NVIDIA Launches DSX Platform, Expanding into AI Factory Infrastructure

marsbitPubblicato 2026-06-01Pubblicato ultima volta 2026-06-01

Introduzione

NVIDIA has unveiled the DSX platform at its GTC Taipei event, marking a strategic expansion from GPU sales into comprehensive AI factory infrastructure solutions. The platform addresses challenges like power supply, cooling, and resource orchestration as AI models scale, shifting the industry focus from single-chip performance to overall infrastructure efficiency. DSX integrates NVIDIA's chips, systems, software, and partner technologies to cover the entire AI factory lifecycle—from design and simulation to deployment and operations. It aims to accelerate deployment, improve reliability and operational efficiency, and reduce the cost per generated token in AI inference. The software suite includes DSX MaxLPS, which uses 45°C liquid cooling and rack-level optimization to allow up to 40% more GPUs per megawatt, and DSX OS, an open-source platform for AI factory operations. The platform also encompasses reference designs, digital twin simulation (DSX Sim), dynamic workload adjustment based on grid conditions (DSX Flex), and data exchange between systems. Early adopters include cloud providers like CoreWeave and Lambda. Major hardware partners, including Dell, HPE, Lenovo, and Supermicro, are developing DSX-ready systems. Pilot projects for DSX Flex are underway with energy providers. Strategically, DSX represents NVIDIA's ongoing transition from an AI chip supplier to a full-stack AI infrastructure platform provider, aiming to set industry standards and solidify its market l...

Summary: Jinshi Data

At the NVIDIA GTC Taipei conference held in Taipei, Taiwan, NVIDIA (NVDA.O) unveiled the NVIDIA DSX platform, further extending its business reach into the field of AI factory infrastructure.

Unlike its past focus on GPU sales, DSX aims to provide enterprises with a complete AI factory solution encompassing design, simulation, deployment, and operational management.

As AI models continue to scale, the challenges faced by data centers are no longer just about chip performance, but also involve power supply, cooling capacity, resource scheduling, and overall operational efficiency. NVIDIA believes that future competition in the AI industry will increasingly shift key metrics from single-chip performance to overall infrastructure efficiency—that is, how to produce more computing power and intelligent services under limited power, space, and resource constraints.

To this end, the DSX platform integrates NVIDIA's chips, systems, software, reference architectures, and partner technologies, covering the entire lifecycle of AI factory construction and operation. By unifying technology stacks across computing, software, and facilities, the platform helps customers improve deployment speed, reliability, and operational efficiency while reducing the cost of generating tokens during AI inference.

Jensen Huang stated:

"We're not just delivering chips—we're providing every infrastructure builder with a complete methodology for building AI factories. With the DSX platform, you can simulate the entire factory without spending a dime, validate performance before installing the first rack, and operate with the reliability required for production-grade AI."

The software suite announced this time primarily includes DSX MaxLPS and DSX OS.

Among them, DSX MaxLPS leverages 45-degree liquid cooling and rack-level power optimization technologies to increase token output per megawatt of power. NVIDIA stated that this technology allows for the deployment of up to 40% more GPUs with minimal impact on performance, thereby further reducing computing costs within a fixed power budget.

DSX OS is an open-source software platform for AI factory operations, supporting functions such as lifecycle management, intelligent scheduling, health automation, multi-tenant operations, and platform services. NVIDIA will also open-source modular software libraries, APIs, reference designs, and accelerated computing platforms to build a unified software architecture.

In addition to the core software, DSX integrates several existing capabilities. DSX Reference Design provides reference architectures covering computing, networking, storage, power, and cooling systems. DSX Sim supports digital twin simulation and optimization from planning to operation. DSX Flex dynamically adjusts workloads based on grid load and electricity price fluctuations. DSX Exchange facilitates data collaboration between computing, networking, energy, and cooling systems.

Regarding commercial deployment, cloud service providers such as CoreWeave, Crusoe, IREN, and Lambda have already deployed core DSX components to improve GPU utilization and shorten the time to market for AI cloud services.

