Urban populations are expected to double by 2050, which means around 2.5 billion people could be added to urban areas by the middle of the century, driving the need for more sustainable urban planning and public services. Cities across the globe are turning to digital twins and AI agents for urban planning scenario analysis and data-driven operational decisions. Building a digital twin of a city and testing smart city AI agents within it, however, is a complex and resource-intensive endeavor, fraught with technical and operational challenges. To address those challenges, NVIDIA today announced the NVIDIA Omniverse Blueprint for smart city AI, a reference framework that combines the NVIDIA Omniverse, Cosmos, NeMo and Metropolis platforms to bring the benefits of physical AI to entire cities and their critical infrastructure. Using the blueprint, developers can build simulation-ready, or SimReady, photorealistic digital twins of cities to build and test AI agents that can help monitor and optimize city operations. Leading companies including XXII, AVES Reality, Akila, Blyncsy, Bentley, Cesium, K2K, Linker Vision, Milestone Systems, Nebius, SNCF Gares&Connexions, Trimble and Younite AI are among the first to use the new blueprint. NVIDIA Omniverse Blueprint for Smart City AI  The NVIDIA Omniverse Blueprint for smart city AI provides the complete software stack needed to accelerate the development and testing of AI agents in physically accurate digital twins of cities. It includes: NVIDIA Omniverse to build physically accurate digital twins and run simulations at city scale. NVIDIA Cosmos to generate synthetic data at scale for post-training AI models. NVIDIA NeMo to curate high-quality data and use that data to train and fine-tune vision language models (VLMs) and large language models. NVIDIA Metropolis to build and deploy video analytics AI agents based on the NVIDIA AI Blueprint for video search and summarization (VSS), helping process vast amounts of video data and provide critical insights to optimize business processes. The blueprint workflow comprises three key steps. First, developers create a SimReady digital twin of locations and facilities using aerial, satellite or map data with Omniverse and Cosmos. Second, they can train and fine-tune AI models, like computer vision models and VLMs, using NVIDIA TAO and NeMo Curator to improve accuracy for vision AI use cases​. Finally, real-time AI agents powered by these customized models are deployed to alert, summarize and query camera and sensor data using the Metropolis VSS blueprint. NVIDIA Partner Ecosystem Powers Smart Cities Worldwide The blueprint for smart city AI enables a large ecosystem of partners to use a single workflow to build and activate digital twins for smart city use cases, tapping into a combination of NVIDIA’s technologies and their own. SNCF Gares&Connexions, which operates a network of 3,000 train stations across France and Monaco, has deployed a digital twin and AI agents to enable real-time operational monitoring, emergency response simulations and infrastructure upgrade planning. This helps each station analyze operational data such as energy and water use, and enables predictive maintenance capabilities, automated reporting and GDPR-compliant video analytics for incident detection and crowd management. Powered by Omniverse, Metropolis and solutions from ecosystem partners Akila and XXII, SNCF Gares&Connexions’ physical AI deployment at the Monaco-Monte-Carlo and Marseille stations has helped SNCF Gares&Connexions achieve a 100% on-time preventive maintenance completion rate, a 50% reduction in downtime and issue response time, and a 20% reduction in energy consumption. https://blogs.nvidia.com/wp-content/uploads/2025/06/01-Monaco-Akila.mp4 The city of Palermo in Sicily is using AI agents and digital twins from its partner K2K to improve public health and safety by helping city operators process and analyze footage from over 1,000 public video streams at a rate of nearly 50 billion pixels per second. Tapped by Sicily, K2K’s AI agents — built with the NVIDIA AI Blueprint for VSS and cloud solutions from Nebius — can interpret and act on video data to provide real-time alerts on public events. To accurately predict and resolve traffic incidents, K2K is generating synthetic data with Cosmos world foundation models to simulate different driving conditions. Then, K2K uses the data to fine-tune the VLMs powering the AI agents with NeMo Curator. These simulations enable K2K’s AI agents to create over 100,000 predictions per second. https://blogs.nvidia.com/wp-content/uploads/2025/06/02-K2K-Polermo-1600x900-1.mp4 Milestone Systems — in collaboration with NVIDIA and European cities — has launched Project Hafnia, an initiative to build an anonymized, ethically sourced video data platform for cities to develop and train AI models and applications while maintaining regulatory compliance. Using a combination of Cosmos and NeMo Curator on NVIDIA DGX Cloud and