customers, and how much the new platform can lower inference costs.<\/p>\n<p>Agentic AI (AI Agent) is another centerpiece of this conference. Unlike traditional chatbots&#8217; one-off Q&amp;A, agent systems need to continuously call models, retrieve data, use tools and handle multi-step tasks; even if the cost of a single inference falls, long-running agents may still significantly increase overall computing volume. Nvidia hopes to use a combination of chips, interconnect equipment, inference software and development tools to spread computing demand from a handful of AI labs to scenarios including customer service, software development, finance, healthcare and enterprise operations.<\/p>\n<p>However, increased computing consumption does not necessarily equal increased commercial value. Speakers on this year&#8217;s security agenda include CrowdStrike CEO George Kurtz, Hugging Face co-founder Thomas Wolf and Red Hat&#8217;s Chris Wright \u2014 a lineup that itself highlights the real-world thresholds for large-scale deployment of agent systems: identity verification, access management and the maturity of enterprise software architecture will determine deployment speed as much as model capabilities.<\/p>\n<p>Physical AI and robotics represent longer-term growth space. This year&#8217;s participants include Agile Robots&#8217; Chen Zhaopeng, Siemens Healthineers&#8217; Arthur Kaindl and Mercedes-Benz Group CEO Ola K\u00e4llenius. Germany&#8217;s vast industrial system makes Berlin a suitable venue for promoting this strategy, but the commercialization cycle for robots and autonomous driving is clearly longer than that for data center chips; product reliability, regulatory approvals and customer validation could all slow revenue realization.<\/p>\n<p>Nvidia&#8217;s official keynote agenda divides the above themes into five chapters:<\/p>\n<p>ChapterThemeSpeakers (Company)Sovereign and Enterprise AICountries and companies building their own intelligencePascal Daloz (Dassault Syst\u00e8mes), Tim H\u00f6ttges (Deutsche Telekom), Arthur Mensch (Mistral AI)AI-Native EnterprisesThe growth engine of European startupsAndreas Cleve (Corti.AI), Jeannette zu F\u00fcrstenberg (General Catalyst), Robin Rombach (Black Forest Labs), David Silver (Ineffable Intelligence)AI Security and ProtectionAccess and protection for enterprise agent systemsBartley Richardson (CrowdStrike), Thomas Wolf (Hugging Face), Chris Wright (Red Hat)AI FactoriesInfrastructure, jobs and economic growthFr\u00e9d\u00e9ric Godemel (Schneider Electric), Siko Rashid (Brookfield), Arkady Volozh (Nebius)Physical and Industrial AIBringing intelligence into the real worldChen Zhaopeng (Agile Robots), Arthur Kaindl (Siemens Healthineers), Ola K\u00e4llenius (Mercedes-Benz), Artem Sokolov (Humanoid)<\/p>\n<p>Germany&#8217;s government also confirmed on Oct. 19 that its Ministry of Economics has signed a memorandum of cooperation with Nvidia to promote AI adoption in industry, though full implementation details have not yet been announced. Huang and German Chancellor Friedrich Merz held a conversation in Berlin that day focused on how to leverage Germany&#8217;s strengths in manufacturing and automobiles to accelerate enterprise AI adoption. On the investment front, Nvidia announced plans to invest in German startups Agile Robots, Avelios Medical, Mirelo and Reflex Aerospace, part of its commitment of nearly $10 billion (about NT$330 billion) to Europe&#8217;s technology ecosystem.<\/p>\n<p>Intel May Be the Underestimated Beneficiary<\/p>\n<p>There is another potential storyline at this year&#8217;s GTC that the market has discussed less: Intel may be the surprise beneficiary. According to Nvidia&#8217;s official specifications, the DGX B300, which uses Blackwell Ultra GPUs, and the DGX Rubin NVL8, which uses Rubin GPUs, are both equipped with Intel Xeon 6776P processors, with the DGX Rubin NVL8 configured with two Xeon CPUs. This means that even as Nvidia continues to expand its own CPU lineup, Intel remains the host processor supplier in some DGX systems.<\/p>\n<p>The use of Xeon in existing products is already public information; the key thing to watch in Berlin is whether this partnership extends to subsequent generations. Nvidia disclosed in March that the next-generation Feynman platform will include its in-house Rosa CPU, and the market speculates that some Feynman-architecture DGX systems may still use Xeon, though this has not been officially confirmed.<\/p>\n<p>Even more intriguing is wafer foundry and advanced packaging. Market rumors persist that Nvidia may manufacture I\/O dies for the Feynman generation at Intel&#8217;s 18A-P or 14A nodes; at the same time, given that Taiwan Semiconductor Manufacturing&#8217;s (2330.TW) CoWoS advanced packaging capacity is becoming tight, whether Nvidia will add Intel&#8217;s EMIB-T or Foveros solutions to the Feynman packaging supply chain is also drawing attention. Any positive confirmation of either item could trigger a repricing of market expectations for Intel&#8217;s revenue. However, the related partnerships remain market rumors for now, with process selection, order scale and production timelines all unclear and awaiting formal announcements from both sides.<\/p>\n<p>What the Market Wants Is Orders and the Cost Curve<\/p>\n<p>Taken together, the core value of this year&#8217;s GTC to the market lies in whether it can answer three of investors&#8217; most pressing questions: whether enterprise AI, autonomous AI and AI factories can create additional infrastructure demand; whether AI Agents can deliver measurable commercial value rather than remaining at the pilot stage; and whether robotics and industrial AI can open new growth paths beyond data centers.<\/p>\n<p>GTC has always been an important forum for technology and industry development, but every advance showcased there does not necessarily translate immediately into revenue or a stock-price catalyst. For the current market, the most consequential content of Huang&#8217;s keynote is probably not the performance curve of the next-generation chips, but whether the AI industry, as it continues to expand, can present a clearer path to commercialization and evidence of investment returns.<\/p>\n","protected":false},"excerpt":{"rendered":"Nvidia&#8217;s Berlin GTC conference runs Oct. 20-22, with CEO Jensen Huang scheduled to deliver the keynote at 11&hellip;\n","protected":false},"author":2,"featured_media":189196,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","_share_on_mastodon":"0"},"categories":[99],"tags":[29634,112,85363,85364,85362,85361,190,79059,43467,18358,15446,63653,84985],"class_list":["post-189195","post","type-post","status-publish","format-standard","has-post-thumbnail","category-berlin","tag-ai-agent","tag-berlin","tag-blackwell","tag-crowdstrike","tag-dgx-rubin-nvl8","tag-feynman","tag-germany","tag-intel","tag-jensen-huang","tag-mercedes-benz","tag-nvidia","tag-taiwan-semiconductor-manufacturing","tag-vera-rubin"],"share_on_mastodon":{"url":"","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/posts\/189195","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/comments?post=189195"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/posts\/189195\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/media\/189196"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/media?parent=189195"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/categories?post=189195"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/dk\/wp-json\/wp\/v2\/tags?post=189195"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}