NVIDIA Reshapes Robotics With Physical AI Data Factory Blueprint

News Summary 

  • Blueprint organizes and processes large data, generates synthetic data, applies reinforcement learning, and evaluates physical AI models for vision agents. Robotics and self-driving vehicles.  
  • Cloud service providers such as Microsoft Azure and Nebius use Blueprint. They turn large-scale computing power into agent-driven data. Production tools are ready for use.  
  • Top physical AI developers are using the blueprint to speed up the development of robotics, vision AI agents, and self-driving vehicles.  

At GTC, NVIDIA announced the NVIDIA Physical AI Data Factory Blueprint. This open reference architecture automates and unifies training data generation, improvement, and evaluation, reducing cost, speeding up processes, and simplifying large-scale physical AI training.  

With the blueprint, developers expand small training datasets into large, varied ones using Nvidia, Cosmos, Open World Base models, and top coding agents, including rare cases that are costly or difficult to collect in real life.  

NVIDIA is working with Microsoft, Azure, and Nebius to connect the open blueprint to their cloud services. This lets developers use powerful computing resources to create large amounts of training data. Companies like FieldAI, Hexagon Robotics, Linker Vision, Milestone Systems, RoboForce Skild AI, Teradyne Robotics, and Uber are already using the blueprint to accelerate development of robotics vision AI agents and self-driving vehicles.  

“Physical AI is the next frontier of the AI revolution, where success depends on the ability to generate massive amounts of data,” said Rev. Lebaredian, Vice President of Omniverse and Simulation Technologies at Nvidia. Together with cloud leaders, we are adding a new kind of agentic engine that transforms compute into high-quality data, enabling the next generation of self-governing systems and robots to come to life. In this new era, compute is data.  

A Unified Engine for Physical AI Development 

Physical AI improves as data, computing power, and model size grow. The Physical AI data factory blueprint acts as a single reference point. It helps teams turn raw data into training sets for models using automated workflows.  

  • Curate and search: Nvidia Cosmos Curator Manages, Improves, and Labels Large Real World and Synthetic (artificially generated) datasets  
  • Augment and multiply: cosmos transfer greatly increases and diversifies the selected data, combining real and simulated inputs to better cover rare and unusual situations across different environments and lighting conditions.  
  • Evaluate and validate Nvidia Cosmos Evaluator, which uses Cosmos Reason and is now on GitHub. Check scores and filter generated data by sensory accuracy and training readiness.  

NVIDIA is using the Physical AI Data Factory Blueprint to train and test NVIDIA Alpamayo. Alpamayo, the world’s first open reasoning-based vision-language-action model for long-tail autonomous driving. Skild AI uses the blueprint to improve general-purpose robot-based models. Uber uses it to speed up autonomous vehicle development.  

Agent Driven Orchestration At Scale 

Many robotics developers do not have the resources to set up and manage complex AI systems needed to generate data at scale.  

NVIDIA Osmo is an open source orchestration framework that brings these workflows together across multiple computing environments. It reduces manual work, allowing developers to focus on building their models.  

Osmo now works with top-count agents like Claude Code, OpenAI Codex, and Cursor. This enables AI agents to manage resources, resolve bottlenecks, and accelerate model deployment at scale.  

Powering The Global Physical AI Ecosystem 

Cloud service providers are essential for fast AI infrastructure, machine learning operations, and orchestration services. Developers use these to build the process of the pro Build and launch physical AI at scale.  

Microsoft Azure is adding the Physical AI Data Factory blueprint to an open Physical AI toolchain. Now on GitHub, the blueprint connects with Azure services such as Azure IoT Operations, Microsoft Fabric, Realtime Intelligence, and Microsoft Foundry to provide businesses with agent-driven workflows for quickly and at-scale training and testing of physical AI systems.  

FieldAI, Hexagon Robotics, Inca, Vision, and Teradyne Robotics are among the first to try the Azure Physical AI tool chain. They use it to speed up and scale data generation, improvement, and evaluation for perception, mobility, and reinforcement learning systems.  

Nebius has added Osmo to AI Cloud, enabling developers to use the blueprint to set up data pipelines ready for production and tailored to their needs. Navy S’s system supports the entire physical AI stack, combining NVIDIA RTX Pro 6000 Blackwell Server Edition GPUs with fast object storage, built-in data management and labeling, serverless execution, and managed inference.  

Early users such as Milestone Systems, Voxel 51, and RoboForce are using the blueprint on Nebius infrastructure to accelerate the development of video analytics, AI agents, self-driving vehicles, and industrial humanoid robots.  

The NVIDIA Physical AI Data Factory Blueprint launches on GitHub in April.  

Source: NVIDIA Announces Open Physical AI Data Factory Blueprint to Accelerate Robotics, Vision AI Agents and Autonomous Vehicle Development 

FBI CISA Alert: State Actors Target Commercial Messaging Apps 

FBI and CISA warn of ongoing Russian-linked phishing targeting messaging accounts.  

Earlier this month, we reported on a large phishing campaign targeting Signal and WhatsApp accounts of senior officials, military personnel, civil servants, and journalists.  

The FBI, CISA, and European intelligence warn that these tactics now target commercial messaging apps rather than breaking end-to-end encryption. Attackers steal access to individual accounts.  

Our last article covered Dutch intelligence warnings on Russian actors contacting high-value targets on Signal and WhatsApp, posing as support or security bots. The new PSA shows these groups now run global phishing campaigns with evidence of thousands of compromised accounts.  

Attackers use social engineering to add devices and listen in without breaking encryption.  

Targets include US officials, military, politicians, journalists, and businesses. These techniques threaten all users.  

This demonstrates that the threat extends far beyond diplomats or generals. Because these techniques are easy to copy, they put all users, including businesses and individuals, at risk.  

How to Protect Your Accounts 

As the PSA puts it:  

Phishing remains one of the most unsophisticated yet effective means of cyber compromise, frequently rendering other protections irrelevant.  

This situation calls for some basic security steps:  

  • Treat unexpected support messages in apps as suspicious. Legitimate support will not ask for verification calls, PINs, or passwords in chat. For account warnings, do not click message links; instead, check access settings or visit the official site yourself.  
  • Never share SMS verification codes or app PINs. These prove phone control—sharing means giving up your account. Treat all requests for codes as scams.  
  • Be careful what you discuss and with whom. Even with encryption, some topics are too sensitive for chat apps.  
  • Use extra security features. Enable registration lock, PIN, and device change alerts to prevent re-registration without a code. Store your PIN in a password manager. If attackers access your chats or backups, they may see content. These measures limit damage but are not foolproof.  