Nebius’ sovereign European cloud infrastructure, Project Hafnia scales up and enables European-compliant training and fine-tuning of video-centric AI models, including VLMs, for a variety of smart city use cases. The project’s initial rollout, taking place in Genoa, Italy, features one of the world’s first VLM models for intelligent transportation systems. https://blogs.nvidia.com/wp-content/uploads/2025/06/03-Milestone.mp4 Linker Vision was among the first to partner with NVIDIA to deploy smart city digital twins and AI agents for Kaohsiung City, Taiwan — powered by Omniverse, Cosmos and Metropolis. Linker Vision worked with AVES Reality, a digital twin company, to bring aerial imagery of cities and infrastructure into 3D geometry and ultimately into SimReady Omniverse digital twins. Linker Vision’s AI-powered application then built, trained and tested visual AI agents in a digital twin before deployment in the physical city. Now, it’s scaling to analyze 50,000 video streams in real time with generative AI to understand and narrate complex urban events like floods and traffic accidents. Linker Vision delivers timely insights to a dozen city departments through a single integrated AI-powered platform, breaking silos and reducing incident response times by up to 80%. https://blogs.nvidia.com/wp-content/uploads/2025/06/02-Linker-Vision-1280x680-1.mp4 Bentley Systems is joining the effort to bring physical AI to cities with the NVIDIA blueprint. Cesium, the open 3D geospatial platform, provides the foundation for visualizing, analyzing and managing infrastructure projects and ports digital twins to Omniverse. The company’s AI platform Blyncsy uses synthetic data generation and Metropolis to analyze road conditions and improve maintenance. Trimble, a global technology company that enables essential industries including construction, geospatial and transportation, is exploring ways to integrate components of the Omniverse blueprint into its reality capture workflows and Trimble Connect digital twin platform for surveying and mapping applications for smart cities. Younite AI, a developer of AI and 3D digital twin solutions, is adopting the blueprint to accelerate its development pipeline, enabling the company to quickly move from operational digital twins to large-scale urban simulations, improve synthetic data generation, integrate real-time IoT sensor data and deploy AI agents. Learn more about the NVIDIA Omniverse Blueprint for smart city AI by attending this GTC Paris session or watching the on-demand video after the event. Sign up to be notified when the blueprint is available. Watch the NVIDIA GTC Paris keynote from NVIDIA founder and CEO Jensen Huang at VivaTech, and explore GTC Paris sessions.

As a global labor shortage leaves 50 million positions unfilled across industries like manufacturing and logistics, Hexagon — a global leader in measurement technologies — is developing humanoid robots that can lend a helping hand. Industrial sectors depend on skilled workers to perform a variety of error-prone tasks, including operating high-precision scanners for reality capture — the process of capturing digital data to replicate the real world in simulation. At the Hexagon LIVE Global conference, Hexagon’s robotics division today unveiled AEON — a new humanoid robot built in collaboration with NVIDIA that’s engineered to perform a wide range of industrial applications, from manipulation and asset inspection to reality capture and operator support. Hexagon plans to deploy AEON across automotive, transportation, aerospace, manufacturing, warehousing and logistics. Future use cases for AEON include: Reality capture, which involves automatic planning and then scanning of assets, industrial spaces and environments to generate 3D models. The captured data is then used for advanced visualization and collaboration in the Hexagon Digital Reality (HxDR) platform powering Hexagon Reality Cloud Studio (RCS). Manipulation tasks, such as sorting and moving parts in various industrial and manufacturing settings. Part inspection, which includes checking parts for defects or ensuring adherence to specifications. Industrial operations, including highly dexterous technical tasks like machinery operations, teleoperation and scanning parts using high-end scanners. “The age of general-purpose robotics has arrived, due to technological advances in simulation and physical AI,” said Deepu Talla, vice president of robotics and edge AI at NVIDIA. “Hexagon’s new AEON humanoid embodies the integration of NVIDIA’s three-computer robotics platform and is making a significant leap forward in addressing industry-critical challenges.” Using NVIDIA’s Three Computers to Develop AEON  To build AEON, Hexagon used NVIDIA’s three computers for developing and deploying physical AI systems. They include AI supercomputers to train and fine-tune powerful foundation models; the NVIDIA Omniverse platform, running on NVIDIA OVX servers, for testing and optimizing these models