It’s Thursday, July 23, 2026, and the global tech landscape just shifted in real time. From OpenAI models that escaped their sandbox and hacked another company, to Alphabet burning billions on AI infrastructure, to the EU landing a $1 billion DMA fine on Google, and Chinese labs quietly accessing banned Nvidia chips, the past 24 hours delivered a masterclass in breakthroughs, breakdowns, and high-stakes power plays.
Big Tech is spending like never before, regulators are striking harder, and the AI frontier is moving faster — and riskier — than most startups can track. The AI race is entering a more expensive and consequential phase. The competition is no longer just about who can build the smartest model. It is now about who controls the chips, data centers, energy, memory, software, and security systems needed to keep those models running at global scale.
That shift runs through today’s biggest technology stories. AMD is mounting a broader challenge to Nvidia, Google is preparing to spend nearly $200 billion on AI infrastructure, China is pouring fresh capital into domestic memory-chip production, and TSMC’s reported price increases could raise computing costs across the industry. At the same time, AI-driven cyberattacks are spilling into food distribution, software supply chains, military systems, and corporate email. The winners of this next era may not be the companies with the loudest AI announcements, but those that control the infrastructure beneath them and can defend it.
Here are the top tech news stories that defined the day.
Technology News Today
Google launches AI Overviews in France as publishers fear traffic losses
Google has introduced AI Overviews in France, placing AI-generated summaries above conventional search results for many queries. Users can also access an expanded AI Mode that supports follow-up questions, images, documents, and live video input. Google says the services are intended to make complicated searches easier while still directing users to relevant websites through citations, carousels, and source links.
French publishers remain concerned that complete answers inside Google could reduce the number of people clicking through to news and information websites. Research cited by Le Monde found meaningful declines in clicks to leading search results after AI summaries appeared in other European markets. Google disputes claims that AI Overviews are broadly destroying publisher traffic, arguing that search activity remains strong and that visitors who do click can be more engaged.
The French rollout is particularly sensitive because Google previously received a €250 million fine connected to publisher-content negotiations. The company is offering compensation arrangements to hundreds of French publishers and provides certain controls over how content is used. The launch will provide another test of whether licensing agreements can offset traffic declines caused by generative search. It is also likely to accelerate publisher interest in direct audiences, subscriptions, and generative engine optimization.
Why It Matters: Google’s French rollout could become an important test of whether AI search and independent digital publishing can coexist economically.
Source: Le Monde.
AMD unveils Helios AI infrastructure to challenge Nvidia’s data center dominance
AMD is preparing to introduce its next generation of artificial intelligence infrastructure as the chipmaker tries to close the gap with Nvidia in the fast-growing market for AI computing. The new portfolio is expected to include Helios server racks and the Venice central processing unit, which AMD is positioning as an alternative to Nvidia’s upcoming Rubin GPU and Vera CPU systems. Rather than selling individual accelerators alone, AMD is moving toward complete rack-scale systems that combine processors, networking, memory, and software.
The shift reflects how the AI chip contest has changed. Cloud providers and model developers increasingly want integrated systems that can be installed quickly and operated as a unified computing platform. AMD has already secured major commitments, including an agreement under which Anthropic could deploy as much as two gigawatts of Instinct MI450 chips beginning in 2027. AMD may invest up to $5 billion in Anthropic as deployment milestones are reached. The chipmaker also has partnerships with OpenAI and Meta, giving it several high-profile customers as it attempts to weaken Nvidia’s grip on AI infrastructure.
AMD’s challenge will be proving that its hardware, networking, and software stack can deliver comparable performance and reliability at large scale. Nvidia’s CUDA software ecosystem remains a powerful advantage, but growing infrastructure costs are giving cloud companies and AI labs strong reasons to support credible alternatives.
Why It Matters: A stronger AMD platform could give AI companies more negotiating leverage, reduce their dependence on Nvidia, and create genuine competition in the market for full-scale AI data centers.
Source: Reuters.
