Spotlight on NVIDIA's GTC Berlin! What are the key takeaways from Jensen Huang's speech?

牛牛課堂

joined discussion · ·

Jensen Huang makes a major appearance on Wednesday! Five key highlights from NVIDIA's GTC revealed: Can the AI rally reverse its downward trend?

Last week, the AI industry faced a double market test: on one hand, AI cloud computing companies were hit by credit risk shocks, raising growing market concerns about the financing costs and debt sustainability of compute infrastructure; on the other hand, the IPO valuations of leading large AI model companies sparked controversy, reigniting market focus on whether AI commercialization can support such high capital expenditures.

Under this dual pressure, the investment logic for AI is undergoing a rare market stress test.The UBS Group trading desk believes that the core shift in the US stock market last week was 'deleveraging, not de-risking': AI and momentum sectors faced targeted selling, with the memory sector dropping 16% in a single week. Capital rotated into defensive and value sectors, but overall risk exposure was not systematically abandoned.

Against this backdrop, $NVIDIA (NVDA.US)$ The upcoming GTC conference may provide important clues for observing the development direction of the AI industry in the next phase.

NVIDIA GTC Berlin 2026 will be held in Berlin, Germany, from October 20 to 22 local time. NVIDIA CEO Jensen Huang willdeliver his keynote speech from 17:00 to 19:00 Beijing Time on October 21(11:00-13:00 local time on October 21), sharing his latest outlook on AI technology and industrial development.

Jensen Huang's Keynote Address at the NVIDIA GTC Conference

Replay

Rather than focusing solely on whether new products will be launched, the market should pay closer attention to:As AI investment enters the stage of return validation, how NVIDIA will further expand demand for AI infrastructure, drive the implementation of AI Agents, and extend AI capabilities into the industrial and physical worlds.

1. AI Investment Enters Validation Phase: Why Is GTC Drawing So Much Attention?

In recent years, the investment thesis in the AI industry has been relatively clear: continuous improvements in large model capabilities have driven growth in computing power demand, thereby boosting capital expenditure in sectors such as GPUs, servers, network equipment, and data centers.

However, as investments in AI infrastructure continue to expand, market focus is shifting.

First,Rising demand for computing power does not guarantee that all related investments will yield ideal returns. Data center construction involves chip procurement, power supply, network equipment, and financing costs. Only when companies can sustain customer demand, improve computing utilization rates, and achieve commercialization will these massive investments have the potential to generate sustainable revenue.

Secondly, AI applications are moving from model training toward inference and practical deployment. Companies are no longer focused solely on model capability; they are also paying attention to inference costs, operational efficiency, data security, and whether AI can truly integrate into business processes.

This means that growth opportunities in the AI industry may no longer be concentrated solely on GPUs, but are gradually extending to areas such as network interconnects, data center infrastructure, inference software, enterprise-grade AI platforms, and end-user applications.

Therefore, a key focus at this year's GTC is how NVIDIA further refines its AI ecosystem spanning chips, systems, software, and applications, and whether these technological advancements can create new demand.

2. Five Major Themes Emerge: Where Is the Next Stop for the AI Industry?

Based on the agenda announced for GTC Berlin, the conference will not only focus on computing infrastructure but also cover directions such as autonomous enterprise AI, AI-native enterprises, security and assurance, and physical AI.

Symbols

1. Sovereign AI vs. Enterprise AI: Can demand for AI infrastructure expand further?

In the session on Sovereign AI and Enterprise AI, the discussion focused on how countries and corporations leverage their own data, industry knowledge, and infrastructure to build and deploy proprietary AI capabilities. Guest speakers included Dassault Systèmes CEO Pascal Daloz, $DEUTSCHE TELEKOM (DTEGY.US)$ CEO Tim Höttges, and Mistral AI CEO Arthur Mensch.

What makes this trend noteworthy is that AI demand is no longer driven solely by a few large model companies. Governments, telecom operators, industrial enterprises, and other major institutions may also build dedicated AI systems due to data sovereignty, compliance requirements, and operational efficiency needs.

For the supply chain, if enterprise-level AI deployment continues to expand, sectors such as computing power infrastructure, private deployment solutions, enterprise-grade software, and data security are poised to see new demand.

2. AI-Native Companies: Can AI transition from technological breakthroughs to commercial growth?

The session on AI-native companies will explore how European entrepreneurs and investors are leveraging AI technology to build the next generation of businesses. Featured guests include Corti.AI co-founder Andreas Cleve, General Catalyst’s Jeannette zu Fürstenberg, Black Forest Labs co-founder Robin Rombach, and Ineffable Intelligence founder David Silver.

For the market, the significance of this session lies in observing the commercialization progress of AI applications.If AI does more than just improve the efficiency of existing software and instead catalyzes new products, services, and business models, the growth drivers of the AI industry could extend beyond infrastructure investment to include application and service revenues.