What To Do If You Think Your Count Was Hijacked 

If you think someone has taken over your messaging account, follow these steps:  

  1. Re-register your number in the app immediately to remove other devices.  
  1. Revoke all linked devices and change app PINs or lock codes.   
  1. Warn contacts that someone may have impersonated you and ask them to be cautious with recent messages.  
  1. Review Recent Conversations for Signs of Data Theft (for Example, Shared IDs, documents, or Passwords that should now be considered exposed).  
  1. Report the incident to the app provider and, if needed, to authorities like the FBI’s IC3 or your national agency.  

Act quickly to limit how long attackers can use your account.  

Source: FBI, CISA warn of Russian hackers hijacking Signal and WhatsApp accounts

Microsoft RSAC alert: AI Evolves from Tool to Cyber-attack Surface. 

Over the past year, the main topic at the intersection of AI and cybersecurity has been speed. While speed is important, it is not the biggest change in today’s threat landscape. Threat actors, from nation-states to cybercrime groups, now use AI to plan, refine, and maintain their cyberattacks. Their goals remain the same, but the pace, repetition, and scale of AI-powered attacks raise the stakes.  

Still, just like defenders, most attackers today have a human involved, not fully autonomous AI running the show. AI is making every stage of the attack process easier, helping attackers research faster, write more convincing vectors, create malware, and sort through stolen data. Security leaders I met at RSAC 2026 this week are now shifting their resources and strategies to stay ahead of these changes.  

The Operational Reality: Embedded, Not Emerging 

The scale of current threats is too big to ignore. VC activity in every region. The United States accounts for almost 25% of what we have, with the United Kingdom, Israel, and Germany following. The volume reflects real economic and geopolitical factors.  

The major shift isn’t location but attackers’ methods; they use AI throughout their processfrom information gathering to malware development and post-breach actions. Stealing credentials, making money, or spying remain the aims, but attacks are now more precise, persistent, and larger in scale.  

Email Is Still the Fastest Inroad 

Email is still the quickest and most affordable way for attackers to get in. What’s different now is how much better AI makes the messages that trick people into clicking.  

With AI in phishing campaigns, click-through rates have jumped to 54% from about 12% with older methods, resulting in a 450% boost in effectiveness. Not because there are more emails, but because the messages are more precise. AI helps attackers fine-tune content and adjust messages for certain roles, making it easier to trick people. When this improved targeting is paired with tools that implement multi-factor authentication (MFA), phishing becomes more resilient, more focused, and much harder to stop on a large scale.  

A450% jump in click-through rates redefines organizational risk, showing that AI enables not just more but better attacks.  

Tycoon2FA: What Industrial Scale Cybercrime Looks Like 

Tycoon 2FA shows how the group we call Storm-1747 has become more refined and resilient. Learning how this operation worked helps us see where threats are going. It also sparked discussions at RSSC 2026 about the broader ecosystem rather than just individual attackers.  

Typhoon 2FA was not a phishing kit; it was a subscription platform that generated tens of millions of phishing emails per month. It was linked to nearly 100,000compromised organizations since 2023. At its peak, it accounted for roughly 62% of all phishing incidents Microsoft blocked each month. This operation specializes in adversary-in-the-middle attacks aimed at defeating MFA. It intercepted credentials and session tokens in real time, allowing attackers to authenticate as legitimate users without triggering alerts even after passwords were reset.  

However, the bigger shift is in group organization. Storm 1747 used specialized services for fishing templates, infrastructure, and email sending. Access, sales, and creating an assembly-line–like approach to identity theft. Services could be mixed, scaled, and subscribed to as needed.  

This model has shifted the conversation. It’s no longer about one skilled attacker but about an entire ecosystem that makes access easier for anyone who joins in. That’s what AI is doing across the threat landscape giving advanced tools to everyone. Key takeaway: AI-driven ecosystems democratize attack capabilities for all threat actors.  

Disruption: Closing the Threat Intelligence Loop 

Earlier this month, our digital crimes unit, working with Europol and industry partners, took down Tycoon 2FA and seized 330 domains. But the real goal wasn’t just to remove websites; it was to put pressure on the supply chain. Today’s cybercrime relies on scalable service models that make it easier for more people to get involved. Identity is the main target, and bypassing MFA is now a standard feature. Shutting down one service forces attackers to adapt, and ongoing pressure breaks up their ecosystem. By hitting the financial side of a tax, we can change the landscape. Key takeaway: Disruption efforts should target criminal supply chains to reduce future risk.  

Every time we disrupt an attack, it generates a signal. The signal feeds intelligence. Each time we stop an attack, we get new information. The information enhances our intelligence, improving our detection. Better detection leads to faster responses. This is how we turn attacker actions into stronger defenses and how our efforts add up over time. Microsoft stands out because we can observe, act, and share intelligence at scale and we have a significant impact when we put it into practice. AI doesn’t appear in just one phase of an attack; it spans the entire life cycle. At RCC 2026, this week, I offered a frame to help defenders rank their response:  

  • In reconnaissance, AI accelerates infrastructure discovery and persona development, compressing the time between target selection and first contact.  
  • In resource development, AI generates forged documents, polishes, social engineering, narratives, and supports infrastructure at scale.  
  • For initial access, AI refines voice-overlays, deepfakes, and message customization using scraped data, producing lures that are increasingly difficult to distinguish from authentic communications.  
  • In persistence and evasion, AI scales fake identities and automates communication, preserving the attacker’s presence while blending into normal activity.  
  • In weaponization, AI enables malware development, payload regeneration, and real-time debugging, producing tooling that adapts to the victim’s environment rather than relying on static signatures.  
  • In post-compromise operations, AI adapts tooling to the specific victim environment and, in some cases, automates ransom negotiation.  

The goals remain: Dash stealing credentials, making money, and spying. What’s new is the pace and scale. Column attackers repeat and improve. Test and refine much more quickly. AI isn’t just enabling faster attacks; it’s making them better.  

