It’s Tuesday, July 28, 2026, and the tech world just shifted under our feet. In the past 24 hours, a Chinese lab dropped the largest open-weight AI model ever released, Microsoft fielded specialized cyber agents to fight AI-powered attacks, Nvidia poured billions into a secretive superintelligence lab, and hyperscalers locked in multi-billion-dollar data-center and fiber deals while regulators, ransomware groups, and satellite operators all made moves that will reshape the next decade.
The AI race is no longer just about who builds the smartest model. It is becoming a global contest over who can finance the data centers, secure the electricity, control the chip supply, own the fiber networks, and defend the systems connecting it all.
That shift is visible across today’s headlines. AMD is locking up as much as 2.5 gigawatts of AI data center capacity, Nvidia is reportedly backing a $50 billion infrastructure lease tied to its own chips, and Amazon is preparing to spend an estimated $200 billion on AI and related infrastructure this year. At the same time, investors are questioning whether the returns can justify the spending, China is pushing deeper into memory chips and domestic lithography, and a major Australian utility is warning that data from roughly 900,000 customers may have been exposed.
Here are the top tech news stories that matter most right now, from AI and startups to cybersecurity, space, robotics, semiconductors, and Big Tech
Technology News Today
Microsoft Introduces AI Security Tools to Automate Exposure Detection
Microsoft has unveiled a new collection of AI-based security capabilities intended to help organizations identify weaknesses, evaluate potential attack paths, and reduce their exposure before hackers can exploit it. The tools are designed to analyze large volumes of security data continuously, prioritize the vulnerabilities that pose the greatest practical risk, and recommend or automate defensive actions.
The announcement comes as corporate security teams struggle with an imbalance between the number of alerts they receive and the people available to investigate them. AI agents could reduce that burden by connecting information across identities, devices, applications, cloud systems, and network configurations. However, automated defense systems will need strong oversight because inaccurate prioritization or poorly executed remediation could disrupt legitimate business operations. Microsoft’s advantage is its access to security signals across Windows, Azure, Microsoft 365, identity services, and enterprise software. That reach gives the company a broad base of data for training and operating defensive AI systems, while also increasing concerns about customers becoming dependent on a single vendor for both infrastructure and security.
Why It Matters: AI security agents could help understaffed teams respond faster, but they will also deepen the role of major cloud providers in deciding how enterprise threats are detected and contained.
Source: Ars Technica.
Verizon Lands $1 Billion Google Fiber Deal as Telecoms Chase AI Data Center Revenue
Verizon is positioning its network infrastructure as a central part of the AI data center boom through a reported $1 billion dark-fiber agreement with Google. Dark fiber refers to installed optical cable that a customer can operate using its own networking equipment, giving hyperscalers greater control over capacity, security, and performance. Verizon expects the arrangement to be the first of several large contracts tied to AI facilities.
The agreement highlights an often-overlooked bottleneck in AI deployment. Advanced processors cannot operate effectively at scale without high-capacity connections linking data centers, cloud regions, power-intensive computing clusters, and end users. As AI models grow and workloads become more distributed, demand for private fiber networks is increasing alongside demand for GPUs and electricity. Verizon is also exploring ways to repurpose selected facilities into smaller data centers, potentially creating new revenue from real estate and network assets that were built for earlier generations of telecommunications. Google, for its part, gains dedicated connectivity for an infrastructure program that requires fast movement of enormous datasets between computing locations.
Why It Matters: Fiber capacity is emerging as one of the critical constraints on AI expansion, creating a new growth opportunity for telecommunications companies.
Source: Ars Technica.
Nvidia Backs $50 Billion Texas AI Data Center Lease Using Its Own Chips
Nvidia is reportedly supporting a roughly $50 billion lease tied to a large Texas data center that will operate with Nvidia processors. The arrangement represents another example of CEO Jensen Huang using the company’s balance sheet and financial strength to help AI infrastructure projects obtain the capital required to move forward. The strategy can stimulate demand for Nvidia hardware while reducing financing barriers for developers facing extraordinary construction and equipment costs.