The hardware ecosystem is also expanding simultaneously. Manufacturers including Dell Technologies (DELL.N), Hewlett Packard Enterprise (HPE.N), Lenovo Group (0992.HK), Supermicro (SMCI.O), ASUS, Foxconn, GIGABYTE, Pegatron, and QCT are developing NVIDIA DSX Ready systems to help customers build full-stack AI factories.

Meanwhile, DSX Flex has initiated commercial pilot projects with Emerald AI and Silicon Valley Power to validate the capability of AI factories to dynamically adjust power consumption based on grid demand.

From a strategic perspective, DSX marks NVIDIA's continued transition from an AI chip supplier to an AI infrastructure platform provider. By incorporating chips, software, data center architecture, operational management, and energy scheduling into a unified system, NVIDIA aims to establish industry standards covering the entire lifecycle of AI factories and further solidify its leading position in the global AI infrastructure market.

Domande pertinenti

QWhat is NVIDIA's DSX platform, and how does it extend the company's business strategy?

ANVIDIA's DSX platform is a comprehensive solution designed for building and operating AI factories. It extends NVIDIA's business strategy by moving beyond just selling GPUs to providing a full-stack infrastructure platform. The platform integrates NVIDIA's chips, systems, software, reference architectures, and partner technologies, covering the entire lifecycle of an AI factory from design and simulation to deployment and operations management.

QAccording to NVIDIA, what is becoming a key competitive metric in the AI industry, and how does DSX address this?

ANVIDIA believes the key competitive metric in the AI industry is shifting from single-chip performance to overall infrastructure efficiency. This focuses on how to produce more computing power and intelligent services under constraints of limited power, space, and resources. The DSX platform addresses this by providing integrated technology stacks to improve deployment speed, reliability, operational efficiency, and reduce the cost per token generated during AI inference.

QWhat are the two main software components announced as part of the DSX platform, and what are their primary functions?

AThe two main software components are DSX MaxLPS and DSX OS. DSX MaxLPS utilizes 45-degree liquid cooling and rack-level power optimization to increase token output per megawatt of power, allowing for up to 40% more GPU deployment with minimal performance impact. DSX OS is an open-source software platform for AI factory operations, supporting lifecycle management, intelligent scheduling, health automation, multi-tenant operations, and platform services.

QWhat is the DSX Flex component, and what is its current commercial status?

ADSX Flex is a component of the DSX platform designed to dynamically adjust AI factory workloads based on grid load and electricity price fluctuations. It is currently involved in commercial pilot projects with Emerald AI and Silicon Valley Power to validate the capability of AI factories to dynamically regulate power consumption according to grid demands.

QHow does the DSX platform represent a strategic shift for NVIDIA, and what is its goal?

AThe DSX platform represents NVIDIA's strategic shift from an AI chip supplier to an AI infrastructure platform provider. By unifying chips, software, data center architecture, operations management, and energy scheduling into a single system, NVIDIA aims to establish industry standards covering the entire AI factory lifecycle and further solidify its leading position in the global AI infrastructure market.

Letture associate

Trend in US Stocks: Jensen Huang's One Sentence Triggers $47 Billion Surge; Google Raises Funds for First Time in 20 Years

U.S. markets reached record highs on June 2nd, but the real story was the intensifying AI arms race, now pivoting from chip supremacy to a scramble for capital to fund compute infrastructure. The day highlighted two stark realities: Nvidia CEO Jensen Huang's endorsement of Marvell Technology as the "next trillion-dollar company" at Computex fueled a historic 32.5% surge, adding $47 billion to its value. Conversely, Alphabet announced its first equity raise in two decades—an $80 billion plan—signaling that even its massive cash flow can't keep pace with soaring AI capital expenditures, forecast to exceed $180 billion in 2026. While the S&P 500 closed above 7,600 for the first time, led by tech and semiconductor stocks (SOXX +5.79%), sector performance was mixed. Alphabet's 4% drop dragged down communications services, illustrating market anxiety over the unsustainable cost of the AI buildout. Hewlett Packard Enterprise soared 25% on stellar earnings, proving AI's benefits extend beyond chip designers to infrastructure providers. Beneath the index highs, concerns linger over extreme concentration in a few AI stocks and geopolitical tensions. The focus now shifts to upcoming economic data, particularly Friday's nonfarm payrolls, which could challenge the market's current "ignore rates, chase AI" mentality.