What Comes Next 

During my sessions at RSSC 2026 this week, I discussed several key themes. That shows how AI is changing the threat landscape, a threat model. The scenarios we prepare for have changed. The barrier to launching sophisticated attacks has collapsed. What once required the resources of a nation-state or well-organized criminal enterprise is now available to a motivated individual with the right tools and the patience to use them. The techniques have not fundamentally changed; the precision, velocity, and volume have.  

The second theme is the software supply chain. It’s not only about compliance, you need to know what software and agents you have and how they behave. The agent ecosystem will soon be the most targeted part of any business. If organizations can’t answer basic questions about their software, they won’t be able to protect it.  

The third theme highlights the value of human talent in security operations using agency systems at scale. The traditional security analyst role is shifting from practitioner to orchestrator; talent models must catch up, and technology now helps prevent errors. Auditability of agent decisions is a governance standard, not just a goal. The future security operations center needs different defenders.  

Now is the time to guide with a clear strategy, set priorities, and build stronger accountability for agentic systems.  

If AI is present throughout the attack life cycle, our intelligence and defenses must be there too. Microsoft threat intelligence will continue to track, share, and act on what we see in real time. The patterns are clear, and the intelligence is available. Key takeaway: Ongoing monitoring and response are essential in the AI-driven threat landscape.  

To find out more about Microsoft security solutions, visit our website. You can also bookmark our security blog for security expert updates and follow us on LinkedIn (Microsoft Security) and X (@MSFTsecurity) for the latest cybersecurity news.

Source: Threat actor abuse of AI accelerates from tool to cyberattack surface

CISA Urges Immediate Hardening After US Organization Endpoint Breach

Following a series of cyberattacks last week, including an ongoing incident at Stryker, the US Cybersecurity and Infrastructure Security Agency (CISA) is urging organizations to rapidly strengthen endpoint management systems to protect against Iran-linked hackers and other threats.  

On March 11th, Michigan-based Stryker was targeted through its Microsoft Intune endpoint management systems. Microsoft devices were wiped, and data was stolen, causing major disruptions and, in some cases, affecting primary healthcare services.  

The Iranian hacktivist group Handala quickly claimed responsibility, saying the attack was retaliation for the ongoing Israeli–US conflict with Iran.  

CISA is working with US partners, including the FBI, to identify further threats and risks.   

To defend against similar malicious activity that misuses legitimate endpoint management software, CISA urges organizations to implement Microsoft’s newly released best practices for securing Microsoft Intune, the agency said in a statement.  

CISA also noted that these recommendations apply not only to Intune but also to other endpoint management software.  

Organizations should use Intune role-based access controls to ensure users have only the permissions they need for daily tasks. They should also enforce phishing-resistant multi-factor authentication and strong privileged access controls. For sensitive or high-impact actions in Microsoft Intune, access policies should require approval from multiple administrators.  

Global Peers 

Offering a global perspective, Keven Knight, CEO of Talion, said CISA’s guidance is relevant outside the US. He expects similar alerts from other agencies worldwide. For example, the UK’s National Cyber Security Center (NCSC) has already issued a wider cyber alert related to the Iran conflict.  

The Stryker attack was striking because its aim was destruction, not money. There was no ransom or way to recover the data. Backups were unavailable, forcing a complete rebuild.  

Given the current geopolitical situation, it’s likely these destructive attacks will occur more frequently. Strengthening endpoints, using least privilege access, making regular backups, and practicing security response plans are all essential steps.  

He added that since these attacks target countries, organizations must be ready.  

Tip Of The Iceberg 

The Stryker attack is the most high-profile case of Iran’s cyber retaliation against the US, occurring two days after progress in nuclear talks. Experts warn this could be just the beginning.  

Michael Smith, CTO at DigiCert, has tracked nearly 4,500 threats from 43 active groups, noting that the most recent regional attacks are attempts to intimidate rather than destroy.  

Smith noted many attacks go unreported. “We’ve seen BDOS attacks stopped before outages. We also monitor active hacktivist discussions for signs and warnings.”  

Smith said these attacks show foreign audiences that geographic boundaries do not limit the reach of threats, reinforcing intimidation.  

Adding to this, Kathryn Raines, who leads Flashpoint’s cyber threat intelligence, said cyber activity related to this conflict is now focusing more on disrupting organizations.  

She continued: “Groups like Handala are making bigger claims about large-scale attacks.” These include destroying data, exposing sensitive information, and information from private companies and individuals. Even if some claims are hard to verify, they still create uncertainty. This can seriously affect trust, operations, and response efforts.  

CISA has identified malicious cyber activity targeting endpoint management systems at US organizations following the March 11, 2020, cyberattack on Stryker Corporation that impacted its Microsoft environment. To help prevent similar incidents, CISA encourages organizations to strengthen their endpoint management system configurations by following the recommendations and resources in this alert. CISA is also working closely with federal partners, including the FBI, to identify additional threats and determine mitigation steps.  

To reduce risk from endpoint threats, promptly apply Microsoft’s best practices for Microsoft Intune and other endpoint tools.  

  • Use Microsoft Intune’s Role-Based Access Control (RBAC) to grant each role only the permissions needed for daily tasks. These permissions define what actions each role can take and which users or devices those actions apply to.  

Enforce phishing-resistant multi-factor authentication for all privileged accounts and follow best practices for privileged access.  

  • Use Microsoft Entra ID features such as conditional access, MFA, risk-based policies, and privileged access policies to prevent unauthorized privileged actions in Intune.  

Require multi-admin approval for Microsoft Intune access policies.  

  • Establish policies requiring secondary administrator approval before executing sensitive or high-impact operations, including device wipes and application deployments.  

Source: CISA Urges Endpoint Management System Hardening After Cyberattack Against US Organization,Cisa tells US organisations to harden endpoint management after Stryker attack 

Tesla Shares Drop as Energy-Storage Deployments Miss Expectations 

Tesla’s shares fell after it released its latest update for investors, which showed that energy-storage deployments were significantly lower than expected given market trends. Although the company has tried to make its energy business a large part of future growth, much like electric vehicles, the fact that it has provided deployment numbers that were much lower than expected raises questions about execution capabilities, demand visibility, and near-term revenue from these projects. Additionally, this situation illustrates just how difficult it is to build and expand energy infrastructure to meet the ever-increasing global demand for clean/renewable energy.  

