It’s Thursday, July 16, 2026, and the tech world isn’t slowing down. While governments, executives, and researchers continue debating the future of AI, the technology is already forcing hard choices today: record-breaking chip demand is colliding with energy constraints, open-weight models are challenging closed AI labs, regulators are rewriting the rules, and billion-dollar investments are reshaping the global race for computing power.
That shift is reflected across today’s biggest stories. TSMC posted another blockbuster quarter as AI chip demand continued to soar. China is pushing to expand its influence over global AI governance. Microsoft is sharpening its AI strategy, Google is opening Android to rival app stores, and startups across healthcare, cybersecurity, and software development continue attracting significant investment as the next wave of AI infrastructure takes shape.
Here are the top technology news stories defining the day, from AI infrastructure and semiconductor breakthroughs to cybersecurity, regulation, Big Tech, and the startups shaping what comes next.
Technology News Today
Microsoft reportedly tells sales teams to challenge OpenAI and Anthropic more directly
Microsoft executives have reportedly instructed sales employees to position the company’s internally developed AI models as more efficient and economical than products from OpenAI, Anthropic, and Google. During an internal strategy meeting, executives compared Microsoft Copilot with rival systems and emphasized integrations with Microsoft’s security, productivity, and enterprise software.
One presentation reportedly argued that Anthropic’s Claude performed less effectively inside Microsoft applications and lacked some of the security connections available through Copilot. The claims reflect Microsoft’s competitive positioning and have not been independently verified across all workloads. Microsoft has not publicly released detailed evidence to support all the comparisons.
The change is significant because Microsoft helped finance OpenAI’s rise and built many of its earliest AI products around OpenAI models. That relationship has become less exclusive as OpenAI has gained additional infrastructure partners and Microsoft has invested more heavily in its own model development.
Microsoft is also under pressure to lower the cost of operating Copilot across Word, Excel, Teams, and other services. Using in-house models for more tasks could improve margins and give the company greater control over product development. It could also turn a once-close partner into a more direct competitor.
Why It Matters: Microsoft’s sales strategy shows that the alliance-driven phase of generative AI is giving way to direct competition among model providers.
Source: Bloomberg.
Google will allow competing Android app stores inside Google Play
Google and Epic Games have withdrawn their remaining challenges to a court order requiring Google to open Android to competing app stores. Beginning July 22, rival marketplaces will be eligible for distribution through Google Play in the United States, giving users a more direct way to install alternative stores without relying on conventional sideloading.
The change follows a long-running antitrust case in which a jury found that Google maintained an illegal monopoly in Android app distribution and in-app billing. Developers’ applications may be made available through participating third-party stores unless developers opt out. Google plans to charge marketplace operators an annual security review fee and to impose rules intended to prevent malware, fraud, and policy violations.
The agreement could create openings for companies such as Microsoft, Epic, Amazon, and independent developers to build Android marketplaces with different fees and payment terms. It may also weaken the economic model under which Google collects commissions from transactions made through Play.
Much will depend on how prominently alternative stores are displayed, whether consumers trust them, and whether Google’s security requirements create practical barriers. Even so, distributing rival marketplaces through Play is a greater concession than merely allowing users to download them from external websites.
Why It Matters: The change could weaken Google Play’s control over Android software distribution and give developers greater leverage over fees, payments, and customer relationships.
Source: The Verge.
TSMC’s AI chip profit surges 77% as it adds another $100 billion to its U.S. expansion
Taiwan Semiconductor Manufacturing Company reported a 77% year-over-year increase in second-quarter net profit, reaching a record NT$706.6 billion, or roughly $22 billion. Revenue climbed 34% to $40.2 billion as demand remained high for the advanced processors and semiconductor packaging used in artificial intelligence data centers. The results surpassed analysts’ profit expectations and prompted TSMC to raise its 2026 capital-spending forecast to between $60 billion and $64 billion.