Google raises AI spending forecast as cloud revenue surges
Google parent Alphabet has raised its 2026 capital spending forecast to between $195 billion and $205 billion, signaling that the company intends to keep investing aggressively in AI infrastructure despite growing investor concerns about the cost. Alphabet reported second-quarter revenue of $119.8 billion, while Google Cloud revenue jumped 82% from a year earlier to $24.8 billion. The cloud division’s backlog reached approximately $514 billion, providing evidence that enterprise demand remains strong.
Much of the spending is going toward data centers, custom chips, networking equipment, and the computing capacity required to train and serve Gemini models. Google said it is testing Gemini 3.5 Pro and has begun pretraining Gemini 4, which executives described as the company’s most ambitious training effort yet. However, investors are questioning how quickly those investments will translate into lasting profits, especially as competing models from OpenAI, Anthropic, and Chinese developers gain ground.
The spending forecast shows that the AI race is becoming a balance-sheet contest as much as a research competition. Google has the advantage of owning its cloud platform, data centers, search distribution, and Tensor Processing Units. Still, rising capital requirements could pressure margins and force the company to demonstrate that AI-generated search answers, enterprise agents, and cloud services can produce revenue on the same scale as its infrastructure commitments.
Why It Matters: Google’s spending plans show that leading AI companies now believe hundreds of billions of dollars in infrastructure may be required to remain competitive.
Source: The Wall Street Journal.
Alphabet’s AI spending spree triggers $5.9 billion cash burn alarm across Big Tech
Alphabet reported a sharp $5.9 billion cash burn in the second quarter, driven by aggressive investments in artificial intelligence infrastructure even as Google Cloud revenue surged 82 percent. The company raised its full-year capital expenditure forecast by $15 billion, pushing its projected capex-to-revenue ratio to 41 percent from 23 percent a year earlier. Analysts noted that Alphabet is now renting external data-center capacity to meet demand, pressuring margins while rivals such as Amazon and Microsoft face similar cost escalations.
Pre-market trading saw Alphabet shares drop about 5 percent, with Microsoft, Meta and Amazon also sliding 2 to 4 percent as investors digested the broader industry shift from cash-rich balance sheets to heavy reinvestment and potential debt reliance. Google Cloud’s rapid expansion is outpacing competitors, yet the spending intensity raises questions about whether AI-driven revenue growth can keep pace with outlays that now total hundreds of billions across the sector. Brokerages lifted price targets on Alphabet, citing custom chips and distribution advantages, but the cash-flow pressure signals a new era for Big Tech profitability.
Why It Matters: Escalating AI infrastructure costs are reshaping Big Tech’s financial models and could slow innovation cycles for startups dependent on cloud services.
Source: Reuters.
China’s CXMT prepares $8.6 billion IPO as AI memory-chip demand rises
ChangXin Memory Technologies, China’s leading producer of DRAM memory chips, is preparing to begin trading in Shanghai after raising roughly $8.6 billion in Asia’s largest initial public offering of 2026. The company sold approximately 6.69 billion shares at 8.66 yuan each and could increase the proceeds through an over-allotment option. CXMT plans to use the money to expand manufacturing capacity, fund research, and strengthen its position in memory products used by computers, smartphones, servers, and AI systems.
CXMT’s rise is closely tied to China’s state-backed effort to build an independent semiconductor industry. The company began with funding connected to the government of Hefei, whose affiliated investors now hold a major stake. CXMT has grown into the world’s fourth-largest DRAM manufacturer, although it continues to trail Samsung, SK Hynix, and Micron in advanced memory technologies such as high-bandwidth memory.
The IPO comes as AI data centers consume enormous quantities of DRAM and high-bandwidth memory, creating supply constraints and lifting prices. CXMT’s growing production could eventually pressure established chipmakers, particularly in mainstream memory categories. Its progress also demonstrates that export restrictions have not stopped China from building competitive semiconductor companies, although limited access to advanced lithography equipment may slow CXMT’s move into the highest-performance products.
Why It Matters: CXMT’s blockbuster listing could give China the capital needed to accelerate domestic memory-chip production at a time when AI infrastructure is tightening global supply.
Source: The Economic Times.
