It’s Friday, July 10, 2026, and the AI frontier just got a serious jolt. In the past 24 hours, OpenAI dropped its most advanced model family yet after U.S. government clearance, Meta green-lit production of its own custom AI chip, and SK Hynix made history with one of the largest foreign listings ever on Nasdaq. At the same time, China flagged security risks in foreign AI tools, Europe took aim at addictive platforms, and public skepticism toward AI is quietly rising. From hyperscale data centers in Canada to power grid strains in Taiwan, the race for compute, control, and sovereignty is accelerating.
The technology race is no longer being shaped by a single breakthrough or company. Meta, OpenAI, Microsoft, ByteDance, and a growing field of global challengers are now competing across AI models, chips, agents, cloud infrastructure, robotics, and regulation—all at once. From billion-dollar infrastructure bets to new rules that could reshape how AI reaches consumers, today’s biggest stories reveal where the next phase of technology is heading and which companies are positioning themselves to lead it.
Here are the top tech news stories shaping the global tech and startup landscape right now
Technology News Today
Microsoft Begins Rolling Out In-House MAI Models in Excel and Outlook
Microsoft is deploying more of its proprietary MAI family of models into Excel and Outlook, reducing previous heavy reliance on OpenAI and Anthropic models. The shift aims to give the company greater control over performance, cost, and integration in productivity tools. lemondeinformatique.fr
This move reflects Big Tech’s broader strategy of building in-house AI capabilities alongside partnerships. It could improve customization for enterprise users and influence pricing negotiations with external model providers.
Why It Matters: Microsoft’s increasing use of proprietary models in core Office products signals maturing in-house AI capabilities and potential shifts in the AI model provider landscape.
Source: Yahoo News.
OpenAI Launches GPT-5.6 Model Family with Tiered Pricing and Global Access
OpenAI officially rolled out its GPT-5.6 AI model family on July 9, introducing the flagship Sol, the balanced Terra, and the budget-friendly Luna after completing a U.S. government review. The models feature improved agentic capabilities in coding, biology, and cybersecurity. Pricing undercuts rivals: Sol at $5/$30 per million tokens, Terra at $2.50/$15, and Luna at $1/$6. Sol matches or exceeds top competitors on key benchmarks while offering strong safety features. Access rolled out across ChatGPT, API, and Codex, with tiered reasoning efforts for different users.
The launch marks a major step in commoditizing advanced AI through differentiated tiers, intensifying competition with Anthropic and others. It accelerates developer adoption and enterprise integration while highlighting ongoing U.S. scrutiny of frontier models. Big Tech ecosystems, including Microsoft’s Copilot, stand to benefit from deeper integration and cost efficiencies.
Why It Matters: This tiered release accelerates AI commoditization, lowers barriers for developers and enterprises, and intensifies the global race for capable, affordable models.
Source: Fundpluse via Reuters.
Tencent Moves to Become the Largest Shareholder in AI Agent Startup Manus
Tencent is in talks to become the largest shareholder in Manus, the AI agent startup that drew global attention for software capable of completing complex, multi-step tasks with limited human supervision. The proposed transaction would give one of China’s largest technology companies a significant position in a business competing in the fast-growing market for autonomous AI agents.
Manus emerged as a prominent Chinese challenger to agent platforms from OpenAI, Anthropic, Google, and Microsoft. Unlike conventional chatbots that mainly respond to prompts, AI agents are built to plan work, operate software, collect information, and execute tasks across several steps. Tencent’s interest suggests China’s leading internet companies increasingly view these systems as strategic infrastructure for cloud computing, workplace software, gaming, advertising, and online commerce.
The investment could also give Manus access to Tencent’s cloud infrastructure, developer network, consumer platforms, and distribution channels. For Tencent, the agreement would offer a faster path to advanced agent technology rather than relying entirely on internally developed products.
The talks underscore a broader consolidation pattern in which cash-rich technology platforms are securing stakes in promising AI startups before independent companies can develop into major competitors.