TSMC also said it plans to invest another $100 billion in its Arizona operations, bringing its total planned U.S. investment to $265 billion. The expansion could include four additional factories alongside facilities already announced, with an emphasis on advanced manufacturing processes such as 2-nanometer chips. The company expects full-year revenue to increase by more than 40%, reflecting orders from customers that include Nvidia, Apple, AMD, and other major chip designers.
The announcement shows how AI infrastructure spending is reshaping both semiconductor economics and industrial policy. TSMC remains the central manufacturer behind much of the global AI industry, but the added U.S. investment also reflects pressure from governments and customers to reduce geographic concentration in Taiwan. The challenge will be scaling U.S. production without weakening the efficiency and supplier network that made TSMC dominant.
Why It Matters: TSMC’s results show that demand for AI chips remains strong while its U.S. expansion could materially alter the global semiconductor supply chain.
Source: Reuters.
U.S. Government Warns of Russian State Hackers Targeting Routers with Residential Proxies
CISA issued an alert about Russian state-sponsored actors increasingly compromising home and small-office routers to build residential proxy networks for malicious activities, including reconnaissance and potential attacks on critical infrastructure. The warning emphasizes the growing use of compromised consumer devices in sophisticated operations.
The advisory reflects the expanding attack surface created by internet-connected devices and the strategic value of residential proxies for evading detection in cyber espionage campaigns.
Why It Matters: Consumer hardware is increasingly being weaponized in state-sponsored cyber operations, raising the stakes for device security and network hygiene.
Source: Ars Technica.
OpenAI Discontinues Atlas AI Browser to Focus Resources on ChatGPT Super App
OpenAI announced it is sunsetting its standalone Atlas AI-powered web browser, redirecting development efforts toward deeper integration of AI capabilities directly into the ChatGPT application and desktop experience. The move follows a short-lived experiment with a dedicated browser product.
The decision signals OpenAI’s strategic pivot toward building an integrated “super app” ecosystem rather than competing directly in the browser market. Users are being encouraged to migrate workflows to the enhanced ChatGPT platform.
Why It Matters: Major AI labs are consolidating around core conversational and agentic platforms instead of fragmenting efforts across multiple consumer products.
Source: Engadget.
China prepares to use the Shanghai AI summit to promote its global governance model
Chinese President Xi Jinping is expected to present China’s vision for international AI governance at the World Artificial Intelligence Conference in Shanghai, which begins Friday. The gathering will bring together government officials, researchers, Chinese technology companies, and representatives from developing economies as Beijing seeks a larger role in setting the rules governing advanced AI systems.
Huawei is expected to demonstrate its Atlas 950 SuperPoD computing system, which uses the company’s Ascend processors and is intended to reduce dependence on restricted U.S. technology. Chinese chipmakers and AI startups will also present domestic computing platforms, open models, and applications. The conference comes before planned U.S.-China discussions on AI and alongside China’s proposal for a World AI Cooperation Organization.
China is positioning low-cost, open-weight AI models as an alternative for countries that cannot afford the infrastructure or commercial licensing associated with leading American systems. That approach gives Beijing a diplomatic opening across Southeast Asia, Africa, the Middle East, and Latin America. It also sets up a contest over whether global AI governance will be led by U.S.-aligned institutions, China-backed frameworks, or a collection of regional rules.
Why It Matters: China is treating AI standards and infrastructure as instruments of foreign policy, widening the technology competition beyond chips and models.
Source: Reuters.
Thinking Machines releases Inkling, its first open-weight AI model
Thinking Machines Lab, the AI startup founded by former OpenAI technology chief Mira Murati, has released its first internally developed model. Called Inkling, the system is open-weight, allowing developers and companies to download, modify, and run it on infrastructure they control rather than relying exclusively on a hosted application programming interface.
The release places Thinking Machines in a different category from frontier-model providers that primarily distribute their strongest systems through closed services. Inkling is intended to support customization, giving organizations more direct control over training, deployment, and model behavior. The company is betting that enterprises will increasingly resist one-size-fits-all systems as AI becomes embedded in specialized workflows, private data environments, and regulated industries.