in simulation environments using real and physically based synthetic data; and NVIDIA IGX Thor robotic computers to run the models. Hexagon is exploring using NVIDIA accelerated computing to post-train the NVIDIA Isaac GR00T N1.5 open foundation model to improve robot reasoning and policies, and tapping Isaac GR00T-Mimic to generate vast amounts of synthetic motion data from a few human demonstrations. AEON learns many of its skills through simulations powered by the NVIDIA Isaac platform. Hexagon uses NVIDIA Isaac Sim, a reference robotic simulation application built on Omniverse, to simulate complex robot actions like navigation, locomotion and manipulation. These skills are then refined using reinforcement learning in NVIDIA Isaac Lab, an open-source framework for robot learning. https://blogs.nvidia.com/wp-content/uploads/2025/06/Copy-of-robotics-hxgn-live-blog-1920x1080-1.mp4 This simulation-first approach enabled Hexagon to fast-track its robotic development, allowing AEON to master core locomotion skills in just 2-3 weeks — rather than 5-6 months — before real-world deployment. In addition, AEON taps into NVIDIA Jetson Orin onboard computers to autonomously move, navigate and perform its tasks in real time, enhancing its speed and accuracy while operating in complex and dynamic environments. Hexagon is also planning to upgrade AEON with NVIDIA IGX Thor to enable functional safety for collaborative operation. “Our goal with AEON was to design an intelligent, autonomous humanoid that addresses the real-world challenges industrial leaders have shared with us over the past months,” said Arnaud Robert, president of Hexagon’s robotics division. “By leveraging NVIDIA’s full-stack robotics and simulation platforms, we were able to deliver a best-in-class humanoid that combines advanced mechatronics, multimodal sensor fusion and real-time AI.” Data Comes to Life Through Reality Capture and Omniverse Integration  AEON will be piloted in factories and warehouses to scan everything from small precision parts and automotive components to large assembly lines and storage areas. Captured data comes to life in RCS, a platform that allows users to collaborate, visualize and share reality-capture data by tapping into HxDR and NVIDIA Omniverse running in the cloud. This removes the constraint of local infrastructure. “Digital twins offer clear advantages, but adoption has been challenging in several industries,” said Lucas Heinzle, vice president of research and development at Hexagon’s robotics division. “AEON’s sophisticated sensor suite enables the integration of reality data capture with NVIDIA Omniverse, streamlining workflows for our customers and moving us closer to making digital twins a mainstream tool for collaboration and innovation.” AEON’s Next Steps By adopting the OpenUSD framework and developing on Omniverse, Hexagon can generate high-fidelity digital twins from scanned data — establishing a data flywheel to continuously train AEON. This latest work with Hexagon is helping shape the future of physical AI — delivering scalable, efficient solutions to address the challenges faced by industries that depend on capturing real-world data. Watch the Hexagon LIVE keynote, explore presentations and read more about AEON. All imagery courtesy of Hexagon.

From the heart of Germany’s automotive sector to manufacturing hubs across France and Italy, Europe is embracing industrial AI and advanced AI-powered robotics to address labor shortages, boost productivity and fuel sustainable economic growth. Robotics companies are developing humanoid robots and collaborative systems that integrate AI into real-world manufacturing applications. Supported by a $200 billion investment initiative and coordinated efforts from the European Commission, Europe is positioning itself at the forefront of the next wave of industrial automation, powered by AI. This momentum is on full display at Automatica — Europe’s premier conference on advancements in robotics, machine vision and intelligent manufacturing — taking place this week in Munich, Germany. NVIDIA and its ecosystem of partners and customers are showcasing next-generation robots, automation and AI technologies designed to accelerate the continent’s leadership in smart manufacturing and logistics. NVIDIA Technologies Boost Robotics Development  Central to advancing robotics development is Europe’s first industrial AI cloud, announced at NVIDIA GTC Paris at VivaTech earlier this month. The Germany-based AI factory, featuring 10,000 NVIDIA GPUs, provides European manufacturers with secure, sovereign and centralized AI infrastructure for industrial workloads. It will support applications ranging from design and engineering to factory digital twins and robotics. To help accelerate humanoid development, NVIDIA released NVIDIA Isaac GR00T N1.5 — an open foundation model for humanoid robot reasoning and skills. This update enhances the model’s adaptability and ability to follow instructions, significantly