Why It Matters: Tencent’s pursuit of Manus shows that autonomous AI agents are becoming strategic assets for China’s largest technology companies.
Source: Reuters.
Taiwan’s Nanya Plans $6 Billion Spending Push as AI Drives Memory Demand
Taiwanese memory-chip producer Nanya Technology plans to spend about $6 billion in 2027 as demand for memory used in AI servers continues to reshape the semiconductor market. The planned investment represents a major expansion for a company historically associated with conventional dynamic random-access memory rather than the premium high-bandwidth memory chips now dominating AI infrastructure.
AI servers require vast quantities of memory to keep processors supplied with data. That demand has tightened availability across the broader memory market, lifting prices and encouraging manufacturers to reopen expansion plans that would have appeared risky during previous semiconductor downturns. Nanya’s spending decision suggests the AI boom is creating opportunities beyond Nvidia, SK Hynix, Micron, and Samsung.
The investment also strengthens Taiwan’s position in another part of the AI hardware supply chain. The island already plays a central role in processor manufacturing, chip packaging, testing, server assembly, and networking equipment. Additional memory capacity would deepen that position even as governments encourage semiconductor companies to spread production across the United States, Japan, Europe, and Southeast Asia.
The central risk is timing. Memory markets have historically moved through sharp cycles, and heavy investment can lead to excess supply if projected AI demand fails to materialize.
Why It Matters: Nanya’s spending plan shows how demand for AI infrastructure is driving more semiconductor companies into a new global capacity race.
Source: Reuters.
SK Hynix Raises $26.5 Billion in the Largest US IPO by a Foreign Company
SK Hynix completed a $26.5 billion US stock offering, setting a record for the largest American initial public offering by a foreign company. The South Korean memory-chip manufacturer priced its American depositary shares at $149 and prepared to trade on Nasdaq under the ticker SKHY.
The listing gives US investors direct access to one of the most important suppliers in the AI infrastructure market. SK Hynix has built a leading position in high-bandwidth memory, or HBM, which sits alongside advanced GPUs and accelerators inside AI servers. Nvidia and other chip developers depend on this memory to move data quickly enough to keep increasingly expensive processors operating efficiently.
The offering also provides SK Hynix with additional capital as it builds factories, purchases chipmaking equipment, and competes with Samsung and Micron for the next generations of HBM. Its rise reflects how the AI boom has transformed memory from a cyclical commodity business into a strategically important layer of the computing stack.
The IPO will now test whether public-market enthusiasm can keep pace with the enormous spending required to support AI data centers. Investors are betting on sustained demand, but memory pricing and technology transitions remain volatile.
Why It Matters: The record offering confirms that AI memory has become one of the most valuable and capital-intensive segments of the global chip market.
Source: Financial Times.
Off-Grid Renewable Energy Emerges as a New Model for AI Data Centers
A new infrastructure thesis is gaining attention: instead of waiting years for connections to congested electricity grids, AI data centers could be built alongside dedicated renewable-energy systems in deserts and other energy-rich locations. Envision founder Zhang Lei is among the executives advocating for data centers powered by off-grid wind, solar, storage, and related energy systems.
The proposal addresses one of the largest constraints facing AI development. Technology companies can buy processors and build server data centers faster than utilities can add generation, transmission lines, substations, and grid connections. In several major data-center markets, power availability has become a more serious bottleneck than access to land or capital.
Locating computing facilities near newly built energy could reduce interconnection delays and prevent some infrastructure costs from being passed to existing electricity customers. It could also encourage companies to schedule flexible AI workloads around periods of strong renewable generation.
The model is not straightforward. Remote data centers require fiber connectivity, cooling systems, equipment maintenance, energy storage, and backup generation. Training workloads can tolerate some scheduling flexibility, but consumer services and real-time inference require continuous availability.
Still, off-grid computing could become an important option as the AI industry searches for electricity outside traditional hyperscale markets.
