It’s Wednesday, June 24, 2026. Today’s tech stories show an industry moving past the demo phase and into the hard infrastructure race, where the winners will control not just smarter models, but the systems that power, secure, deploy, and regulate them.
Today crystallized the next phase of the AI buildout in real time. OpenAI revealed its first custom inference chip, Micron delivered record-shattering results on the back of insatiable memory demand, SK Hynix moved to raise nearly $30 billion to scale production, and Qualcomm struck a nearly $4 billion deal to strengthen its AI software layer. At the same time, Anthropic accused a major Chinese rival of large-scale model extraction, while fresh cybersecurity alerts and supply-chain incidents underscored the mounting complexity of securing the systems powering this transformation.
From AI and startups to regulation and Big Tech, here are the top tech news stories that actually matter right now.
Technology News Today
OpenAI and Broadcom Unveil First Custom AI Inference Chip “Jalapeño”
OpenAI and Broadcom on Wednesday unveiled the design of Jalapeño, OpenAI’s first custom AI chip developed in partnership with the semiconductor giant. The chip is optimized specifically for inference workloads—running trained models to generate responses in systems like ChatGPT. OpenAI has received initial samples and is testing them, with deployment planned by the end of the year. The design targets high efficiency in power and performance for large-language-model inference, drawing on Broadcom’s expertise in custom silicon.
OpenAI engineers collaborated closely with Broadcom, leveraging the company’s AI tools in the design process itself. This move marks a significant step in OpenAI’s strategy to build more of its own AI infrastructure stack, reducing long-term dependence on third-party chip suppliers amid surging demand for compute. It aligns with similar efforts by other AI leaders to create tailored hardware that better matches their specific model architectures and workloads.
Why It Matters: Custom AI chips like Jalapeño accelerate the shift toward specialized hardware ecosystems, lowering costs and improving efficiency in AI deployment while intensifying competition in the semiconductor space that is critical to startups and hyperscalers alike.
Source: Bloomberg.
Qualcomm Buys AI Startup Modular in $4 Billion Push to Challenge Nvidia’s Software Moat
Qualcomm said it will acquire AI chip software startup Modular in an all-stock deal valued at nearly $4 billion, giving the mobile chip giant a deeper software stack for running AI models across different processors. Modular’s technology is designed to make AI workloads portable across chips without requiring developers to rewrite code for every processor family.
The deal matters because Nvidia’s dominance in AI has never been just about chips. CUDA, its developer software platform, has locked millions of developers into Nvidia hardware. Qualcomm’s Modular bet is a direct move against that advantage as the company looks beyond smartphones into data centers, cars, edge devices, and AI infrastructure.
Why It Matters: Qualcomm is trying to turn AI software portability into a weapon against Nvidia’s lock-in.
Source: Reuters.
EU Targets Amazon and Microsoft Cloud Units Under Big Tech Gatekeeper Rules
European regulators are moving to bring Amazon Web Services and Microsoft Azure under tougher Big Tech “gatekeeper” rules, widening the EU’s focus from app stores and social platforms to the cloud infrastructure powering the AI boom. The move reflects growing concern that cloud giants can use scale, pricing, data access, and bundled services to shape competition across the digital economy.
For startups, this is more than a Brussels regulatory story. AI companies depend heavily on cloud platforms for training, inference, storage, and enterprise distribution. If EU rules force more openness, portability, or fairer access terms, it could affect cloud pricing, AI model deployment, and the bargaining power of smaller companies building on top of Amazon and Microsoft.
Why It Matters: Europe is now treating cloud infrastructure as a core layer of competition in the AI economy.
Source: Reuters.
SK Hynix Plans Massive $29 Billion Nasdaq ADR Listing to Fund AI Expansion
South Korea’s SK Hynix, the world’s second-largest memory chipmaker and a key Nvidia supplier for HBM, announced on June 24 plans to raise up to $29.4 billion through a secondary listing of American Depositary Receipts on Nasdaq. The company will issue up to 17.79 million new shares, with trading expected to begin in July. Proceeds are earmarked for expanding production capacity to meet surging demand for AI chips, including new facilities in South Korea and the US.