The deal also draws attention to increasingly interconnected AI financing structures. Chip suppliers, cloud providers, model developers, infrastructure companies, and lenders are entering agreements in which the same companies may act as investors, customers, suppliers, and financial guarantors. These arrangements can accelerate construction, but they also make it harder to separate organic customer demand from demand supported by vendor financing. Investors are becoming more sensitive to that distinction as projected AI capital expenditures move into the hundreds of billions of dollars. For Nvidia, the Texas agreement could preserve its dominant position as AMD and other chipmakers assemble their own infrastructure partnerships. It also demonstrates that access to credit is becoming another competitive advantage in the AI hardware market.
Why It Matters: Nvidia is using financing, rather than chips alone, to shape where AI computing capacity is built and whose hardware powers it.
Source: Financial Times.
Moonshot AI Releases Open Weights for 2.8-Trillion-Parameter Kimi K3 Model
Chinese AI lab Moonshot AI has made the full weights of its Kimi K3 model publicly available for download, modification, and independent hosting under a modified MIT-style license. The 2.8-trillion-parameter mixture-of-experts system activates only about 104 billion parameters per token, features a native 1-million-token context window, and includes multimodal vision capabilities. Moonshot claims architectural advances such as Kimi Delta Attention deliver roughly 2.5 times better intelligence per unit of compute compared with prior versions, positioning the model near frontier closed systems on coding and agentic benchmarks while running at a fraction of the cost.
The release expands access for developers and enterprises seeking high-capability open models without relying solely on proprietary APIs. It intensifies competition in the global AI race, particularly as Chinese open-weight systems challenge U.S. closed-model economics and raise questions about infrastructure demands for self-hosting such large systems. Startups and researchers gain tools for fine-tuning and specialized applications, potentially accelerating innovation outside the largest labs.
Why It Matters: Open-weight frontier-scale models lower barriers for startups and shift power dynamics in the AI ecosystem toward greater accessibility and cost efficiency.
Source: Fundpluse via Moonshot AI.
AMD Secures Up to 2.5 Gigawatts of AI Data Center Capacity From Core Scientific
AMD has signed a major infrastructure agreement with Core Scientific that could give the chipmaker access to as much as 2.5 gigawatts of data center capacity. The first phase is expected to provide 500 megawatts of AI-ready capacity beginning in 2027, with additional deployments planned if customer demand supports the expansion. The companies will also work together on facility design, hardware deployment, and the software needed to run AMD’s AI accelerators at scale.
The agreement shows how the AI chip competition is moving beyond processor specifications. AMD must also help customers secure electricity, land, cooling, networking, and buildings if it wants to challenge Nvidia in large AI deployments. Core Scientific, meanwhile, is accelerating its shift from cryptocurrency mining toward high-performance computing and AI hosting, where long-term contracts can generate more predictable revenue. AMD will receive market-priced warrants to purchase Core Scientific shares, giving the chipmaker a financial interest in the infrastructure provider’s growth. Core Scientific’s shares rose in premarket trading following the announcement, while AMD shares declined amid a broader retreat in semiconductor stocks.
Why It Matters: AI infrastructure has become a full-stack contest in which chipmakers must secure data center capacity as aggressively as they design processors.
Source: Reuters.
Global AI Chip Selloff Deepens as Investors Question Data Center Returns
Semiconductor stocks came under fresh pressure Tuesday as concerns about the scale, financing, and eventual returns of AI infrastructure spending spread through global markets. The decline followed losses in South Korean technology shares and extended a broader retreat among companies tied to AI processors, memory, networking equipment, and chipmaking machinery. China’s booming semiconductor market also contributed to the shift in sentiment by highlighting the possibility of stronger competition and additional manufacturing capacity.
The selloff does not necessarily indicate that demand for AI computing has weakened. Data center developers are still ordering chips and securing vast amounts of electricity. Instead, investors appear to be questioning how quickly enormous capital investments will translate into profitable AI services. The distinction matters because technology companies are committing to multiyear projects that require heavy borrowing, long leases, and infrastructure that may take years to become fully operational. A reduction in expected returns could affect valuations across the entire supply chain, from Nvidia and memory-chip manufacturers to power providers and construction companies. The market reaction also shows that public investors are beginning to scrutinize the financial structure of AI expansion rather than automatically rewarding every infrastructure announcement.