Why It Matters: The AI infrastructure race may increasingly be determined by which companies can build dedicated energy systems rather than simply secure more chips.
Source: Financial Times.
EU Accuses Meta of Failing to Address Addictive Design on Facebook and Instagram
European Union regulators have issued preliminary findings indicating that Meta may have violated the Digital Services Act by failing to assess and mitigate mental-health risks associated with Facebook and Instagram. The European Commission pointed to features such as infinite scrolling, autoplay, and highly personalized recommendations that can encourage compulsive platform use.
The investigation moves Europe’s technology enforcement beyond familiar disputes over advertising, competition, and data privacy. Regulators are now examining the underlying mechanics used to keep people engaged, including recommendation algorithms and interface decisions that influence how long users remain on a platform.
Meta will have an opportunity to respond before the Commission reaches a final decision. A confirmed violation could lead to required product changes and substantial financial penalties. The company may argue that its services include controls for users and parents and that engagement features are not inherently harmful.
The case could affect the wider consumer-technology industry. TikTok, YouTube, Snapchat, gaming platforms, and other services use similar recommendation and engagement systems. Any regulatory standard developed through the Meta investigation could therefore serve as a reference point for assessing whether platforms have adequately addressed behavioral and mental health risks.
Why It Matters: Europe is beginning to regulate the engagement systems at the heart of social-media business models, not merely the content they distribute.
Source: The Guardian.
US Senator Proposes Broad AI Accountability Package Covering Data Centers and Hiring
US Senator Ed Markey has introduced an AI accountability agenda containing proposals aimed at data centers, workplace surveillance, automated hiring, algorithmic discrimination, healthcare decisions, and emotionally manipulative chatbots.
One proposal would require certain data-center operators to obtain certification demonstrating that their projects do not impose unacceptable costs on electricity customers, water supplies, air quality, or local ecosystems. Other measures would restrict employers from making important employment decisions solely through automated systems and would require stronger testing for discrimination.
The package reflects a change in the Washington debate. Federal AI policy previously centered on model safety, national security, and competition with China. Lawmakers are now placing greater emphasis on the effects AI infrastructure and automated decision-making have on workers, families, consumers, and local communities.
The bills face an uncertain path in Congress, where lawmakers remain divided over whether federal rules should limit state regulation or set a national minimum standard. Even without immediate passage, the proposals could shape state legislation and future agency enforcement.
For AI companies, the agenda signals that accountability obligations may eventually extend beyond model developers to employers, data center owners, software vendors, and organizations that deploy automated decisions.
Why It Matters: AI regulation is expanding from abstract model risks to the physical, economic, and workplace consequences of deploying the technology.
Source: The Guardian.
European Parliament Extends Voluntary Scanning of Private Messages
The European Parliament has approved an extension allowing technology platforms to continue voluntarily scanning certain private communications for suspected child sexual abuse material. The temporary authorization had expired in April and will now remain available until 2028 or until lawmakers agree on a permanent framework.
The decision is politically contentious because more lawmakers reportedly voted against the extension than supported it. Opponents nevertheless failed to reach the absolute majority required to block the measure. Privacy advocates argue that allowing platforms to inspect private messages risks creating surveillance infrastructure that could eventually be expanded to other forms of content.
Supporters say the authorization gives online services a legal basis to detect and report abusive material while European institutions negotiate longer-term legislation. End-to-end encrypted services such as Signal and WhatsApp are treated differently under the temporary arrangement, although encryption remains central to the wider policy debate.
The dispute illustrates the difficulty of reconciling child protection, platform responsibility, cybersecurity, and private communication. Technology companies must comply with reporting obligations while avoiding systems that weaken encryption or expose sensitive conversations.
Europe’s eventual permanent law could influence messaging services globally, as platforms frequently adopt common technical systems across multiple markets.
Why It Matters: Europe’s message-scanning debate could determine how far governments can push online-safety enforcement without weakening digital privacy.
Source: WIRED.