Energy Storage as a Growth Pillar  

To continue developing its strategic imperatives, Tesla’s Energy Storage Division, which includes both Commercial Battery Systems and Residential Products, has become an increasingly important aspect of the company. By providing effective means to store electricity generated from renewable energy sources (wind, solar, etc.) and deliver it back to the grid once stored, these divisions also assist in fulfilling Tesla’s obligations. 

Investors generally see this division as a very important source of diversification beyond automotive revenues, especially as the world moves towards cleaner, more sustainable forms of energy production. Recent deployment levels, however, illustrate the differences between the long-term potential and the short-term results of the division’s activity and cast significant doubt in the minds of Tesla’s investors on how quickly the company will be able to achieve the level of profitability currently anticipated from its energy business.  

Missed Expectations and Market Reaction  

Analysts had estimated deployment numbers to be much higher than the actual numbers reported for Q3’20, causing the stock market to react negatively. Since investors are constantly looking for confirmation of Tesla’s ongoing growth in its energy operations, any divergence from analyst expectations may affect their perception of Tesla stock and the direction of the overall market. 

The decreases in share price point to greater execution risks (e.g., production capacity constraints, supply chain disruptions, and/or delays in completing projects). While Tesla’s long-term outlook for its energy storage efforts is very good, I was surprised that investors were as sensitive to Q3’20 operational performance metrics and delivery timeframes.  

Factors Behind the Shortfall  

Low levels of energy storage installations are due to a variety of factors, including supply chain disruptions affecting key battery components and, in turn, production schedules and project due dates.  

In addition to supply chain issues, the complexity of installation and associated regulatory approvals for larger-scale energy systems creates delays in their installation and integration into existing infrastructure. For instance, utility changes in financial support or incentives to establish an energy storage or production facility can create fluctuations in installation activity, notwithstanding significant demand for increased energy storage capacity.  

Balancing Automotive and Energy Operations  

Tesla is focused on electric vehicles and energy solutions, which present both opportunities and challenges. While the automotive business continues to generate revenue and attention, the energy segment requires substantial investments and operational coordination to scale appropriately.  

By effectively balancing resources between these two areas, they can maintain reliable performance across the company. An operationally efficient Tesla will experience significant delays for either segment due to a manufacturing or supply chain constraint.  

Long-Term Potential of Energy Storage  

Although recent obstacles may have deterred some investors from Tesla’s success, the company still has significant long-term potential, given the growing demand for energy storage as global renewable energy adoption continues to expand. One way this growing demand will affect Tesla is through its innovative battery technology, which enables large-scale battery storage and efficient power dispatch.  

As such, it is anticipated that continued research and manufacturing investment, as well as partnerships, will drive significant growth over time, even if the company experiences short-term fluctuations.  

Competitive Landscape  

Newly competitive markets in the energy storage arena are currently being developed, with many companies investing in battery technology and attempting to create solutions at the grid scale. Increasingly, utility companies, industrial and large-enterprise companies, and technology providers are seeking to capitalise on the rapid growth in demand for energy storage.  

Tesla has a first-mover advantage and an integrated strategy, so they will need to focus on consistent execution and innovation to maintain their market leadership. Their competitors are also evolving their capabilities, placing even greater competitive pressure on Tesla in pricing, performance, and delivery speed.  

Operational and Execution Challenges  

Deploying large-scale energy storage entails manufacturing, transporting, installing, and connecting energy storage systems to other energy facilities; each of these areas presents operational challenges that could hinder performance as a large-scale project.  

From ensuring high-quality production to complying with regulations to managing the various parties supporting the system’s deployment, addressing these challenges will improve reliability in future periods and reflect the experiences of the current period.  

Investor Outlook and Confidence  

Investors’ faith in Tesla’s energy division hinges on its ability to demonstrate consistent growth and execution. Although short-term misses can vary considerably and affect investors’ sentiment, long-term investor confidence (over the 6-year period) stems from the company’s marketing of its strategy and overall technological capabilities.  

Maintaining investor trust and enabling future growth of the energy division (which has been hampered by poor execution) requires Tesla to provide clear communication, transparent reporting, and consistent performance.  

Future Developments and Strategy  

Tesla anticipates ongoing investment in its energy storage business by increasing manufacturing and supply chain resilience, and in their manufacturing could increase the efficiency of the battery system while lowering costs over time.  

Additionally, the company might work more closely with utilities and governments to help facilitate large-scale energy projects, aligning itself with the worldwide movement towards renewable energy systems.  

A Critical Phase for Tesla’s Energy Business  

Tesla’s stock price decline after its missed deployment reiterates how vital its energy segment is relative to the market’s overall perception. If the company is going to expand on its energy presence, it must continue delivering on its commitments and maintain momentum.  

As competition increases and demand for sustainable energy products and solutions grows, Tesla’s energy-storage division faces both opportunities and operational challenges.

Sources: Investor Relations

NVIDIA Shifts PC Computing Toward Agent Computers Via NemoClaw 

NemoClaw can be installed with a single command, making it easy to add security and privacy for always‑on OpenClaw agents. It runs in the cloud, on premises, and on NVIDIA GeForce RTX PCs, NVIDIA DGX Station, and NVIDIA DGX Spark.  

At GTC, Nvidia announced the Nvidia NemoClaw stack for the OpenClaw Agent platform. With NemoClaw, users can install NVIDIA Nemotron models and the new NVIDIA OpenShell runtime in one step. This update adds privacy and security controls. As a result, self-evolving autonomous AI agents, called Claws, are now more trustworthy, scalable, and accessible.  

“OpenClaw opened the next frontier of AI to everyone and became the fastest growing open source project in history,” said Jensen Huang, Founder and CEO of Nvidia. Mac and Windows are operating systems for the personal computer. OpenClaw is the operating system for personal AI. This is the moment when the industry is very important. The industry has been waiting for the beginning of a new renaissance in software.  

OpenClaw brings people closer to AI and helps build a world where everyone has their own agents, said Peter Steinberger, creator of OpenClaw. With NVIDIA and the wider ecosystem, we are building the claws and guardrails that let everyone create powerful, secure AI assistants.  

NemoClaw uses the Nvidia Agent Toolkit to optimize OpenClaw with one command. It installs OpenShell, which offers open models and a secure sandbox. This protects data privacy and security for autonomous agents. The Name of Law adds an important infrastructure layer; it gives them the access they need to work well while enforcing security, network, and privacy rules.  