improving its performance in material handling and manufacturing tasks. To help post-train GR00T N1.5, NVIDIA has also released the Isaac GR00T-Dreams blueprint — a reference workflow for generating vast amounts of synthetic trajectory data from a small number of human demonstrations — enabling robots to generalize across behaviors and adapt to new environments with minimal human demonstration data. In addition, early developer previews of NVIDIA Isaac Sim 5.0 and Isaac Lab 2.2 — open-source robot simulation and learning frameworks optimized for NVIDIA RTX PRO 6000 workstations — are now available on GitHub. Image courtesy of Wandelbots. Robotics Leaders Tap NVIDIA Simulation Technology to Develop and Deploy Humanoids and More  Robotics developers and solutions providers across the globe are integrating NVIDIA’s three computers to train, simulate and deploy robots. NEURA Robotics, a German robotics company and pioneer for cognitive robots, unveiled the third generation of its humanoid, 4NE1, designed to assist humans in domestic and professional environments through advanced cognitive capabilities and humanlike interaction. 4NE1 is powered by GR00T N1 and was trained in Isaac Sim and Isaac Lab before real-world deployment. NEURA Robotics is also presenting Neuraverse, a digital twin and interconnected ecosystem for robot training, skills and applications, fully compatible with NVIDIA Omniverse technologies. Delta Electronics, a global leader in power management and smart green solutions, is debuting two next-generation collaborative robots: D-Bot Mar and D-Bot 2 in 1 — both trained using Omniverse and Isaac Sim technologies and libraries. These cobots are engineered to transform intralogistics and optimize production flows. Wandelbots, the creator of the Wandelbots NOVA software platform for industrial robotics, is partnering with SoftServe, a global IT consulting and digital services provider, to scale simulation-first automating using NVIDIA Isaac Sim, enabling virtual validation and real-world deployment with maximum impact. Cyngn, a pioneer in autonomous mobile robotics, is integrating its DriveMod technology into Isaac Sim to enable large-scale, high fidelity virtual testing of advanced autonomous operation. Purpose-built for industrial applications, DriveMod is already deployed on vehicles such as the Motrec MT-160 Tugger and BYD Forklift, delivering sophisticated automation to material handling operations. Doosan Robotics, a company specializing in AI robotic solutions, will showcase its “sim to real” solution, using NVIDIA Isaac Sim and cuRobo. Doosan will be showcasing how to seamlessly transfer tasks from simulation to real robots across a wide range of applications — from manufacturing to service industries. Franka Robotics has integrated Isaac GR00T N1.5 into a dual-arm Franka Research 3 (FR3) robot for robotic control. The integration of GR00T N1.5 allows the system to interpret visual input, understand task context and autonomously perform complex manipulation — without the need for task-specific programming or hardcoded logic. Image courtesy of Franka Robotics. Hexagon, the global leader in measurement technologies, launched its new humanoid, dubbed AEON. With its unique locomotion system and multimodal sensor fusion, and powered by NVIDIA’s three-computer solution, AEON is engineered to perform a wide range of industrial applications, from manipulation and asset inspection to reality capture and operator support. Intrinsic, a software and AI robotics company, is integrating Intrinsic Flowstate with  Omniverse and OpenUSD for advanced visualization and digital twins that can be used in many industrial use cases. The company is also using NVIDIA foundation models to enhance robot capabilities like grasp planning through AI and simulation technologies. SCHUNK, a global leader in gripping systems and automation technology, is showcasing its innovative grasping kit powered by the NVIDIA Jetson AGX Orin module. The kit intelligently detects objects and calculates optimal grasping points. Schunk is also demonstrating seamless simulation-to-reality transfer using IGS Virtuous software — built on Omniverse technologies — to control a real robot through simulation in a pick-and-place scenario. Universal Robots is showcasing UR15, its fastest cobot yet. Powered by the UR AI Accelerator — developed with NVIDIA and running on Jetson AGX Orin using CUDA-accelerated Isaac libraries — UR15 helps set a new standard for industrial automation. Vention, a full-stack software and hardware automation company, launched its Machine Motion AI, built on CUDA-accelerated Isaac libraries and powered by Jetson. Vention is also expanding its lineup of robotic offerings by adding the FR3 robot from Franka Robotics to its ecosystem, enhancing its solutions for academic and research applications. Image courtesy of Vention. Learn more about the latest robotics advancements by joining NVIDIA at Automatica, running through Friday, June 27. 