This would rank among the largest equity offerings globally in recent years and reflects intense investor appetite for AI-exposed semiconductor plays. It also diversifies SK Hynix’s investor base beyond Asia while providing capital for the capital-intensive memory business.
Why It Matters: The listing highlights how demand for AI is reshaping global capital flows toward chipmakers, enabling faster scaling of critical memory infrastructure that underpins AI data centers worldwide.
Source: Reuters.
Former Infosys CEO’s AI Startup Hang Ten Raises $32 Million to Rebuild Enterprise Software Services
Hang Ten Systems, founded by former Infosys CEO Vishal Sikka, raised $32 million in seed funding led by Mayfield, with strategic backing from Aramco Ventures and participation from prominent angel investors. The startup says it helps enterprises continuously build, modify, and operate software using AI-driven development and automation.
The round is notable because it targets one of the largest and most vulnerable segments of global tech: IT services. For decades, enterprises have relied on armies of consultants, engineers, and outsourcing firms to maintain software systems. Hang Ten’s thesis is that AI agents and automation can compress that work into a more continuous, software-driven operating model.
Why It Matters: AI is moving from coding assistants into the heart of enterprise IT services, where billions of dollars in legacy spending are up for grabs.
Source: TechCrunch.
Anthropic Veterans Launch Mirendil With $200 Million to Help Scientists Build Their Own AI Models
Mirendil, a new AI startup founded by former Google and Anthropic researchers Behnam Neyshabur and Harsh Mehta, raised $200 million in seed funding at a reported $1 billion valuation. Investors include Andreessen Horowitz, Kleiner Perkins, and Nvidia. The startup wants to give scientists tools to build and improve their own AI models rather than rely entirely on closed frontier labs.
The company is positioning itself around scientific AI and recursive model improvement, with potential applications in medicine, materials science, and AI research itself. Its launch also signals a growing backlash against restrictions imposed by major AI labs that prevent customers from using their models to develop competing systems.
Why It Matters: Mirendil reflects a growing push to decentralize AI research tools beyond the largest closed labs.
Source: The Wall Street Journal.
Anthropic Accuses Alibaba of Large-Scale Illicit Extraction of Claude AI Capabilities
Anthropic on Wednesday publicly detailed accusations against Alibaba, alleging that operators linked to the Chinese company’s Qwen AI lab conducted the largest known “distillation” campaign against its Claude models. The effort reportedly used thousands of fraudulent accounts and involved nearly 29 million interactions to extract advanced capabilities in software engineering and agentic reasoning. Anthropic sent a letter to US senators and officials describing the activity as brazen and illicit, aimed at training weaker models on outputs from Claude.
The incident highlights growing geopolitical tensions around AI model security and intellectual property protection, as Chinese labs seek to close capability gaps through sophisticated extraction techniques rather than independent development. It comes amid broader US efforts to restrict advanced AI technology flows.
Why It Matters: Such attacks underscore the national security and competitive stakes in frontier AI, prompting stricter access controls, monitoring, and policy responses that could reshape global AI collaboration and startup strategies for model protection.
Source: Bloomberg.
Agility Robotics to Go Public in $2.5 Billion Humanoid Robot SPAC Deal
Agility Robotics, maker of the warehouse robot Digit, announced plans to go public through a SPAC merger valuing the company at $2.5 billion. The deal would make Agility one of the first pure-play public-market companies in humanoid robotics. The company’s investors include Amazon, Nvidia, SoftBank, and Foxconn.
Digit is already being tested or deployed in logistics and manufacturing environments, including warehouse tasks such as moving totes and handling repetitive material movement. The timing is important: robotics is shifting from demos to commercial deployments as labor shortages, reshoring, and AI advances push companies to automate physical work.