China Recovers First Stage of Long March Rocket in Major Reusability Test
China successfully recovered the first stage of a Long March-10B rocket after launch, marking an important step in its effort to develop reusable launch vehicles. The booster separated from the upper stage before returning to a platform at sea, according to Chinese state media.
Reusable rockets can substantially reduce launch costs by enabling companies and space agencies to reuse expensive engines and booster structures multiple times. SpaceX demonstrated the commercial importance of this approach through Falcon 9, forcing launch providers in China, Europe, the United States, and elsewhere to accelerate their own reusable-rocket programs.
The Long March-10 family is also connected to China’s broader human-spaceflight ambitions, including future lunar missions and expanded operations in low Earth orbit. Demonstrating controlled recovery does not, by itself, establish operational reuse: engineers must inspect the stage, assess the damage, refurbish it, and prove that it can launch safely again.
The test nevertheless shows China narrowing an important technology gap. Reusable rockets could help the country deploy satellite constellations more economically, increase launch frequency, support commercial space startups, and strengthen independent access to orbit.
Success would add further competitive pressure to an industry already undergoing rapid expansion.
Why It Matters: A reusable Long March system could lower China’s launch costs and intensify competition with SpaceX and other commercial space providers.
Source: Associated Press.
Asian Technology Stocks Rise as Investors Return to the AI Trade
Technology-related shares helped lift Asian markets Friday following a tech-led rally in the United States. The move came as investors assessed the record SK Hynix listing, strong demand for AI memory and networking equipment, and signs that capital remains willing to support companies tied to data-center construction.
The gains show that the AI investment cycle has become a global market force rather than a narrow US software trend. Semiconductor manufacturers in South Korea and Taiwan, electronics suppliers in Japan, server manufacturers, optical networking companies, and energy providers are increasingly trading as part of the same AI infrastructure ecosystem.
That concentration also creates risk. Technology indexes can rise even when most companies in the broader market decline, leaving market performance dependent on a relatively small group of AI-linked businesses. Any reduction in capital spending by hyperscalers could therefore affect suppliers across several countries.
Investors are also balancing the AI rally against energy price volatility and geopolitical tensions, which can raise manufacturing, transportation, and data center operating costs.
For startups, continued strength in public AI stocks can improve private-market fundraising conditions and acquisition prospects. It can also make investors less tolerant of businesses that use AI branding without clear revenue, technical differentiation, or exposure to infrastructure.
Why It Matters: The AI boom is increasingly connecting stock markets across the United States and Asia through a shared semiconductor and infrastructure supply chain.
Source: Associated Press.
ByteDance and Alibaba Restrict Humanlike AI Agents Ahead of New China Rules
ByteDance and Alibaba are disabling or modifying customizable, humanlike AI agent features as new Chinese regulations governing anthropomorphic AI services are set to take effect on July 15. ByteDance’s Doubao and Alibaba’s Qwen are among the platforms affected by the approaching rules.
China’s regulations target AI systems designed to simulate human personalities, relationships, and emotional interaction. Authorities are concerned that highly persuasive or emotionally responsive services could manipulate vulnerable users, create unhealthy dependency, impersonate real people, or distribute content that violates national rules.
The response demonstrates how regulation can alter AI products before enforcement formally begins. Instead of risking noncompliance, major platforms are removing features, limiting customization, and reviewing how agents present themselves to users.
China remains committed to building a competitive domestic AI sector, but its government is drawing firmer boundaries around consumer-facing behavior. That approach differs from the US model, where federal regulation remains fragmented, and from the European Union’s risk-based AI framework.
The changes could also influence developers outside China. Companies offering AI companions, customer-service characters, gaming personalities, or virtual influencers may eventually need separate product configurations for different countries.
The emerging regulatory category is no longer just artificial intelligence. It is an artificial personality.
Why It Matters: China’s rules could establish an early regulatory template for AI companions and agents that imitate human identity and emotion.
Source: South China Morning Post.