marsbit3 min fa

Trend in US Stocks: Jensen Huang's One Sentence Triggers $47 Billion Surge; Google Raises Funds for First Time in 20 Years

marsbit3 min fa

Variant: Bitcoin, Ethereum, and ZCash Highly Likely to Become Major Stores of Value

At Variant, our investment philosophy centers on enabling individuals to own their money, identity, and data. We believe all tokens fall into two categories: stores of value (SOV) or equity-like instruments. The SOV framework is particularly useful for evaluating Layer 1 (L1) blockchains, which are major monetary coordination protocols. A good store of value should possess: 1) Technical durability, 2) Scarcity, 3) Censorship resistance, 4) Economic productivity, 5) Strong memetics, and 6) Liquidity. The total addressable market for SOV assets is massive, with gold alone valued at $31 trillion. Three L1 assets stand out as prime candidates for becoming major stores of value: Bitcoin (BTC), Ethereum (ETH), and ZCash (ZEC). Bitcoin excels in memetic dominance as "digital gold." Ethereum demonstrates superior technical durability and adaptability through its upgradeable roadmap. ZCash offers exceptional censorship resistance and privacy via its shielded transactions. While digital SOVs outperform traditional ones like gold on many metrics, they still represent a small fraction of the overall SOV market, presenting a significant growth opportunity.

marsbit46 min fa

Variant: Bitcoin, Ethereum, and ZCash Highly Likely to Become Major Stores of Value

marsbit46 min fa

Can DeepSeek Save China One Trillion Dollars?

"DeepSeek and the $1 Trillion Infrastructure Question" The article examines whether DeepSeek's AI optimization breakthroughs could potentially save China $1 trillion in future AI infrastructure costs. The analysis begins with Nvidia's upcoming Vera Rubin AI platform, costing ~$7.8 million, where memory (HBM4/LPDDR5X) constitutes $2 million—a 435% cost increase in one year, highlighting how AI hardware spending is shifting toward expensive memory components. DeepSeek's approach works in the opposite direction. Through three key technical innovations showcased in DeepSeek V4, the company dramatically improves hardware efficiency: 1. **Memory Compression (MLA)**: Re-engineers the attention mechanism to compress long-context memory (KV Cache) by over 90%, drastically reducing expensive HBM usage. 2. **Selective Activation (MoE)**: Employs Mixture-of-Experts architecture where only a small fraction of parameters (e.g., 49B out of 1.6T in V4-Pro) are activated per token, allowing most parameters to reside in cheaper memory/SSD. 3. **Computation Caching**: Reuses previously computed results via cache hits, replacing expensive GPU computations with cheap memory reads. Combined, these optimizations allow the same hardware to produce approximately 4x more tokens, effectively reducing required hardware investment by 75%. DeepSeek's pricing reflects this: a 10-billion token workload costs ~$522 monthly versus ~$9,000-$10,000 for competitors. The $1 trillion savings projection stems from McKinsey's estimate that global AI infrastructure will require ~$5.2 trillion investment by 2030. As China's daily token consumption grows toward quadrillions, even marginal efficiency gains scale massively. With a conservative 4x throughput improvement, China could avoid building tens of thousands of AI data centers equivalent to ~7 trillion RMB ($1 trillion) in saved investment. Critically, this strategy shifts dependency from scarce, expensive GPU/HBM—where China lags—toward more accessible storage, caching, and systems engineering where domestic suppliers like CXMT are gaining strength. Rather than "replacing Nvidia," DeepSeek rebalances AI's value chain away from monolithic hardware dependency. Ultimately, DeepSeek's technical breakthroughs could lower the barrier to AI adoption across Chinese industries by making advanced capabilities affordable at scale—transforming who can access next-generation AI.