Why It Matters: Agility’s public-market move could become a test case for investor appetite in real-world robotics beyond software-only AI.
Source: Associated Press.
Amazon Commits Another $13 Billion to India Cloud and AI Infrastructure
Amazon said it will invest an additional $13 billion in India, expanding its cloud, AI, and digital infrastructure footprint in one of the world’s most important technology markets. The investment comes as India pushes deeper into sovereign AI, digital public infrastructure, and enterprise cloud adoption.
The move strengthens AWS’s position in a market where Microsoft, Google, domestic cloud providers, and telecom-backed infrastructure players are all competing for AI workloads. India’s large developer base, fast-growing startup ecosystem, and public-sector digitization make it a critical battleground for cloud and AI platforms.
Why It Matters: Amazon is doubling down on India as AI infrastructure becomes a geopolitical and commercial race.
Source: Reuters.
Raimondo-Backed RAISE US Launches $500 Million AI Workforce Push
RAISE US, a new bipartisan nonprofit backed by major technology and corporate players, launched with more than $500 million to help workers adapt as AI changes the labor market. The group is starting with state-level education and retraining programs in Arkansas, Connecticut, Maryland, and Utah.
The initiative arrives as AI adoption accelerates inside customer support, software development, marketing, finance, logistics, and administrative work. The political significance is clear: AI labor disruption is becoming a mainstream policy issue, not just a Silicon Valley concern. Companies want to show they are addressing workforce risk before public backlash hardens into regulation.
Why It Matters: The AI jobs debate is shifting from theory to large-scale worker retraining programs.
Source: Associated Press.
Ornn Raises $33 Million to Build an AI Compute Marketplace
AI startup Ornn raised $33 million in seed funding to build a marketplace where companies can buy and sell AI compute capacity more like a commodity. The round was backed by Andreessen Horowitz’s crypto-focused fund and other investors.
The idea speaks directly to one of the biggest bottlenecks in AI: access to GPUs and affordable compute. As startups struggle to secure capacity from hyperscalers and chip supply remains tight, marketplaces for unused or contracted compute could become a new infrastructure layer. The challenge will be reliability, pricing transparency, security, and whether enterprises trust third-party compute pools for sensitive AI workloads.
Why It Matters: Compute is becoming the oil of the AI economy, and startups are racing to financialize access to it.
Source: SiliconANGLE.
Agentic Infrastructure Startup Seltz Raises $12.5 Million to Help AI Agents Search the Web
Seltz raised $12.5 million to build infrastructure that helps AI agents search the web and retrieve answers more effectively. The startup is targeting a key weakness in agentic systems: AI agents often need fresh, structured, verifiable information, but the open web is messy, inconsistent, and increasingly hostile to automated retrieval.
As enterprises experiment with AI agents, search and retrieval quality will matter as much as model intelligence. Agents that cannot reliably find, verify, and act on information will remain limited to narrow workflows. Seltz is betting that web search for agents becomes a distinct infrastructure category, separate from consumer search and traditional enterprise search.
Why It Matters: AI agents need better information pipelines before they can handle serious enterprise work.
Source: SiliconANGLE.
Open-Source Software Security Crisis Deepens as AI Raises the Stakes
CyberScoop reported that governments are struggling to address security issues in open-source software, even as AI increases both the speed of software development and the risk of exploitation. The report points to a fragmented ecosystem in which widely used open-source components are maintained by small teams, volunteers, or underfunded organizations.
That matters because nearly every modern software company depends on open-source code. AI coding tools may help developers move faster, but they can also spread insecure patterns, introduce hidden dependencies, and accelerate attacks against vulnerable packages. Government efforts such as software bills of materials and secure-by-design guidance are useful, but enforcement and funding remain hard problems.
Why It Matters: Open-source security is becoming a national infrastructure issue as AI accelerates software production and exploitation.
Source: CyberScoop.