Why It Matters: The AI investment debate is shifting from whether demand exists to whether the industry can earn adequate returns on unprecedented infrastructure spending.
Source: Financial Times.
Origin Energy Warns Data From About 900,000 Customers May Have Been Exposed
Australian electricity and gas provider Origin Energy said information belonging to approximately 900,000 current and former customers may have been accessed during a cybersecurity incident. The potentially exposed records include personal information and limited financial details, such as partial bank account or payment-card numbers. Origin has begun notifying affected customers and is offering identity-protection and cybersecurity assistance while forensic specialists investigate the breach.
The incident is significant because energy providers hold extensive identity, billing, address, and payment data while operating services that households cannot easily abandon. That combination makes utilities attractive targets for attackers seeking information for identity theft, phishing, account takeovers, or financial fraud. Origin first investigated a potential threat earlier in July but reportedly reassessed the situation after receiving additional information on July 22. Its shares declined following the disclosure. The breach adds to mounting pressure on companies to investigate warnings quickly and communicate clearly when customer information may be at risk. It also illustrates how cyber incidents affecting critical-service providers can create consequences extending far beyond temporary website outages or internal IT disruption.
Why It Matters: A breach involving a major utility can expose customers to long-term fraud risks while testing public confidence in the security of essential infrastructure providers.
Source: Reuters.
China’s CXMT Debut Shakes Global Memory-Chip Stocks
Shares of major global memory-chip companies fell after Chinese DRAM manufacturer ChangXin Memory Technologies, or CXMT, surged 466% during its Shanghai market debut. The rally briefly made CXMT one of China’s most valuable listed companies and renewed investor concerns about the country’s ability to build semiconductor competitors despite restrictions on advanced Western equipment.
CXMT remains behind the largest international manufacturers in process technology and does not have access to the most advanced extreme-ultraviolet lithography systems. Even so, its public-market valuation gives the company another channel for raising capital, recruiting engineers, and increasing production. That matters because memory demand is being reshaped by AI. Manufacturers are prioritizing high-bandwidth memory used in accelerators, contributing to tighter supplies and higher prices in other portions of the market. A well-funded Chinese supplier could increase output in mainstream DRAM while gradually moving toward more advanced products. The market reaction affected companies including Micron, SK Hynix, and SanDisk, showing that investors increasingly view China’s semiconductor expansion as a commercial threat rather than a distant policy ambition.
Why It Matters: CXMT’s market debut gives China’s semiconductor strategy additional capital and could reshape pricing and competition across the global memory industry.
Source: MarketWatch.
China Begins Producing Domestic DUV Chipmaking Equipment in Challenge to ASML
A state-backed Chinese company has reportedly begun manufacturing immersion deep-ultraviolet lithography systems, marking a potentially important step in Beijing’s effort to build a domestic semiconductor supply chain. DUV machines project circuit patterns onto silicon wafers and can manufacture advanced chips through repeated patterning, although the process is generally less efficient than production using extreme-ultraviolet equipment.
The development does not mean China has matched ASML, the Dutch company that dominates advanced lithography. Manufacturing reliable chipmaking tools at scale requires precision optics, light sources, software, chemicals, and thousands of tightly integrated components. But a functioning domestic DUV system could reduce reliance on imported equipment for a broad range of semiconductors and provide a foundation for further technical improvements. It could also help Chinese foundries replace or supplement foreign machines that are becoming harder to service because of export restrictions. The broader significance lies in China’s movement across the full semiconductor stack—from AI models and processors to memory chips and the equipment used to manufacture them.
Why It Matters: Domestic lithography production could weaken one of the most important pressure points in Western semiconductor export controls.
Source: Fundpluse via The Information.