marsbit46 min fa

Can DeepSeek Save China One Trillion Dollars?

marsbit46 min fa

Overturning the Mainstream Approach to Hallucinations: Metacognition is the New Solution for Large Models to Break the Hallucination Barrier

This paper, "Hallucinations Undermine Trust; Metacognition is a Way Forward," proposes a paradigm shift in combating AI hallucination. It argues that the current mainstream approaches—striving for omniscience by scaling data/models or having AI abstain from uncertain answers—are fundamentally flawed. The former has inevitable knowledge gaps, while the latter imposes a crippling "utility tax," requiring the rejection of many correct answers to achieve high accuracy, due to models' poor "discrimination" (the ability to distinguish correct from incorrect answers internally). The core contribution is redefining hallucination not as "being wrong," but as "expressing false information with unwarranted certainty." The proposed solution is **Faithful Uncertainty** or **Metacognition**: enabling AI to accurately perceive its internal uncertainty and honestly express it in its language (e.g., using hedging phrases when unsure). This creates a more reliable assistant that provides useful information while signaling its confidence, minimizing harm from errors. The paper emphasizes that metacognition is critical for the era of AI Agents. Without it, Agents cannot intelligently decide when to use tools like search engines, leading to inefficiency and misuse. Key implementation challenges are highlighted: the "bootstrapping paradox" of training with static uncertainty data, the "alignment distortion signal" where human preference training suppresses internal uncertainty cues, and the difficulty of causally evaluating true metacognition vs. its superficial imitation. The paper concludes that the goal should not be an infallible AI, but one that is honest about the limits of its knowledge, thereby building user trust through transparent communication of its certainty.

marsbit51 min fa

Overturning the Mainstream Approach to Hallucinations: Metacognition is the New Solution for Large Models to Break the Hallucination Barrier

marsbit51 min fa

Hedge by Buying Gold and Oil, Chase Soaring Returns with AI. ‘Dated’ Bitcoin Enters a Bear Market

Bitcoin has recently declined, hitting a two-month low near $66,123, while Ethereum fell to a three-month low around $1,837. Analysts suggest the drop is not merely due to factors like ETF outflows or MicroStrategy's selling but reflects a deeper issue: Bitcoin is losing a broader asset competition. In a near-zero interest rate environment, Bitcoin previously thrived as an outlet for investor dissatisfaction with inflation and limited options. However, the market landscape has shifted. Bitcoin now occupies an "awkward middle ground," facing competition on three fronts. For inflation hedging, investors prefer gold, energy stocks, and commodity producers—assets with tangible backing and clearer pricing power. For growth exposure, AI-related companies with actual revenues and profits are more attractive. Even within crypto, investors can choose stablecoins, exchanges, or infrastructure firms tied directly to adoption, offering clearer business models and leverage. Thus, Bitcoin is no longer the top choice for hedging, growth, or crypto exposure. This shift is evident in market reactions: despite recent warnings about persistent inflation from a Fed official, Bitcoin did not rally as it might have in the past. Instead, capital flowed to assets with direct commodity or energy exposure. The recent ETF outflows and MicroStrategy sales are symptoms, not causes, of this new reality. Investors are becoming more selective, demanding clearer value propositions beyond mere scarcity. The emerging bear case for Bitcoin is not about it being a bubble or failed technology, but that scarcity alone is no longer sufficient.

华尔街日报54 min fa

Hedge by Buying Gold and Oil, Chase Soaring Returns with AI. ‘Dated’ Bitcoin Enters a Bear Market

华尔街日报54 min fa

Trading

Spot
Futures
活动图片