Editor’s note: This blog is a part of Into the Omniverse, a series focused on how developers, 3D practitioners and enterprises can transform their workflows using the latest advances in OpenUSD and NVIDIA Omniverse. Simulated driving environments enable engineers to safely and efficiently train, test and validate autonomous vehicles (AVs) across countless real-world and edge-case scenarios without the risks and costs of physical testing. These simulated environments can be created through neural reconstruction of real-world data from AV fleets or generated with world foundation models (WFMs) — neural networks that understand physics and real-world properties. WFMs can be used to generate synthetic datasets for enhanced AV simulation. To help physical AI developers build such simulated environments, NVIDIA unveiled major advances in WFMs at the GTC Paris and CVPR conferences earlier this month. These new capabilities enhance NVIDIA Cosmos — a platform of generative WFMs, advanced tokenizers, guardrails and accelerated data processing tools. Key innovations like Cosmos Predict-2, the Cosmos Transfer-1 NVIDIA preview NIM microservice and Cosmos Reason are improving how AV developers generate synthetic data, build realistic simulated environments and validate safety systems at unprecedented scale. Universal Scene Description (OpenUSD), a unified data framework and standard for physical AI applications, enables seamless integration and interoperability of simulation assets across the development pipeline. OpenUSD standardization plays a critical role in ensuring 3D pipelines are built to scale. NVIDIA Omniverse, a platform of application programming interfaces, software development kits and services for building OpenUSD-based physical AI applications, enables simulations from WFMs and neural reconstruction at world scale. Leading AV organizations — including Foretellix, Mcity, Oxa, Parallel Domain, Plus AI and Uber — are among the first to adopt Cosmos models. Foundations for Scalable, Realistic Simulation Cosmos Predict-2, NVIDIA’s latest WFM, generates high-quality synthetic data by predicting future world states from multimodal inputs like text, images and video. This capability is critical for creating temporally consistent, realistic scenarios that accelerate training and validation of AVs and robots. In addition, Cosmos Transfer, a control model that adds variations in weather, lighting and terrain to existing scenarios, will soon be available to 150,000 developers on CARLA, a leading open-source AV simulator. This greatly expands the broad AV developer community’s access to advanced AI-powered simulation tools. Developers can start integrating synthetic data into their own pipelines using the NVIDIA Physical AI Dataset. The latest release includes 40,000 clips generated using Cosmos. Building on these foundations, the Omniverse Blueprint for AV simulation provides a standardized, API-driven workflow for constructing rich digital twins, replaying real-world sensor data and generating new ground-truth data for closed-loop testing. The blueprint taps into OpenUSD’s layer-stacking and composition arcs, which enable developers to collaborate asynchronously and modify scenes nondestructively. This helps create modular, reusable scenario variants to efficiently generate different weather conditions, traffic patterns and edge cases. Driving the Future of AV Safety To bolster the operational safety of AV systems, NVIDIA earlier this year introduced NVIDIA Halos — a comprehensive safety platform that integrates the company’s full automotive hardware and software stack with AI research focused on AV safety. The new Cosmos models — Cosmos Predict- 2, Cosmos Transfer- 1 NIM and Cosmos Reason — deliver further safety enhancements to the Halos platform, enabling developers to create diverse, controllable and realistic scenarios for training and validating AV systems. These models, trained on massive multimodal datasets including driving data, amplify the breadth and depth of simulation, allowing for robust scenario coverage — including rare and safety-critical events — while supporting post-training customization for specialized AV tasks. At CVPR, NVIDIA was recognized as an Autonomous Grand Challenge winner, highlighting its leadership in advancing end-to-end AV workflows. The challenge used OpenUSD’s robust metadata and interoperability to simulate sensor inputs and vehicle trajectories in semi-reactive environments, achieving state-of-the-art results in safety and compliance. Learn more about how developers are leveraging tools like CARLA, Cosmos, and Omniverse to advance AV simulation in this livestream replay: Hear NVIDIA Director of Autonomous Vehicle Research Marco Pavone on the NVIDIA AI Podcast share how digital twins and high-fidelity simulation are improving vehicle testing, accelerating development and reducing real-world risks. Get Plugged Into the World of OpenUSD Learn more about what’s next for AV simulation with OpenUSD by watching the replay of NVIDIA founder and CEO Jensen Huang’s GTC Paris keynote. Looking for more live opportunities to learn more about OpenUSD? Don’t miss sessions and labs happening at SIGGRAPH 2025, August 10–14. Discover why developers and 3D practitioners are using OpenUSD and learn how to optimize 3D workflows with the self-paced “Learn OpenUSD” curriculum for 3D developers and practitioners, available for free through the NVIDIA Deep Learning Institute. Explore the Alliance for OpenUSD forum and the AOUSD website. Stay up to date by subscribing to NVIDIA Omniverse news, joining the community and following NVIDIA Omniverse on Instagram, LinkedIn, Medium and X.