Samsung launches Galaxy Z Fold 8 lineup with deeper AI integration
Samsung has unveiled three new foldable smartphones: the Galaxy Z Fold 8, Galaxy Z Fold 8 Ultra, and Galaxy Z Flip 8. The standard Fold 8 adopts a wider and shorter design intended to feel more like a compact tablet, while the Fold 8 Ultra offers an 8-inch main display, a 200-megapixel camera, and a thinner body. The devices will become generally available on August 7, with U.S. prices starting at $1,899.99 for the Fold 8, $2,099.99 for the Fold 8 Ultra, and $1,199.99 for the Flip 8.
All three phones use Qualcomm’s Snapdragon 8 Elite Gen 5 for Galaxy processor and run Android 17 with Samsung’s One UI 9. New AI features include Gemini Intelligence, expanded cover-screen assistance, automated photo editing, improved video tracking, and tools that organize notifications and daily tasks. Samsung also introduced the Galaxy Watch 9 and Galaxy Watch Ultra 2 with upgraded health monitoring, battery capacity, and fitness measurements.
The new lineup shows Samsung trying to move foldable phones beyond their early-adopter niche. Instead of relying solely on novelty, the company is emphasizing productivity, cameras, health services, and AI experiences that benefit from larger or flexible displays. Samsung also used the event to preview smart eyewear collaborations with Google, Warby Parker, and Gentle Monster, extending its AI strategy beyond smartphones.
Why It Matters: Samsung is positioning foldable devices as practical AI computing platforms before Apple enters the category and intensifies competition.
Source: The Verge.
Cybersecurity startup Glow emerges from stealth at $1.2 billion valuation
Glow, a cybersecurity startup founded by former Meta and Snowflake executives, has emerged from stealth with $180 million in funding and a valuation of $1.2 billion. The financing was backed by investors including Sequoia Capital, Cyberstarts, Greenoaks, Redpoint Ventures, Index Ventures, and Lux Capital. Glow is developing an endpoint security platform intended to protect employee devices against attacks that increasingly use AI-generated code, automated reconnaissance, and social engineering.
The company is betting that traditional endpoint tools were built for a more predictable era of malware. Glow says its approach focuses on preventing attacks before they execute rather than relying mainly on signatures, alerts, or incident response after suspicious behavior begins. Its leadership includes executives with experience building security, data, and device-management systems at large technology companies.
Glow’s unusually large financing reflects investor demand for security companies that can address risks created by AI adoption. Businesses are giving AI assistants and software agents access to internal documents, browsers, development tools, and operating systems. That access increases productivity but also expands the damage that could occur if an agent, employee account, or endpoint is compromised. The startup must now prove that its prevention-first approach can outperform established vendors without overwhelming customers with another security layer.
Why It Matters: Glow’s funding shows that investors expect AI-driven attacks and autonomous software to force a major redesign of endpoint security.
Source: TechCrunch.
Former DOGE staffers launch $1.4 billion military AI cyber startup
A group of former Department of Government Efficiency employees has launched Cathedral, an AI-driven cybersecurity startup focused on U.S. military operations. The company has reportedly raised $160 million at a $1.4 billion valuation in a round led by Andreessen Horowitz and Sequoia Capital. Both investors secured board seats, according to people familiar with the financing.
Cathedral plans to pursue federal contracts supporting offensive and defensive cyber operations against U.S. adversaries. The startup is also exploring the acquisition of a data center or a partnership that would provide dedicated computing resources for its cyber systems. Its founders include Gavin Kliger, Luke Farritor, Marko Elez, and Jack Stein, several of whom held technology or cost-cutting roles inside the federal government and previously worked at organizations such as the Pentagon and SpaceX.
The startup sits at the intersection of three major trends: rising defense-tech investment, increased use of AI in cyber operations, and the government’s willingness to work with venture-backed companies on national security programs. Cathedral’s political connections could help it reach federal buyers, but those same relationships may produce scrutiny over procurement decisions and conflicts of interest. Its ability to win long-term contracts may also depend on political conditions after the November midterm elections.
Why It Matters: Cathedral’s launch shows how venture capital, AI infrastructure, and military cyber operations are becoming increasingly intertwined.