Korean Researchers Develop Hierarchical AI Agent ‘ReAcTree’ for Complex Tasks
Researchers in South Korea unveiled ReAcTree, a hierarchical AI technology that breaks down complex, sequential errands into subgoals for autonomous planning. The system significantly reduces errors and improves success rates in long-term missions for robots and agents.
The advancement addresses a key limitation in current AI agents — reliably handling extended, multi-step workflows. It has strong potential applications in robotics, automation, and enterprise task orchestration.
Why It Matters: Hierarchical agent architectures like ReAcTree represent important progress toward more reliable, long-horizon AI systems capable of real-world deployment.
Source: TechXplore.
PLAUD Achieves $100 Million ARR with AI Hardware-Subscription Model
PLAUD has reached $100 million in annual recurring revenue by combining AI hardware with subscription services, capitalizing on advances in generative AI and agentic tools for voice, meetings, and conversational workflows.
The milestone demonstrates viable hardware-plus-AI business models beyond pure software. It shows how specialized devices can capture value in the growing agentic and productivity AI segment.
Why It Matters: Hardware-AI hybrid models achieving significant scale validate new monetization paths for startups in the voice and productivity AI space.
Source: Digitimes.
US AI Regulation Shifts Toward Voluntary Government Review of Frontier Models
The United States is developing an alternative approach to AI regulation centered on cooperation between frontier-model developers and government agencies. Recent releases from OpenAI and Anthropic have involved government review or testing before wider deployment, even though Congress has not enacted a comprehensive national AI safety law.
Supporters argue that voluntary review can respond faster than traditional legislation and give officials early visibility into cyber, biological, and national-security capabilities. Companies also have strong incentives to cooperate because government approval can reduce political risk and reassure enterprise customers.
Critics warn that a voluntary system depends heavily on trust and may lack consistent standards, enforcement mechanisms, technical staffing, and transparency. Government agencies must evaluate systems developed by companies that employ many of the country’s leading AI researchers and possess far greater computing resources.
The arrangement could also favor the largest AI laboratories. Smaller developers may struggle to participate in extensive evaluation processes or interpret informal government expectations, while established companies can maintain dedicated policy and security teams.
The immediate result is a hybrid system: advanced models remain commercially developed, but their release increasingly resembles a regulated national-security event rather than a routine software update.
Why It Matters: Frontier AI governance in the United States is being built through government-industry negotiation before Congress has agreed on permanent rules.
Source: Axios.
European Startup Funding Reaches Strongest Quarter in Four Years
European startups recorded their strongest venture-funding quarter in four years during the second quarter of 2026, raising approximately $24 billion, according to Crunchbase data. The improvement was supported by larger technology rounds, resilient merger activity, and stronger fundraising in the United Kingdom.
The figures suggest that Europe’s venture market is recovering from the sharp slowdown that followed the low-interest-rate funding surge. Investors remain selective, but capital is returning to businesses in AI infrastructure, defense technology, climate and energy systems, fintech, robotics, and enterprise software.
Europe still faces structural challenges. Growth-stage companies frequently turn to US or Middle Eastern investors for larger rounds, while fragmented capital markets make it harder to finance companies from startup through public listing. Many of Europe’s strongest technology businesses eventually list in New York or sell to foreign buyers.
Governments are responding with sovereign investment vehicles, defense funds, AI infrastructure programs, and proposals aimed at deepening European capital markets. The stronger quarter provides evidence that those efforts are underway alongside renewed private-sector interest.
The quality of the recovery will depend on whether funding reaches a broad base of companies or remains concentrated in a small number of unusually large AI and defense transactions.
Why It Matters: Europe’s funding rebound could help more regional startups scale independently rather than relocate or sell to larger foreign technology companies.
Source: Crunchbase News.
Post-Quantum Cybersecurity Startup QIZ Security Raises $17 Million
QIZ Security has secured $17 million to help critical-infrastructure operators prepare for cybersecurity threats posed by future quantum computers. The company is developing tools intended to identify vulnerable cryptographic systems and support migration to algorithms that can withstand attacks from sufficiently advanced quantum machines.