El estudio barcelonés Domestic Data Streamers lanza un proyecto que usa inteligencia artificial generativa para reconstruir recuerdos perdidos o nunca registrados. “No son fotografías del pasado, son espejos poéticos del recuerdo”, explican. Con la c

Gardenful / TAOASave this picture!© Tao LeiLandscape Architecture•Beijing, China Architects: TAOA Area Area of this architecture project Area:  227 m² Year Completion year of this architecture project Year:  2024 Photographs Photographs:Tao LeiMore SpecsLess Specs Save this picture! Text description provided by the architects. This is an urban garden built for private use. As a corner of the city, I hope to fill the whole garden with abundant nature in this small space. The site is an open space in a villa compound, surrounded by a cluster of European-style single-family villas typical of Chinese real estate. Modern buildings greatly meet the requirements of indoor temperature and humidity comfort because of their complete facilities, but the building also has a clear climate boundary, cutting off the connection between indoor and outdoor, but also cut off the continuity of nature and life.Save this picture!Save this picture!There is no simple definition of the project as a garden or a building, too simple definition will only fall into the narrow imagination, the purpose is only to establish a place that can accommodate a piece of real nature, can give people shelter, can also walk in it. It is the original intention of this design to build a quiet place where you can be alone, a semi-indoor and semi-outdoor space, and re-lead the enclosed life to the outdoors and into the nature.Save this picture!Save this picture!The square site in the middle of the garden, which is a relatively independent space, the top shelter provides a comfortable life and cozy, the middle of the garden exposed a sky, sunshine and rain and snow will be staged here. With the corresponding land below, the trees and vegetation of the mountains are introduced into it, maintaining the most primitive wildness. To remain wild in this exquisite urban space, in this abstract geometric order, will naturally get rid of the wild gas of the original nature. A spatial transformation is made on both sides to the north, through the stairway and the upward pull of the roof space, extending the narrow auxiliary garden, which has no roof and is therefore bright, maintaining a different light and shade relationship from the central garden, which is filled with rocks and plants transplanted from the mountains.Save this picture!Save this picture!Save this picture!The structure of the garden is thin and dense synthetic bamboo, and the cross combination of dense structures forms a partition of the space, like a bamboo fence, forming a soft boundary. The interior of the space is lined with wooden panels, and the exterior is covered with thin and crisp aluminum panels. The "bridge" made of stone panels passes through different Spaces, sometimes standing between the bamboo structures, sometimes crossing the rocks, walking between them. Moving between order and wildness.Save this picture!Nature is difficult to measure, and because of its rich and ever-changing qualities, nature provides richness to Spaces. This is from the mountains to large trees, rocks, small flowers and plants, as far as possible to avoid artificial nursery plants. The structure of the garden will geometrically order the nature, eliminating the wild sense of nature. The details of nature can be discovered, and the life force released can be unconsciously perceived. The nature of fragments is real, is wild, and does not want to lose vitality and richness because of artificial transplantation. The superposition of wild abundance and modern geometric space makes it alive with elegance and decency.Save this picture!Save this picture!The nature is independent of the high-density urban space, becoming an independent world, shielding the noise of the city. These are integrated into a continuous and integral "pavilion" and "corridor" constitute the carrier of outdoor life of the family, while sheltering from the wind and rain, under the four eaves also create the relationship between light and dark space, the middle highlights the nature, especially bright, and becomes the center of life. From any Angle one can see a picture of hierarchy and order, a real fragment of nature, built into a new context by geometric order. The richness of nature is therefore more easily perceived, and the changes of nature are constantly played out in daily life and can be seen throughout the year.Save this picture! Project gallerySee allShow less Project locationAddress:Beijing, ChinaLocation to be used only as a reference. It could indicate city/country but not exact address.About this officeTAOAOffice••• Published on June 15, 2025Cite: "Gardenful / TAOA" 15 Jun 2025. ArchDaily. Accessed . <https://www.archdaily.com/1028408/gardenful-taoa&gt ISSN 0719-8884Save想阅读文章的中文版本吗?满园 / TAOA 陶磊建筑是否 You've started following your first account!Did you know?You'll now receive updates based on what you follow! Personalize your stream and start following your favorite authors, offices and users.Go to my stream