NemoClaw works with any coding agent. With open agents, it can use open models, including Nvidia and Nemo Tron, running locally on a user’s system through a privacy router. Agents can also access advanced models in the cloud by combining local and cloud models. Agents can develop new skills and complete tasks. They can still follow set privacy and security rules.  

Always-on agents need dedicated computing to build software, tools, and complete tasks. Name of claw for open claw can run on any dedicated platform, such as NVIDIA GE4s, RTX PCs and laptops, NVIDIA RTX Pro, Power World Workstations, and NVIDIA DGX Station or DGX Spark AI supercomputers. This setup enables local computation, allowing autonomous agents to run continuously. Stop by NVIDIA, build a Claw event in the GTC Park March 16 to 19 – 1 to 5 p.m. on Monday and 8 a.m. to 5 p.m. on Tuesday through Thursday to customize and deploy an active, always-on AI assistant with NemoClaw or for OpenClaw.  

Source: NVIDIA Announces NemoClaw for the OpenClaw Community 

Amazon Pours $1.4B into US Housing Equity to Create 14,000 Homes  

Amazon has announced a $1.4 billion expansion of its Housing Equity Fund, which will develop and maintain more than 14000 affordable housing units throughout major US metropolitan areas. The investment builds on the company’s earlier commitments to address housing affordability challenges in areas where it has a significant corporate presence. Amazon collaborates with local governments, developers, and non-profit organisations to develop affordable housing solutions for low- and moderate-income communities.  

Expanding Investment in Affordable Housing  

The investment will assist in expanding Amazon’s Housing Equity Fund to combat the American housing crisis by providing additional funds for constructing new housing units and retaining current affordable housing units that would otherwise be sold in the marketplace due to high demand for housing in certain geographic areas and due to rapid growth in those geographic regions. 

The investment is structured to support a mix of housing solutions, including workforce housing for middle-income earners and deeply affordable units for underserved populations. The dual approach exists because housing affordability challenges impact various communities that need different types of solutions that can grow with their needs.  

Addressing the Housing Supply Gap  

The United States has been experiencing a continuing affordable housing crisis. There are many factors contributing to the current imbalance of supply and demand; however, rising construction costs, a lack of available land for development, and government regulation all play significant roles in exacerbating this issue. To help address this housing gap, Amazon is investing in ventures that provide developers with capital to build their projects more quickly, while still enabling them to scale up at a larger scale than previously possible. 

The Housing Equity Fund enables projects to start by providing funding that would otherwise be delayed. The solution increases housing supply while helping maintain stability in rental markets in areas experiencing high demand but facing growing affordability issues for residents.  

Partnerships with Local Communities  

The development of Amazon’s overall plan is largely a collaborative effort with other agencies; therefore, all projects addressed through the partnership process should be aligned to reflect both community and local development priorities to maximise their effectiveness.   

The partnership also enables Amazon to support local communities through investments in housing solutions that follow sustainable development practices, meet residents’ needs, and expand access to critical services, public transportation, and employment opportunities. 

Economic and Social Impact  

Creating affordable housing developments provides economic and social benefits to communities, going beyond just providing residential units. With more housing options, workers have a greater chance of keeping their jobs because they can locate housing closer to their workplaces, thereby eliminating the hassle of commuting. 

Affordable housing also helps build better communities by providing greater access to educational institutions, medical facilities, and vital community resources. Amazon invests in housing infrastructure to advance its mission of advancing economic mobility and social inclusion, especially in areas experiencing rapid development and rising living costs.  

Corporate Role in Housing Initiatives  

Amazon established its Housing Equity Fund to address social problems, including housing affordability. Large employers who expand into urban regions must meet rising community demands for solutions that mitigate the effects of their growth.  

Amazon establishes itself as a community development partner through its commitment to supporting housing programmes with extensive funding. The system tackles present problems while building enduring solutions that foster beneficial connections with community partners.  

Challenges in Affordable Housing Development  

The process of creating affordable housing continues to require substantial resources despite receiving significant financial backing. The project needs to complete three different requirements from regulatory bodies, zoning regulations, and construction obstacles, which will impact both costs and project duration. Maintaining long-term housing affordability requires both strategic development and continuous monitoring.  

Investors face difficulties balancing their financial returns with their social responsibility goals. Amazon leverages its partnerships with seasoned organisations to achieve community outcomes.  

Scaling the Housing Equity Model  

Amazon demonstrates a commitment to expanding its Housing Equity Fund through its ongoing efforts to scale this model into new territories. The company plans to use its financial resources and partnerships to replicate its successful projects while extending their benefits to areas facing similar housing challenges.  

The model requires three elements for expansion: improved investment methods, better project management, and new partnership development. The programme will create a model that other businesses can use to address housing affordability issues as it continues to develop.  

Future Outlook for Affordable Housing  

The United States will continue to face high demand for affordable housing due to population growth, urbanisation, and economic disparities. The demand for affordable housing must be addressed through Amazon’s Housing Equity Fund and other solutions that involve public-private partnerships and policy reforms.  

Sustainable housing solutions will only succeed if ongoing investment, innovation, and collaboration efforts continue.  

Looking Ahead: Building Inclusive Communities  

Amazon received $1.4 billion in investment, indicating that the growing role of business in solving major social challenges is a driving force behind many major companies. Amazon is involved in helping people access affordable housing and establish a more equitable and sustainable community for everyone. 

The initiative demonstrates how private sector investment can support public initiatives by creating permanent benefits that enhance residents’ living conditions. 

Source: https://www.aboutamazon.com/ 

OpenAI Surges with $122 Billion to Lead Next Phase of AI Scaling 

OpenAI has achieved a $122 billion valuation, demonstrating that investors believe in the company’s leadership in artificial intelligence. The increased valuation also underscores OpenAI’s growing influence across the sectors it serves and its commitment to advancing the development of AI systems designed to run applications across a wide range of industries, including productivity and scientific research. The growing demand for generative AI will enable OpenAI to establish itself as the leading organisation advancing AI through its infrastructure development, partnership creation, and product innovation.  

Scaling AI for the Next Generation  

With the rapid emergence of AI technologies, there is a demand for scalable infrastructure to support new levels of complexity in application development across an even broader range of areas than previously possible. As OpenAI continues through its current phase of growth and development, the company is now looking to begin building and deploying more sophisticated models that offer much higher-quality reasoning, the ability to work with multiple modalities (text, video, and images), and instantaneous user interactions. To achieve this expanded capability through their scaling efforts, OpenAI is focused not only on enhancing the quality of performance of each model but also on increasing the overall trustworthiness, security, and dependability of all AI systems developed by OpenAI’s clients. 