Today’s quantum computers are not capable of breaking widely used public-key encryption at a meaningful scale. The concern is that attackers can collect encrypted information now and store it until more capable systems become available, a strategy commonly described as “harvest now, decrypt later.”
That risk is particularly serious for governments, utilities, telecommunications providers, financial institutions, healthcare organizations, and industrial operators whose sensitive information may remain valuable for decades. Migrating their cryptographic infrastructure can take years because encryption is embedded across applications, devices, identity systems, certificates, network equipment, and supplier relationships.
Funding for QIZ reflects growing demand for practical post-quantum transition products rather than purely theoretical research. Organizations need inventories of where vulnerable algorithms are used, plans for replacing them, and systems that can accommodate future cryptographic changes.
The opportunity is likely to grow as governments set deadlines and procurement requirements aligned with post-quantum standards.
Why It Matters: Post-quantum security is turning into an enterprise migration market long before cryptographically relevant quantum computers arrive.
Source: Industrial Cyber.
US AI Cybersecurity Clearinghouse Plan Begins to Take Shape
The Trump administration’s plan for an AI cybersecurity clearinghouse is moving closer to implementation as agencies consider how the government can collect, analyze, and distribute information about AI-enabled cyber threats.
A centralized clearinghouse could help federal agencies, critical infrastructure operators, security researchers, and AI companies exchange information on model vulnerabilities, malicious use, autonomous attack tools, defensive discoveries, and emerging techniques. The initiative follows a presidential order directing agencies to strengthen cyber defenses and coordinate responses to threats created or accelerated by advanced AI systems.
The practical challenge will be deciding what information companies must provide and who can access it. AI laboratories may possess sensitive details about model capabilities, internal testing, and previously undisclosed vulnerabilities. Security agencies may hold classified intelligence that cannot be broadly shared. Companies will also want protection from liability and public disclosure when voluntarily reporting weaknesses.
A well-designed clearinghouse could reduce duplicated work and shorten the time between discovering an AI-related threat and deploying defenses. A poorly defined program could become another reporting obligation with limited operational value.
The initiative reflects a broader recognition that AI security cannot be handled by model companies, software vendors, or government agencies working independently.
Why It Matters: Coordinated threat sharing could become essential as AI systems enable cyberattacks to be faster, cheaper, and easier to scale.
Source: Inside Cybersecurity.
NASA Upgrades Quantum Laboratory Aboard the International Space Station
NASA’s Jet Propulsion Laboratory has highlighted an upgrade to a quantum research facility operating aboard the International Space Station. The work involves cooling experimental equipment to extremely low temperatures, allowing researchers to study quantum behavior in microgravity for longer periods and under conditions that are difficult to reproduce on Earth.
Microgravity enables clouds of ultracold atoms to remain suspended rather than fall quickly under Earth’s gravity. Scientists can therefore observe them for longer durations, increasing the precision of experiments involving quantum states, atomic interactions, and matter-wave behavior.
The research could contribute to the development of highly sensitive quantum sensors capable of measuring gravity, acceleration, rotation, and magnetic fields. Future space-based instruments may help map underground water, monitor geological changes, improve navigation without GPS, test fundamental physics, and detect subtle variations in Earth’s gravitational field.
The upgrade does not mean a general-purpose quantum computer is operating aboard the station. The laboratory is focused on scientific experiments and sensing technologies rather than commercial quantum computing.
Still, the project shows how space is becoming an important testing environment for quantum science. As governments and startups pursue quantum navigation, communications, clocks, and sensors, orbital laboratories may provide capabilities that terrestrial facilities cannot match.
Why It Matters: Microgravity quantum experiments could lead to a new generation of space-based sensors, navigation systems, and scientific instruments.
Source: NASA Jet Propulsion Laboratory.
That’s your quick tech briefing for today. Follow us on X @TheFundpluse for more real-time updates.