Inkling will still need to prove that it can attract a serious developer community and deliver competitive performance. Open-weight distribution can accelerate adoption, but it also shifts hosting, security, and maintenance costs to users. For Thinking Machines, the release provides the first meaningful evidence of its technical direction after the company raised substantial capital while revealing relatively little about its products.
The model also adds another competitor to an open ecosystem already populated by Meta, Mistral, DeepSeek, Alibaba, and several smaller AI labs.
Why It Matters: Inkling gives developers another alternative to closed AI platforms and offers the first concrete view of Mira Murati’s strategy after leaving OpenAI.
Source: Fundpluse via Thinking Machines.
OpenAI enters consumer hardware with a $230 Codex control keyboard
OpenAI has introduced its first branded hardware product, a compact keyboard called the Codex Micro. The $230 device is built for developers who use multiple Codex coding agents and includes programmable controls, status indicators, and RGB lighting to show what each agent is doing without requiring users to constantly switch between windows.
Rather than replacing the computer, the keyboard acts as a physical control surface for supervising AI software. Developers can assign buttons to common actions, approve or stop tasks, move between agents, and monitor whether an agent is working, waiting for input, or has encountered an error. The product reflects the growing challenge of managing multiple autonomous coding processes simultaneously.
The Codex Micro is a modest entry into hardware compared with reports that OpenAI is developing broader consumer devices with former Apple design chief Jony Ive. Still, it provides a useful indication of how the company views the relationship between people and AI agents. As software moves from responding to individual prompts to performing longer sequences of work, physical interfaces may become another way to make those systems observable and easier to interrupt.
The device’s appeal may initially be limited to dedicated Codex users, but the concept could spread to other agent platforms and professional software.
Why It Matters: OpenAI’s first branded device turns AI-agent supervision into a hardware problem and offers an early look at interfaces built for autonomous software.
Source: Ars Technica.
China completes what it calls the first commercial invasive brain-chip implant
Chinese surgeons have implanted a coin-sized brain-computer interface in a patient with a spinal cord injury in what local authorities and researchers describe as the first commercial use of an invasive BCI system. Unlike experimental procedures conducted solely through clinical trials, the Chinese device reportedly received regulatory approval for use under a commercial medical program.
The implant records neural activity and translates signals into commands that can operate external devices. Its immediate purpose is to help people with severe paralysis regain communication or control capabilities. China has already approved several brain-computer interface products, including non-invasive rehabilitation systems and a semi-invasive implant, while research centers in Shanghai, Guangdong, and Jiangsu have built a growing clinical pipeline.
The development places additional competitive pressure on Neuralink and other American and European neurotechnology companies. Neuralink has implanted devices in trial participants, but broader commercial access remains subject to regulatory review.
Commercial approval does not eliminate unresolved questions about long-term implant stability, surgical risk, data privacy, device maintenance, and the reliability of neural decoding across different patients. Independent clinical evidence will be needed to evaluate the system’s safety and effectiveness.
Why It Matters: China’s move suggests that brain-computer interfaces are beginning to cross from experimental research into regulated medical treatment.
Source: South China Morning Post.
Microsoft Issues Record Patch Tuesday with 570 Vulnerabilities Fixed, Many Discovered via AI
Microsoft released its July 2026 Patch Tuesday updates addressing a record 570 security flaws across Windows and related products. The company credited internal AI systems with helping identify and prioritize a significant portion of the vulnerabilities. The update also addressed a new zero-day and long-standing issues with Secure Boot components, illustrating both the benefits and persistent challenges of AI-assisted security research.
Why It Matters: AI is transforming vulnerability discovery at scale, but the overall attack surface continues to expand rapidly.
Source: gHacks.
Neko Health raises $700 million to bring AI body-scanning clinics to the U.S.