2026 FIFA World Cup Faces Surge in Cyber Threats Across North America
Cybercriminals are targeting the 2026 FIFA World Cup with phishing, fraud, DDoS attacks, ticket scams, and other cyber threats across host cities in the United States, Canada, and Mexico. The tournament’s scale makes it a high-value target, with millions of fans, sponsors, broadcasters, vendors, travel platforms, and payment systems in play.
The risk extends beyond fans losing money to scams. Major global events create temporary digital ecosystems that are difficult to secure: ticketing portals, hotel bookings, stadium networks, transportation systems, and media platforms all become attack surfaces. Nation-state actors may also exploit the event for espionage, disruption, or influence operations.
Why It Matters: Mega-events are now cybersecurity stress tests for cities, platforms, and payment infrastructure.
Source: Dark Reading.
Cordyceps CI/CD Flaws Expose 300+ GitHub Repositories to Supply-Chain Attacks
Security researchers flagged a new class of CI/CD workflow weaknesses dubbed Cordyceps that could expose more than 300 GitHub repositories to supply-chain attacks. The issue centers on how attackers can hijack automation workflows that developers use to build, test, and deploy software.
This is exactly the kind of vulnerability that keeps enterprise security teams up at night. Modern software moves through automated pipelines, and if attackers compromise those pipelines, they may inject malicious code before it reaches production. The broader lesson is that software supply-chain security is no longer just about scanning dependencies. It now includes workflow permissions, secrets management, developer identity, and build system integrity.
Why It Matters: Attackers are increasingly targeting the machinery that ships software, not just the software itself.
Source: The Hacker News.
Cisco SD-WAN Flaw Was Exploited Before Public Disclosure
Attackers reportedly exploited a Cisco Catalyst SD-WAN Manager flaw months before public disclosure, gaining root-level access in at least one communications provider environment. The vulnerability, tracked as CVE-2026-20245, was later added to CISA’s known exploited vulnerabilities catalog, with federal agencies ordered to patch or stop using affected systems by the deadline.
The incident is a reminder that networking infrastructure remains one of the most valuable targets in cybersecurity. SD-WAN tools are closely tied to enterprise traffic, remote access, branch connectivity, and cloud routing. A compromise at that layer can provide attackers with deep visibility and persistence, especially within telecom and communications environments.
Why It Matters: Exploited network management flaws can give attackers high-leverage access across enterprise and telecom systems.
Source: Dark Reading.
SPRIND Launches Quantum Sensing Initiative to Back European Projects and Startups
Germany’s federal innovation agency, SPRIND, launched a quantum sensing initiative to support European projects and startups in a field that could affect defense, navigation, medical imaging, energy, and industrial measurement. The program is designed to support commercialization, not just academic research.
Quantum sensing is less flashy than quantum computing, but it may reach practical markets sooner. Sensors based on quantum properties can detect tiny changes in gravity, magnetic fields, time, or motion. That could help with underground mapping, submarine detection, precision navigation without GPS, and advanced diagnostics. Europe’s push reflects a broader strategy to avoid falling behind the U.S. and China in frontier tech infrastructure.
Why It Matters: Quantum sensing may become one of the first quantum technologies to move from labs into real-world markets.
Source: The Quantum Insider.
China’s Future Industries Investment Boom Raises AI, Space, and Quantum Valuation Concerns
China is seeing a surge of investment into “future industries” such as AI, commercial space, quantum computing, nuclear fusion, and advanced manufacturing. The capital wave is tied to Beijing’s push for technological self-reliance and global leadership in strategic sectors.
But the rush is also raising concerns about overheated valuations, especially for startups with limited revenue and long commercialization timelines. The pattern is familiar: when governments, local funds, and private investors all chase strategic technologies at once, capital can move faster than technical progress. Still, China’s ability to coordinate industrial policy, manufacturing capacity, and domestic capital means even speculative bubbles can leave behind lasting infrastructure and talent.
Why It Matters: China’s deep-tech funding boom could fuel breakthroughs, but it also risks inflating valuations before markets are ready.
Source: Economic Times.