Join the event trusted by enterprise leaders for nearly two decades. VB Transform brings together the people building real enterprise AI strategy. Learn more When DeepSeek released its R1 model this January, it wasn’t just another AI announcement. It was a watershed moment that sent shockwaves through the tech industry, forcing industry leaders to reconsider their fundamental approaches to AI development. What makes DeepSeek’s accomplishment remarkable isn’t that the company developed novel capabilities; rather, it was how it achieved comparable results to those delivered by tech heavyweights at a fraction of the cost. In reality, DeepSeek didn’t do anything that hadn’t been done before; its innovation stemmed from pursuing different priorities. As a result, we are now experiencing rapid-fire development along two parallel tracks: efficiency and compute.  As DeepSeek prepares to release its R2 model, and as it concurrently faces the potential of even greater chip restrictions from the U.S., it’s important to look at how it captured so much attention. Engineering around constraints DeepSeek’s arrival, as sudden and dramatic as it was, captivated us all because it showcased the capacity for innovation to thrive even under significant constraints. Faced with U.S. export controls limiting access to cutting-edge AI chips, DeepSeek was forced to find alternative pathways to AI advancement. While U.S. companies pursued performance gains through more powerful hardware, bigger models and better data, DeepSeek focused on optimizing what was available. It implemented known ideas with remarkable execution — and there is novelty in executing what’s known and doing it well. This efficiency-first mindset yielded incredibly impressive results. DeepSeek’s R1 model reportedly matches OpenAI’s capabilities at just 5 to 10% of the operating cost. According to reports, the final training run for DeepSeek’s V3 predecessor cost a mere $6 million — which was described by former Tesla AI scientist Andrej Karpathy as “a joke of a budget” compared to the tens or hundreds of millions spent by U.S. competitors. More strikingly, while OpenAI reportedly spent $500 million training its recent “Orion” model, DeepSeek achieved superior benchmark results for just $5.6 million — less than 1.2% of OpenAI’s investment. If you get starry eyed believing these incredible results were achieved even as DeepSeek was at a severe disadvantage based on its inability to access advanced AI chips, I hate to tell you, but that narrative isn’t entirely accurate (even though it makes a good story). Initial U.S. export controls focused primarily on compute capabilities, not on memory and networking — two crucial components for AI development. That means that the chips DeepSeek had access to were not poor quality chips; their networking and memory capabilities allowed DeepSeek to parallelize operations across many units, a key strategy for running their large model efficiently. This, combined with China’s national push toward controlling the entire vertical stack of AI infrastructure, resulted in accelerated innovation that many Western observers didn’t anticipate. DeepSeek’s advancements were an inevitable part of AI development, but they brought known advancements forward a few years earlier than would have been possible otherwise, and that’s pretty amazing. Pragmatism over process Beyond hardware optimization, DeepSeek’s approach to training data represents another departure from conventional Western practices. Rather than relying solely on web-scraped content, DeepSeek reportedly leveraged significant amounts of synthetic data and outputs from other proprietary models. This is a classic example of model distillation, or the ability to learn from really powerful models. Such an approach, however, raises questions about data privacy and governance that might concern Western enterprise customers. Still, it underscores DeepSeek’s overall pragmatic focus on results over process. The effective use of synthetic data is a key differentiator. Synthetic data can be very effective when it comes to training large models, but you have to be careful; some model architectures handle synthetic data better than others. For instance, transformer-based models with mixture of experts (MoE) architectures like DeepSeek’s tend to be more robust when incorporating synthetic data, while more traditional dense architectures like those used in early Llama models can experience performance degradation or even “model collapse” when trained on too much synthetic content. This architectural sensitivity matters because synthetic data introduces different patterns and distributions compared to real-world data. When a model architecture doesn’t handle synthetic data well, it may learn shortcuts or biases present in the synthetic data generation process rather than generalizable knowledge. This can lead to reduced performance on real-world tasks, increased hallucinations or brittleness when facing novel situations.  