With the increasing use of AI in everyday processes, the need for scalable, high-availability infrastructure is becoming imperative. The OpenAI approach acknowledges that the future of AI technology depends on both developing new technologies and implementing them at an accelerated pace and scale.  

Investment in Infrastructure and Computer  

OpenAI heavily invests in computing infrastructure, as it is critical for training and developing large AI models. The reason it requires so much investment is that advanced AI systems of the future will require significant processing power, data, and energy. As a result, computing infrastructure is a competitive advantage or differentiator.  

Therefore, increasing their compute capacity will enable OpenAI to accelerate model training cycles, improve operational efficiency, and handle a rapidly expanding number of end users. OpenAI will partner with cloud service providers and hardware vendors in order to meet these needs. These partnerships and relationships will also allow OpenAI to build a scalable business while maintaining performance levels and reliability.  

This emphasis on infrastructure underscores the growing importance of computing resources in the broader AI development landscape.  

Expanding Product Ecosystem  

OpenAI is expanding its product ecosystem by integrating AI capabilities into a wide range of applications and services. The components of the ecosystem are designed for individual consumers (e.g., chatbots) and provide developer tools and enterprise solutions, thereby creating a single, unified platform on which consumers and businesses can rely.  

Many people are looking for ways to automate their daily lives, create new content, and support decision-making. By providing a scalable, versatile platform for integrating AI into organisations, OpenAI will help improve organisational effectiveness and drive creative solutions across industries. 

With its unified product ecosystem, OpenAI can offer a range of value-based solutions, thereby establishing itself as a major player in this space.  

Enterprise Adoption and Industry Impact  

Organisations are quickly adopting AI across their businesses to leverage sophisticated models for productivity, customer engagement, and data analysis. OpenAI’s tools and services are being embraced across many industries, including finance, health care, education, and software development, demonstrating their versatility and impact.  

As OpenAI provides businesses with enterprise-grade solutions for integrating AI into their existing processes, they contribute to smarter decisions, efficiency through automation, and better outcomes across all aspects of business operation. Businesses universally recognise AI’s potential to transform how they operate and gain a competitive advantage.  

The rapid proliferation of AI across industries increases the need for scalable, reliable, and secure AI systems.  

Competition in the AI Landscape  

The increasing competitiveness of the AI industry has led technology companies to invest significant sums in research facilities and product development. OpenAI has been valued highly for its strong market position; however, to maintain its lead in this sector, it must continually innovate and efficiently execute its strategies.  

The open artificial intelligence market currently has an abundance of competitors, which are driving competition through the introduction of many new products and services. In addition to making ongoing investments in new technology through research, most of OpenAI’s competitors will quickly update their existing products through advances in generative AI models, multimodal technology, and enterprise solutions. Long-term success will depend on differentiating themselves through performance, user experience, and/or the extent of integration of their ecosystems within the overall marketplace. 
 
OpenAI will need to find the right balance between innovating new products and using current technology in ways that are practical to continue competing. 

Challenges in AI Scaling  

AI systems face significant challenges when scaled. The need for increased computing power, along with the costs of energy and data processing, creates unique challenges in maintaining efficient, low-cost models as they grow larger and more complex.  

Safety issues such as ethics and governance are also important to consider in the face of growing AI capabilities. As AI grows more powerful, addressing bias, misinformation, and proper usage will become increasingly important. OpenAI has thus highlighted the need to build safeguards and governance mechanisms for the safe deployment of AI technologies.  

The industry faces very significant challenges in balancing rapid technological innovation with a responsible approach to development.  

Partnerships and Collaboration  

Collaboration is an important part of OpenAI’s strategy because it enables it to leverage diverse skills, expertise, and resources from the broader technology community. OpenAI collaborates with public cloud providers, large companies, and research/academic institutions for AI system creation and deployment on a massive scale.   

OpenAI collaborates with many different types of partners to effectively deploy AI across a broader range of applications and use cases, ultimately delivering real value through technological innovation. By collaborating with others, OpenAI can innovate faster and have a greater impact across multiple sectors and industries. 

Future Developments in AI  

By addressing specific areas where improvements could be made (e.g., reasoning and language capabilities and real-time interaction), OpenAI has made significant advancements over the last several years in building and enhancing its models, with applications (including complex virtual assistants) being employed to help researchers conduct scientific research more easily.  

In addition to building larger models, advancements during the next wave of AI scaling will also include developing more efficient architectures and better integrating them with hardware- and software-based systems. A strong commitment to ongoing research and development will enable the uncovering of new potential and sustaining growth in the rapidly changing AI environment.  

Looking Ahead: The Next Phase of AI Growth  

The rise in OpenAI’s valuation and ongoing investment in scalable building technology demonstrate that AI can be a disruptive force in shaping our society. As OpenAI builds upon its capabilities and infrastructure, it is shaping the future of artificial intelligence by influencing how technology is developed and delivered worldwide.  

The next stage of growth will require OpenAI to provide users/organisations with powerful, reliable, and responsible AI systems that satisfy their needs. 

Source: OpenAI raises $122 billion to accelerate the next phase of AI

Apple Accelerates App Store Growth With 100 New Monetization Metrics 

Discover how users find and interact with your app or game using analytics in App Store Connect. Use these insights to process refunds effectively. Find your app and grow your business on the App Store.  

What’s New 

  • New cohort features let you analyze user behavior by attributes like download date, source, offer, or start date. For example, after launching in a new region, see how quickly those users make purchases compared to users in other regions.   
  • Peer group benchmarks now include download-to-paid conversion and proceeds per download, so you can compare your app with others.  
  • Export new subscription reports via the Analytics Reports API for offline analysis or integration with your data systems.  
  • Apply up to seven filters to your metrics for deeper insights.  

User Acquisition 

Discover how people find and download your apps and games. Improve your marketing by leveraging this information. Access acquisition data for iOS, iPadOS, macOS, tvOS, and VisionOS apps.  

App Downloads 

To see your app or game’s total downloads, check Total downloads in the Overview section of Analytics. This includes new and repeat installs. Break down the data by territory, device, and source for more detail.  

On the Sources page in the Acquisition section, view downloads by browsing, referral, or search. Assess how effective each channel is.  