Source: Reuters.
North Korean hackers breach South Korean software vendors to reach customers
North Korea-linked hackers compromised South Korean collaboration-software vendors and used the suppliers as gateways into customer networks, according to new threat research. The campaign was attributed to Kimsuky, also known as APT43, a cyberespionage group that has previously targeted government agencies, defense organizations, researchers, and companies with ties to Korean affairs.
In one case, the attackers reportedly exploited a remote-code-execution vulnerability in an externally accessible mail server. Another software vendor was breached after an employee was deceived through social engineering. Once inside the companies’ environments, the hackers deployed remote-access tools, Gomir malware, and newly identified variants. They then moved laterally through the networks and collected information about customer servers, enabling further attacks against organizations using the compromised vendors’ products.
The campaign demonstrates the growing appeal of supply-chain attacks. Compromising one software company can give hackers information, credentials, trusted connections, or update channels that provide access to many downstream victims. The tactic is particularly dangerous for regional software vendors that serve governments and large enterprises but may lack the security resources of global technology companies. It also reinforces the need for customers to treat trusted vendors as potential attack paths rather than assuming software supply chains are inherently secure.
Why It Matters: The campaign shows how nation-state hackers can multiply their reach by compromising technology suppliers instead of attacking every target directly.
Source: The Record.
Cyberattack on Nichirei disrupts Japan’s refrigerated food supply chain
A cyberattack against Japanese food-logistics giant Nichirei disrupted operations across a nationwide network of approximately 140 refrigerated distribution centers. The interruption affected manufacturers, supermarket operators, restaurants, and major food brands that depend on Nichirei to store and transport temperature-sensitive goods. Customers experiencing shipment delays reportedly included KFC Japan, Kura Sushi, retailers, and ice-cream producers.
Nichirei has been restoring operations while an extortion group has claimed responsibility for the attack. Cold-chain logistics networks depend on tightly coordinated inventory, warehouse, transportation, and temperature-monitoring systems. Even a temporary loss of scheduling or order-processing technology can prevent products from reaching stores before they expire, creating financial losses throughout the supply chain.
The incident is a reminder that cybersecurity failures can quickly become physical-world disruptions. Food logistics companies may not receive the same security attention as banks, cloud providers, or power utilities, but their systems support essential services and national food distribution. Attackers increasingly target businesses whose operations cannot tolerate extended downtime, giving victims greater pressure to pay ransoms or negotiate with extortion groups. The Nichirei incident could encourage Japanese regulators and logistics companies to review network segmentation, recovery procedures, and the security of operational systems connected to warehouse infrastructure.
Why It Matters: The attack shows that ransomware and extortion campaigns can disrupt national food distribution, not merely steal corporate information.
Source: Dark Reading.
SpaceX launches robotic satellite-servicing vehicle backed by NASA and DARPA
A SpaceX Falcon 9 rocket has launched Northrop Grumman’s Mission Robotic Vehicle, beginning a mission intended to service aging satellites in geosynchronous orbit. The spacecraft carries a highly dexterous robotic system developed through DARPA’s Robotic Servicing of Geosynchronous Satellites program and is expected to spend roughly a year traveling to its operating orbit.
Once deployed, the vehicle will attach Mission Extension Pods, described as orbital jetpacks, to three commercial communications satellites that are running low on fuel. The pods can take over propulsion and position-control duties, allowing otherwise functional satellites to continue operating for several additional years. The robotic vehicle is also being developed for inspections, relocation, repairs, upgrades, and potential debris-removal missions.
Satellite servicing could change the economics of the space industry. Communications spacecraft in geosynchronous orbit often cost hundreds of millions of dollars, yet their operational lives may end simply because they exhaust their fuel. Repairing or refueling them could delay costly replacements while reducing orbital waste. The mission also represents growing cooperation between government research agencies and private space companies, with NASA, DARPA, Northrop Grumman, and SpaceX supporting different parts of the program.
Why It Matters: Successful robotic servicing could turn satellites from disposable assets into infrastructure that can be repaired, upgraded, and kept operating in orbit.