Still, DeepSeek’s engineering teams reportedly designed their model architecture specifically with synthetic data integration in mind from the earliest planning stages. This allowed the company to leverage the cost benefits of synthetic data without sacrificing performance. Market reverberations Why does all of this matter? Stock market aside, DeepSeek’s emergence has triggered substantive strategic shifts among industry leaders. Case in point: OpenAI. Sam Altman recently announced plans to release the company’s first “open-weight” language model since 2019. This is a pretty notable pivot for a company that built its business on proprietary systems. It seems DeepSeek’s rise, on top of Llama’s success, has hit OpenAI’s leader hard. Just a month after DeepSeek arrived on the scene, Altman admitted that OpenAI had been “on the wrong side of history” regarding open-source AI.  With OpenAI reportedly spending $7 to 8 billion annually on operations, the economic pressure from efficient alternatives like DeepSeek has become impossible to ignore. As AI scholar Kai-Fu Lee bluntly put it: “You’re spending $7 billion or $8 billion a year, making a massive loss, and here you have a competitor coming in with an open-source model that’s for free.” This necessitates change. This economic reality prompted OpenAI to pursue a massive $40 billion funding round that valued the company at an unprecedented $300 billion. But even with a war chest of funds at its disposal, the fundamental challenge remains: OpenAI’s approach is dramatically more resource-intensive than DeepSeek’s. Beyond model training Another significant trend accelerated by DeepSeek is the shift toward “test-time compute” (TTC). As major AI labs have now trained their models on much of the available public data on the internet, data scarcity is slowing further improvements in pre-training. To get around this, DeepSeek announced a collaboration with Tsinghua University to enable “self-principled critique tuning” (SPCT). This approach trains AI to develop its own rules for judging content and then uses those rules to provide detailed critiques. The system includes a built-in “judge” that evaluates the AI’s answers in real-time, comparing responses against core rules and quality standards. The development is part of a movement towards autonomous self-evaluation and improvement in AI systems in which models use inference time to improve results, rather than simply making models larger during training. DeepSeek calls its system “DeepSeek-GRM” (generalist reward modeling). But, as with its model distillation approach, this could be considered a mix of promise and risk. For example, if the AI develops its own judging criteria, there’s a risk those principles diverge from human values, ethics or context. The rules could end up being overly rigid or biased, optimizing for style over substance, and/or reinforce incorrect assumptions or hallucinations. Additionally, without a human in the loop, issues could arise if the “judge” is flawed or misaligned. It’s a kind of AI talking to itself, without robust external grounding. On top of this, users and developers may not understand why the AI reached a certain conclusion — which feeds into a bigger concern: Should an AI be allowed to decide what is “good” or “correct” based solely on its own logic? These risks shouldn’t be discounted. At the same time, this approach is gaining traction, as again DeepSeek builds on the body of work of others (think OpenAI’s “critique and revise” methods, Anthropic’s constitutional AI or research on self-rewarding agents) to create what is likely the first full-stack application of SPCT in a commercial effort. This could mark a powerful shift in AI autonomy, but there still is a need for rigorous auditing, transparency and safeguards. It’s not just about models getting smarter, but that they remain aligned, interpretable, and trustworthy as they begin critiquing themselves without human guardrails. Moving into the future So, taking all of this into account, the rise of DeepSeek signals a broader shift in the AI industry toward parallel innovation tracks. While companies continue building more powerful compute clusters for next-generation capabilities, there will also be intense focus on finding efficiency gains through software engineering and model architecture improvements to offset the challenges of AI energy consumption, which far outpaces power generation capacity.  Companies are taking note. Microsoft, for example, has halted data center development in multiple regions globally, recalibrating toward a more distributed, efficient infrastructure approach. While still planning to invest approximately $80 billion in AI infrastructure this fiscal year, the company is reallocating resources in response to the efficiency gains DeepSeek introduced to the market. Meta has also responded, With so much movement in such a short time, it becomes somewhat ironic that the U.S. sanctions designed to maintain American AI dominance may have instead accelerated the very innovation they sought to contain. By constraining access to materials, DeepSeek was forced to blaze a new trail. Moving forward, as the industry continues to evolve globally, adaptability for all players will be key. Policies, people and market reactions will continue to shift the ground rules — whether it’s eliminating the AI diffusion rule, a new ban on technology purchases or something else entirely. It’s what we learn from one another and how we respond that will be worth watching. Jae Lee is CEO and co-founder of TwelveLabs. Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.