Conversion Rate  

Your conversion rate shows how often people download your app after seeing it on the app store. It’s calculated by dividing total downloads (both first-time and re-downloads) by unique impressions, which count daily app icon views. For example, if your app icon has 100 unique impressions and 50 downloads in a day, your conversion rate is 50%.  

Measure and track your performance over time, and view conversion rates by source type to identify which sources work best. Filter by a specific source for more details. Use this data to assess how changes to your product page, like a new app icon, affect your conversion rate. Compare your conversion rate to similar apps or games with peer group benchmarks, and make improvements as needed. Apply app store features, including:  

  • Monitor your published in-app events to assess engagement and identify top performers.  
  • Review each custom product page to evaluate its effectiveness in driving downloads and re-downloads.  
  • Compare how different versions of your product page perform and decide whether to implement changes to your main page based on data.  

Campaign Performance 

Add campaign links to your marketing channels social media, email, ads, or cross-promotion to track effectiveness. These direct users to your app’s product page. Create unique links for each variant to track what drives downloads. Apply successful elements to custom pages or tests. For campaign details, check conversion rates and filter by campaign.  

Pre-orders 

Before releasing your app, offer it as a pre-order on the App Store so users can order early and have it automatically download to their devices when it becomes available. Analytics tracks the number of pre-orders; after release, view the downloads and sales from pre-orders.  

User Participation  

Track user activity by monitoring active devices, sessions, retention, and source across Apple platforms. Data only includes users who come to the U. Consented to share diagnostics and usage information.  

Active Devices and Sessions 

Use active devices and sessions to measure app usage. Filter or view by:  

  • Filter by campaign to pinpoint which campaigns attract the most engaged users.  
  • View by territory to assess regional engagement. If sessions per device are low, consider localization, onboarding clarity, or regional preferences.  
  • Filter by device to determine which types, such as iPhone, yield higher engagement. Use these insights to prioritize platform optimization.  

Retention Rate  

Track your retention rate to see the percentage of devices continuing to use your app after download. If retention is low, improve onboarding to boost understanding. Filter by app version to determine whether updates improve daily retention.  

  • Filter by source type or campaign to see how retention rates vary by acquisition channel or campaign.  

Monetization And Business Performance 

Track daily app and in-app purchase performance across Apple platforms. Monetization is tracked when transactions begin in the App Store, Apple Games app, or your app. iMessage apps and sticker packs are included with iOS.  

In-App Purchases 

The App Store provides a powerful commerce system to offer your digital goods, content, and services worldwide. Your overview page summarizes in-app purchase data, including revenue, paying users, and sales, based on the selected time frame across your app, the App Store, and the Games app. You can get additional insights by viewing or filtering this data. For example:  

  • Measure the growth of a newly launched in-app purchase by territory to understand where users have made the most purchases.  
  • Filter by name to evaluate the performance of specific in-app purchases.  

Sales 

View total sales for in-app purchases and bundles under sales metrics in USD, using last month’s rolling-average exchange rates, with refunds included. Sales are logged when transactions start, not when billing completes.  

The sales section provides total sales for your chosen time frame and the percent change from the previous period. A daily graph is included. Filter or view by different categories for details like…  

  • View by territory to see which areas generate the most sales over time. For instance, if a particular region shows higher sales, consider whether local marketing efforts, pricing strategies, or culturally relevant features are contributing to this trend. Click a new territory to view a sales summary and compare it with the previous period. You can also filter by territory and pick specific regions for more details.  

Proceeds Per Paying User 

A paying user is someone with a unique Apple account who has paid for your app, game, or in‑app purchase, and the proceeds per paying user reflect that. Select the ‘Proceeds ‘metric in the Metrics tab and compare it to the number of paying users. Look for changes in this data across several app versions to see whether the updates have affected user spending. You can also filter this data by different categories for more insight. For example:  

  • View proceeds per paying user by territory to identify regions with the highest spending. Target these territories for further growth.s.  
  • Filter by source type to find which sources attract the most paying users.  

Take Action on Insights from Peer Group Benchmarks 

Peer group benchmarks use data from all apps on the App Store to provide accurate, privacy-focused comparisons across categories, business models, and download volumes. You can see how your app performs compared to similar apps on key metrics, such as conversion rate, retention, crash rate, and average proceeds per paying user. These benchmarks are shown weekly and use the same definitions as app metrics in Analytics.  

How Peer Groups Are Created 

Peer groups use several attributes to match your app with relevant benchmarks, including:  

  • App Store category view benchmarks for each app store category you belong to, as long as there are enough apps within that peer group to ensure individual app performance is kept protected. Only apps in the same app store category are grouped within a peer group. So your app will not be included in a category you haven’t selected. You can view which app store categories your app is included in by reviewing your app information in App Store Connect.  
  • Apps are grouped by business model. In-app purchase apps are freemium or premium; paid apps are considered paid or premium. Subscription business models require freemium or premium apps with at least 50% revenue from subscriptions.  
  • App store download volume. Apps with lower weekly download volumes are grouped into low, medium, or high.  

Improve Your Performance 

If you want to improve a specific metric, you can use tools in App Store Connect, like product page optimization, custom product pages, and TestFlight. Here are some options to try:  

Conversion Rate 

Conversion rate is the percentage of people who download your app after seeing it in the app store. For example, if 100 people view your app and 25 download it, your conversion rate is 25%. A higher rate means you gain users more efficiently. To improve this, you might…   

  • Set up a product page optimization test. This helps you see which app icons, screenshots, and previews work best. For example, try changing your app icon’s style or color to see if it increases conversions—show or gameplay feature.  
  • Improve your localization. Make sure your product page and app experience are well localized so they adapt to a variety of cultures and languages. You might prioritize your localization efforts in regions where your app has proven popular.  

Monetization Performance 

Analytics gives you three monetization benchmarks to help you measure your business performance: conversion to paying users, average proceeds per paying user, and revenue by source or group. These data show where you are strong and where you can improve. If your results are below average, you might  

  • Identify the top groups and sources driving Day 35 proceeds, and prioritize them for marketing.  
  • Show the value of in-app purchases if your D35 download-to-paid rate is low and you offer subscriptions or consumable purchases. Consider offering a free trial or a free item during onboarding. This helps users see the value of paid items and may encourage them to purchase.  
  • If proceeds per paying user lag behind peers, adjust in-app purchase prices or test regional subscription pricing in app analytics.  