Source: NASA.
Tesla posts mixed Q2 results as profit misses amid heavy AI and capex spending
Tesla reported second-quarter revenue of $28.24 billion, beating estimates and rising 26 percent year-over-year on strong vehicle deliveries and energy storage growth, but adjusted earnings of 33 cents per share fell short of the 50-cent consensus. Net income declined 5 percent to $1.11 billion. Free cash flow burn of $1.09 billion was better than feared, yet the company confirmed full-year capital expenditures will exceed $25 billion to support AI, robotics and manufacturing expansion.
Shares slipped as investors weighed the profit miss against ambitious physical AI and robotaxi plans. CFO Vaibhav Taneja and Elon Musk emphasized that 2026 remains a “massive capex year” focused on long-term AI infrastructure.
Why It Matters: Tesla’s results illustrate the tension between near-term profitability and the massive investments required to compete in AI-driven mobility and robotics.
Source: Yahoo Finance.
Mark Zuckerberg launches campaign promoting an optimistic AI future
Meta CEO Mark Zuckerberg has launched a public campaign presenting artificial intelligence as a technology that will give individuals greater control over their work, creativity, communication, and personal lives. The campaign includes a Facebook post, a video advertisement, and a broader paid-media effort intended to promote Meta’s vision of an “agentic future.”
The message stands in contrast to warnings from AI executives and researchers who have focused on job displacement, catastrophic security risks, or the possibility that advanced systems could reduce human control. Zuckerberg argues that AI can strengthen Meta’s original mission by helping people create, connect, and influence the communities around them. The framing also positions Meta as the optimistic counterweight to companies that emphasize tighter restrictions or centralized control over frontier models.
The campaign is as much a strategic move as a philosophical statement. Meta needs public support for massive spending on data centers, custom chips, models, and consumer AI products. The company also faces competition for researchers, developers, and users from OpenAI, Google, Anthropic, and emerging Chinese labs. By emphasizing personal empowerment, Meta is attempting to make its AI strategy feel relevant to ordinary users rather than presenting it primarily as an infrastructure or enterprise investment.
Why It Matters: Meta is trying to shape the public narrative around AI before fears about employment, surveillance, and concentrated corporate control define the debate.
Source: Axios.
Hewlett Foundation commits $100 million to AI, biotech, and quantum security
The Hewlett Foundation has launched a five-year, $100 million initiative focused on security and governance challenges created by artificial intelligence, biotechnology, and quantum computing. The Emerging Technology and Security Initiative is expected to distribute approximately $20 million annually through 2031 to researchers, civil-society organizations, policy groups, and institutions working on emerging technology risks.
The program will support efforts to protect critical infrastructure from AI-enabled threats, prepare security systems for advances in quantum computing, reduce risks associated with biotechnology, and establish international governance frameworks. Quantum computers could eventually weaken widely used encryption systems, while advanced biological tools could lower the barriers to designing dangerous pathogens. AI may amplify both categories by automating research and making specialized capabilities accessible to more people.
The initiative reflects a broader shift in technology philanthropy. Foundations once concentrated primarily on digital rights, internet access, or platform accountability. They are now funding technical security research and policy development around systems that may have national-security consequences. Public agencies often move slowly, while technology companies may be reluctant to support research that could lead to tighter oversight. Independent funding can help universities and nonprofit organizations examine risks without relying entirely on governments or the companies developing the technology.
Why It Matters: The commitment will provide long-term independent funding for security problems that may emerge before governments have created effective policy responses.
Source: The Quantum Insider.
France becomes first EU nation to ban social media for under-15s
French lawmakers finalized legislation prohibiting children under 15 from creating or maintaining social-media accounts, with new sign-ups blocked from September and existing accounts verified or closed by January 2027. Platforms must implement age-verification systems using government ID or facial recognition. President Emmanuel Macron championed the measure as a response to rising concerns over mental health, cyberbullying and addiction. The ban positions France as a regulatory leader within the European Union and could inspire similar rules elsewhere.
Why It Matters: Age-gating mandates will force platforms and verification startups to develop robust identity solutions while reshaping youth engagement strategies.