Amazon’s $200 Billion AI Infrastructure Plan Pushes It to Top of Global 500
Amazon has taken the top position in Fortune’s latest Global 500 ranking as the company prepares to spend an estimated $200 billion on AI and related infrastructure during 2026. The planned expenditure places Amazon at the front of a Big Tech capital-spending race that is expected to exceed $700 billion across the largest hyperscalers this year.
Much of Amazon’s investment is directed toward AWS data centers, custom AI processors, networking equipment, and electricity supply. The company is trying to protect its cloud leadership while supporting increasingly demanding AI workloads and reducing its dependence on third-party chip suppliers. Amazon’s scale gives it the ability to spread infrastructure costs across cloud services, e-commerce, advertising, subscriptions, and logistics. Yet the strategy also raises the financial stakes. AWS must attract enough model developers, enterprises, and AI applications to justify years of heavy construction and depreciation expenses. Amazon is competing against Microsoft’s OpenAI relationship and Google’s combination of proprietary models, TPUs, and cloud infrastructure.
Why It Matters: Amazon’s spending shows that the AI race is becoming a test of financial endurance, operational scale, and the ability to turn infrastructure into durable cloud revenue.
Source: Fortune.
India’s Space Fundpluse Raise $113 Million as Private Sector Gains Ground
Indian space technology startups have raised approximately $113 million in equity funding so far in 2026, bringing total investment in the sector to about $871 million since 2021, according to data from Tracxn. The figures reflect growing private-sector participation in a space industry historically dominated by the Indian Space Research Organization.
Capital is moving into satellite manufacturing, launch services, propulsion, Earth observation, communications, and software that analyzes data collected from orbit. India offers founders access to experienced aerospace engineers, relatively low development costs, a large domestic market, and an increasingly supportive policy environment. The government has opened more activities to private operators while encouraging companies to commercialize technologies developed around the national space program. Funding remains smaller than the capital flowing into U.S. space companies, but the direction is strategically important. Indian startups could become lower-cost suppliers to global satellite and launch markets while serving domestic needs in agriculture, weather forecasting, defense, navigation, and telecommunications.
Why It Matters: India’s expanding private space sector could create a lower-cost global competitor across satellite, launch, and Earth-observation markets.
Source: The Times of India.
Britain’s Jodrell Bank Observatory Faces Closure After Science Funding Cuts
Jodrell Bank Observatory, one of the world’s best-known radio astronomy centers, is facing an uncertain future after UK Research and Innovation decided to end funding for the e-MERLIN telescope network in March 2028. The decision could directly affect 28 jobs and disrupt work by thousands of researchers who rely on the network’s high-resolution observations.
The observatory hosts the Lovell Telescope and plays an important role in studying black holes, stellar formation, pulsars, and distant galaxies. E-MERLIN links multiple radio telescopes across Britain, effectively allowing them to operate as a single instrument with much greater resolving capability. Scientists have warned that shutting down or reducing the network would weaken Britain’s position in international astronomy and waste decades of investment in scientific infrastructure and expertise. The funding decision comes as the UK reduces its participation in several major research projects because of budget pressures and rising operating costs. Those cuts could affect universities, scientific startups, instrument manufacturers, and the pipeline of researchers entering advanced fields.
Why It Matters: Cutting established scientific infrastructure can weaken the research base that later produces commercial breakthroughs in computing, communications, imaging, and space technology.
Source: The Guardian.
AI Startup COR Secures $30 Million to Build Profitability Software for Professional Services
COR has secured a $30 million investment from FTV Capital to expand its AI-based project profitability platform for agencies and professional-services firms. The company’s software combines project management, staffing, time tracking, budgeting, and financial analysis to help service businesses understand which clients and projects generate sustainable margins.
Professional-services companies frequently struggle to measure profitability because employee time, changing project scopes, unbilled work, and uneven resource allocation are spread across several systems. COR is betting that AI agents can bring those records together, identify problems earlier, and automate administrative tasks that reduce billable capacity. The new capital will support product development and the addition of more autonomous capabilities, according to the company. The investment also reflects a wider shift in enterprise AI funding. Investors are increasingly backing industry-specific platforms that attach AI to measurable business outcomes rather than general-purpose assistants. COR’s challenge will be proving that its recommendations improve margins consistently and that customers can trust automated decisions affecting staffing, pricing, and client relationships.