However, what truly needs validation is customers' willingness to pay, revenue growth, and profitability, rather than mere user numbers or technological hype.

3. AI Security and Enterprise Deployment: What’s Missing for Agent Adoption?

As AI agents gradually gain the ability to invoke tools, access data, and execute tasks, security issues have become increasingly critical. This session features: $CrowdStrike (CRWD.US)$ CEO George Kurtz, Hugging Face co-founder Thomas Wolf, and Red Hat’s Chris Wright.

As AI shifts from answering questions to executing tasks, enterprises must ensure that AI accesses data and performs operations within authorized scopes, supported by robust security mechanisms.

Therefore, the widespread adoption of AI agents involves not only model capabilities but also identity authentication, permission management, cybersecurity, and enterprise software architecture.

For investors, this means AI security and enterprise-grade software could be promising areas to monitor, though specific commercial value will depend on actual deployment scale and customer demand.

4. AI Factories: Competition in Computing Infrastructure Will Shift Toward System Efficiency

AI factories are another key topic at this GTC, with discussions focusing on how infrastructure development can boost productivity and drive economic growth. Featured speakers include Frédéric Godemel from Schneider Electric, $Brookfield (BN.US)$ Siko Rashid from [Company Name], $NEBIUS (NBIS.US)$ and Arkady Volozh, founder and CEO of [Company Name].

From a supply chain perspective, AI factories are not just about procuring GPUs; they encompass a complete system including computing, networking, storage, power supply, cooling, and software orchestration.

As the scale of AI clusters continues to expand, the performance of a single chip is no longer sufficient to determine overall efficiency. Factors such as improving cluster utilization, reducing computational cost per unit, and shortening data center deployment cycles will all impact the return on investment in computing power.

This also means that, beyond focusing on GPU demand, the market needs to pay attention to developments in high-speed interconnects, networking equipment, power supply, and data center infrastructure.

5. Physical AI and Industrial AI: AI Moving Beyond Data Centers

Physical AI extends AI capabilities into real-world scenarios such as robotics, automobiles, and industrial equipment. Speakers in this session include Zhaopeng Chen from Agile Robots, Arthur Kaindl from Siemens Healthineers, $MERCEDES-BENZ GROUP AG UNSP ADR EACH REP 0.25 ORD SHS (MBGYY.US)$ CEO Ola Källenius, and Artem Sokolov from Humanoid.

Unlike generative AI, which primarily processes text, images, and data, Physical AI requires an understanding of the physical environment and completes tasks in the real world through perception, reasoning, simulation, and control.

This direction is expected to extend AI demand into robotics, industrial automation, intelligent vehicles, and related computing platforms.

However, from an investment perspective, it is still necessary to distinguish between technical demonstrations and scalable commercial deployment. The actual operational capabilities, cost, reliability, and customer adoption rate of robots are key to determining whether the industry can achieve sustainable growth.

III. More Important Than New Products: The Continued Validation of AI Return on Investment

Based on the agenda of this conference, the market should focus on three key questions:

First, is the demand for computing power still expanding? Beyond monitoring NVIDIA's chip and system strategies, it is crucial to observe whether enterprise AI, sovereign AI, and AI factories are driving new infrastructure demands.

Second, can AI Agents create tangible commercial value? As AI gradually evolves from an auxiliary tool to autonomous task execution, areas such as inference computing, security software, and enterprise deployment may see more opportunities. However, the relationship between costs and revenue still needs validation.

Third, can Physical AI unlock new growth potential? Scenarios such as robotics, industrial automation, and intelligent vehicles could broaden the application boundaries of AI, but the pace of commercialization and profitability models remain to be seen.

It is important to note that GTC serves as a vital window for technology and industry exchange; it does not imply that every technological advancement showcased will immediately translate into corporate revenue or act as a catalyst for stock prices.

As the investment logic for AI undergoes re-evaluation, the long-term trajectory of the industry will ultimately depend on whether computing demand remains sustainable, whether capital expenditures yield returns, and whether AI applications can establish viable business models.

For investors, the most noteworthy aspect of this GTC is not just what new technologies NVIDIA launches, but whether the AI industry can provide clearer answers regarding commercialization and return on investment amidst its continued expansion.

Celebrating Futu's 14th Anniversary! Thank you to our fellow investors for your continued support! From October 12 to November 21, Futu's 3rd Global Paper Trading Competition brings together contestants from 8 major regions!

Join the paper trading competition and prediction games now for a chance to shareUp to $200,000 in cash rewards！

The top 3 global finalists will even have the chance to go international and appear on the Nasdaq billboard in Times Square, New York! Earn competition points and win rich rewards—sign up now!

📢Event Period: October 12, 2026, 10:00 – November 21, 2026, 09:00

*This event is open only to Futu Hong Kong users and existing investors.

304K Views