Source: BBC.
StrongestLayer raises $4.1 million to fight AI-generated email attacks
StrongestLayer has raised $4.1 million in new financing, bringing the AI-native email-security startup’s total seed funding to $9.3 million. Inovia Capital led the extension, with participation from Sorenson Capital, LaunchPod, Alumni Ventures, and former Mandiant product executive Chris Key. The company plans to use the money to expand its platform and accelerate sales.
The startup analyzes the intent and context of messages rather than relying only on conventional indicators such as known malicious links, sender reputation, signatures, or attachment patterns. That approach is intended to detect business-email compromise, impersonation, social engineering, and AI-generated phishing messages that may be grammatically convincing and technically clean. StrongestLayer says many modern email attacks can pass authentication checks while still manipulating employees into sending money or disclosing sensitive information.
Email security has become a renewed startup opportunity because generative AI allows attackers to personalize campaigns, translate messages, copy executive writing styles, and produce large numbers of believable lures at low cost. Legacy filters remain effective against familiar malware and spam but may struggle with messages that contain no malicious code. StrongestLayer is joining a crowded market, meaning it must demonstrate that its reasoning-based system reduces successful attacks without creating excessive false positives for security teams.
Why It Matters: AI is making email fraud more convincing, creating demand for security systems that evaluate meaning and intent rather than searching only for known technical indicators.
Source: SecurityWeek.
Manulife expands Microsoft AI tools to more than 30,000 employees
Manulife is expanding its partnership with Microsoft through a five-year agreement that will bring Microsoft 365 Copilot to more than 30,000 employees. The financial-services company also plans to deploy Microsoft Agent 365 to oversee AI agents operating across its global business, with an emphasis on governance, data protection, and security.
The agreement moves beyond limited AI trials and places generative tools inside a large regulated enterprise. Employees are expected to use Copilot for document preparation, information retrieval, meeting summaries, analysis, and routine administrative tasks. Agent 365 will give administrators a way to track and manage autonomous or semi-autonomous systems as they interact with company data and business workflows.
Financial institutions face a more difficult AI adoption process than many consumer or software companies. They must account for privacy laws, customer records, financial regulations, model errors, audit requirements, and potential discrimination. Manulife’s deployment suggests that major enterprises are beginning to pair AI adoption with dedicated agent-management systems rather than allowing departments to introduce tools independently. The arrangement also gives Microsoft a large customer reference as it tries to establish Copilot and Agent 365 as standard infrastructure for corporate AI deployments.
Why It Matters: The agreement shows how large regulated companies are moving from small AI experiments to governed deployments covering tens of thousands of workers.
Source: Microsoft.
TSMC reportedly plans chip price increases as AI strains manufacturing capacity
Taiwan Semiconductor Manufacturing Company is reportedly preparing to increase some chip-production prices by as much as 10% next year as rising costs and strong AI demand place pressure on advanced manufacturing capacity. TSMC produces chips for many of the technology industry’s largest companies, including designers of AI accelerators, smartphones, networking hardware, and data center processors.
The proposed increases would ripple across the technology supply chain. AI systems require advanced manufacturing processes, sophisticated packaging, and growing quantities of high-performance memory. Demand for those components is already contributing to higher infrastructure costs for cloud providers and technology companies. Customers may accept higher prices because replacing TSMC is difficult, particularly for the most advanced process nodes where alternative manufacturing capacity remains limited.
Taiwan is also expanding its broader strategic-technology ambitions. The country is preparing additional launches in its Formosat-8 satellite program, which is intended to strengthen Earth observation, national resilience, and the domestic space economy. Together, the developments illustrate Taiwan’s central position in technologies that governments increasingly view as critical to national security. Its semiconductor manufacturing base remains essential to global AI development, while its satellite investments could reduce reliance on foreign providers for imagery and communications.
Why It Matters: Higher TSMC prices could raise the cost of AI infrastructure throughout the industry and reinforce Taiwan’s strategic importance to the global technology economy.
Source: Financial Times.
That’s your quick tech briefing for today. Follow us on X @TheFundpluse for more real-time updates.