Why It Matters: Enterprise AI investment is moving toward applications that can demonstrate direct financial returns inside specific industries.
Source: Business Wire.
Cybersecurity Startup Frenos Raises $1.52 Million for AI-Powered OT Penetration Testing
Frenos has closed a $1.52 million funding round to develop an AI-based simulated penetration-testing platform for operational technology environments. OT systems control industrial equipment used in manufacturing, energy, transportation, logistics, and other physical operations, making security failures potentially more disruptive than breaches confined to conventional corporate networks.
Traditional penetration testing can be expensive, time-consuming, and difficult to conduct safely inside facilities where an interruption could halt production or damage equipment. Frenos is developing software intended to model attacks and identify exploitable paths without exposing live industrial systems to unnecessary risk. The company’s focus reflects increased demand for security products capable of protecting the convergence of information technology, connected sensors, industrial control systems, and AI-driven automation. However, automated simulations will need to account for legacy equipment, proprietary protocols, and plant-specific configurations that are often poorly documented. The modest funding round gives Frenos resources to validate its approach, but industrial customers will require strong evidence that its testing is both accurate and operationally safe.
Why It Matters: As factories and infrastructure become more connected, AI-based security testing could help identify cyber risks before they create physical disruption.
Source: Business Wire.
AI Security Startups Attract $855 Million as Investors Target the Next Cybersecurity Platform
Startups working at the intersection of artificial intelligence and cybersecurity have raised approximately $855 million across more than 150 reported seed-stage rounds in 2026, according to Crunchbase data. The activity suggests investors are searching for companies that can secure AI models, generated code, autonomous agents, and the infrastructure on which those systems operate.
The category includes startups monitoring model behavior, protecting AI applications from prompt injection, evaluating generated software, managing access to agents, and using AI to automate threat detection. The large number of early-stage rounds indicates that the market remains fragmented, with no consensus yet on which product category will become the defining AI security layer. Many companies will also face competition from Microsoft, Google, Palo Alto Networks, CrowdStrike, and other established security vendors adding similar capabilities. Still, major technology transitions have historically created new cybersecurity leaders. Cloud computing helped establish companies such as Wiz, while endpoint security helped CrowdStrike grow into a major platform. Investors are betting that AI will produce another opening of similar scale.
Why It Matters: The next major cybersecurity company may emerge from tools built specifically to secure AI agents, models, and machine-generated software.
Source: Crunchbase News.
Agility Robotics Expands Silicon Valley AI Hub as Humanoid Robot Commercialization Accelerates
Agility Robotics has opened a 60,000-square-foot engineering facility in Fremont, California, as the company works to increase development and deployment of its Digit humanoid robot. The site is expected to support engineering, testing, customer demonstrations, and training close to Silicon Valley’s robotics talent, investors, and technology suppliers.
Fremont has become an important center for physical AI because it combines large industrial properties with proximity to software and hardware engineering expertise. Tesla, Zoox, Pony.ai, DoorDash, and other companies already operate in the area, creating a concentration of autonomous-vehicle and robotics activity. Agility’s expansion comes as humanoid-robot developers move from controlled demonstrations to warehouse and manufacturing deployments where reliability, safety, maintenance costs, and measurable productivity matter more than visual novelty. A market update distributed Tuesday also referenced Agility Robotics’ proposed public listing, although investors will need to evaluate any transaction using formal regulatory disclosures rather than promotional announcements. The company must show that Digit can perform repetitive tasks consistently and economically before humanoid robots become a mainstream industrial category.
Why It Matters: Agility’s new hub shows the humanoid robotics race shifting from prototypes and fundraising toward engineering, manufacturing, and real customer deployments.
Source: GlobeNewswire.
That’s your quick tech briefing for today. Follow us on X @TheFundpluse for more real-time updates.