Crash Rate  

Crash rate is defined as the average number of crashes per session during a selected period. For example, if your app has 10 crashes and 100 sessions, your crash rate is 10%. A lower crash rate indicates better app stability. To improve this, you can  

  • Review crash and deletion data in analytics by platform, app version, and OS to find causes. Use Xcode crash logs for details.  
  • Use TestFlight to gather useful feedback and identify issues before launch. Invite up to 10,000 testers via email or public link.  

User Retention 

Retention rate is defined as the percentage of users who return to use your app over a specific period. It is calculated by dividing the number of devices that opened your app after downloading it by the total number of devices that downloaded it on the same day and used it in the last 30 days. For example, if 100 devices downloaded your app on May 1 and 20 are still active on May 8, your day 7 retention rate is 20%.  

  • Offer in-app events. Use in-app events to highlight new and current updates in your app or game. These events appear on the App Store and the Games app, helping new users find your app, keeping current users informed, and bringing back previous users.  
  • Review your onboarding to ensure it clearly shows users how to use your app and its value.  

Final Proceeds in Payments and Financial Reports 

You can view your total monthly proceeds, total units sold, and payment details by territory. Proceeds are paid in your bank account’s currency. To get this information as a monthly report, click the Create Reports button or automate it with the App Store Connect API or Reporter.  

You will see the total estimated proceeds at the top of the page until payment is sent to your bank. After the transfer, you will see the total proceeds, which shows the final amount paid for each region and reporting period. The amount you receive may change due to exchange rates and any withholding tax applied at the time of payment.  

Data Privacy 

At Apple, we build data privacy protections into each of our products and services, including analytics in App Store Connect. App Usage and App Clip Metrics only use data from users who have agreed to share their diagnostics and usage information to protect privacy. Some sources, such as app referrals, web referrals, and campaign links, require a minimum amount of data to appear in analytics. These protections apply throughout analytics, including subscription and cohort views. Peer group benchmarks use differential privacy, the gold standard for protecting individual data. Benchmarks are only derived from data from users who have agreed to share their analytics. When needed, Apple’s own apps may be included in benchmarks to give a more complete view of performance for a business category, business model, or download volume tier. 

Source: Measuring app performance 

Amazon Transforms Kindle With First Ever Color Scribe and AI Search 

Key Takeaways 

  • Amazon introduces the redesigned Kindle Stripe lineup to boost productivity.  
  • Kindle Scribe Colorsoft Soft delivers comfortable, colorful writing.  
  • Order both Kindle Scribe models now. A front light-free version arrives in 2026.  

We’re excited to introduce our new Kindle Scribe lineup, designed to boost your productivity. The new lineup includes the next-generation Kindle Scribe (available with or without a front light) and our first color model, Kindle Scribe Colorsoft.  

A New Paper-Like Design That’s Thinner, Lighter, and Faster 

The new Kindle Scribe features a sleek, paper-like design. It’s just 5.4 mm thick, weighs only 400 g, and is 40% faster for writing and turning pages. The large 11-inch glare-free display is the same size as a sheet of paper, making it great for reviewing documents, taking notes, and taking with you wherever you go.  

Our new lineup is packed with innovation:  

  • A new front light system with miniaturized LEDs that fit tightly against the display to create a narrower bezel and uniform lighting.  
  • The new textured molded glass gives the pen just the right amount of friction as it moves across the screen, unlike other tablets that can feel smooth or slippery.  
  • The redesigned display stack reduces parallax to almost zero, so it feels like you are writing right on the page.  
  • With a new quad-core chip, more memory, and our latest Oxide display technology, everything feels faster and more responsive.  

A Color Writing Experience That’s Easy on the Eyes 

Kindle Scribe ColorSoft has the same new design as the standard Kindle Scribe but lets you write in color smoothly. Its custom ColorSoft display technology uses a special color filter and light guide with nitride LEDs, delivering a soft, eye-friendly color display unlike the monochrome screen on the standard model. A new redesign engine makes writing in color fast, smooth, and natural.  

Kindle Scribe Colorsoft also offers weeks of battery life and has no distracting apps or notifications, so you can stay focused.  

All New Productivity Features, Including an AI-Powered Notebook 

Our new lineup includes a smart AI-powered notebook and updated software and tools to help you get more done.  

  • The all-new Home Quick Notes lets you quickly jot down notes whenever inspiration strikes. You can also access recently opened and added books. Humans and notebooks.  
  • You can easily import documents from Google Drive and Microsoft OneDrive for markup and export your annotated PDFs.  
  • The AI-powered search lets you browse your notes naturally and provides easy-to-understand summaries. You can also ask follow-up questions to learn more.  
  • You can send your notes and documents from Kindle Scribe to Alexa Plus and talk about them there (US only).  
  • Export your notes as text or images to OneNote so you can keep everything in one place and continue editing on your laptop.  
  • Color pen: You can write, draw, and annotate using different pen colors or highlights with five highlighter colors. Artists and creators can create smooth gradients and subtle tones with our new sharper tone, giving you even more control over the depth and richness of your art.  
  • You can organize your documents, notebooks, books, and more—all in one folder.  

Kindle Scribe combines a notebook and an e‑reader, giving you access to the world’s largest eBooks store. It also includes a three-month Kindle Unlimited subscription and the latest reading features.  

New AI-powered reading features preserve the magic of reading on Kindle. The story so far helps you catch up on your book without spoilers. For those who want to explore further, ask this book: you can highlight any passage and receive a spoiler-free answer, such as a character’s motive or the meaning of a scene. These features will be available for thousands of questions and Kindle books in the US and will be accessible on books purchased or borrowed via the Kindle iOS app and Kindle devices.  

Pricing And Availability 

All of our devices come with a new pen that feels amazing in your hands and seamlessly attaches to Kindle Scribe. Every device comes with a new pen that feels great to use, attaches securely to Kindle Scribe so you won’t lose it, and never needs charging. We are also launching new folio covers made from premium, plant-based leather, plus an exclusive executive notebook-style swap portfolio.  

Kindle Scribe Color Soft starts at $629.99 (US), £569.99 (UK), and €649.99 (Germany), shipping from April 8, 2026. The front light-free Kindle Scribe will also be available soon.  

Source: Amazon unveils redesigned Kindle Scribe lineup with first-ever color Scribe