Neko Health, the preventive-care startup co-founded by Spotify founder Daniel Ek and Hjalmar Nilsonne, has raised $700 million in Series C funding. The round values the Stockholm-based company at close to $7 billion and will support its planned U.S. launch, beginning with clinics in New York.
Neko combines full-body imaging, skin analysis, cardiovascular measurements, blood tests, and clinician consultations to identify health risks before symptoms become serious. The company already operates eight clinics in Sweden and the United Kingdom and says more than 100,000 people have completed scans. Its waiting list has grown to more than 350,000 prospective customers.
The financing reflects strong investor interest in technology-enabled preventive medicine, but it also raises questions about affordability, medical necessity, false positives, and the clinical value of broad screening for otherwise healthy consumers. Full-body screening can identify genuine risks, but additional findings may lead to unnecessary follow-up tests or anxiety.
Neko will also face a more complex regulatory, insurance, and healthcare provider environment in the United States than in its existing European markets.
Why It Matters: Neko’s expansion will test whether AI-assisted preventive screening can become a mainstream healthcare service rather than a premium wellness product.
Source: The Verge.
AI security acquisitions put cybersecurity M&A on course for a record year
Cybersecurity companies completed 219 mergers and acquisitions during the first half of 2026, placing the industry on track to exceed the roughly 400 transactions recorded last year. Strategic buyers are increasingly seeking companies that can secure AI models, autonomous agents, data pipelines, industrial systems, and the growing number of non-human identities operating inside corporate networks.
Deals involving AI security companies increased from 10 during all of last year to 29 during the first half of 2026, according to data cited by Momentum Cyber. Recent transactions include Akamai’s $205 million acquisition of browser-security company LayerX and Accenture’s $4.2 billion industrial cybersecurity purchase. Technology companies, consultancies, and established security vendors have been more active than private equity firms in several of the largest deals.
The buying activity reflects a structural gap in enterprise security. Existing identity and access-management products were largely built for employees and conventional software accounts, not for agents that can independently access files, write code, communicate with other systems, and execute transactions.
Acquirers are therefore paying for technical teams and specialized capabilities even when target companies have limited revenue. The result could be a faster consolidation of the security market around large platforms.
Why It Matters: AI agents are creating security problems that older products were not built to handle, driving established vendors to buy capabilities rather than develop them internally.
Source: The Wall Street Journal.
Indian AI coding startup Emergent becomes a unicorn after raising $130 million
Emergent has raised $130 million in Series C funding at a $1.5 billion post-money valuation, making the Indian AI coding company a unicorn just over a year after launch. The valuation is five times the $300 million figure attached to its previous funding round in January.
Private equity firm Creaegis led the new investment, joined by MNI Ventures-Claypond, Sentinel Global, Khosla Ventures, SoftBank Vision Fund 2, Lightspeed, and Y Combinator. The round brings Emergent’s total capital raised to $230 million. Its platform allows users to describe software in natural language and then generates, tests, and deploys applications with limited conventional coding.
Emergent’s rise demonstrates how quickly investors are backing AI development tools outside the United States. India has a large base of software engineers, IT-service companies, small businesses, and first-time entrepreneurs, giving local platforms a substantial potential market.
The company still faces competition from Cursor, Replit, Lovable, Bolt, GitHub Copilot, and AI features built directly into cloud platforms. Its valuation will ultimately depend on whether generated applications remain reliable after deployment and whether customers continue paying after the novelty of creating an initial prototype fades.
Why It Matters: Emergent’s funding shows that AI coding platforms are becoming a major global startup category rather than a market confined to Silicon Valley.
Source: The Economic Times.
China reportedly permits limited Nvidia H200 imports under tighter controls
Chinese authorities are reportedly allowing selected domestic companies to import Nvidia’s H200 AI processors, reversing some earlier restrictions while maintaining government oversight of who receives the chips. The approvals are expected to focus on organizations whose computing requirements cannot yet be met by domestically produced alternatives.
The H200 is less advanced than Nvidia’s newest processors but remains highly valuable for training and operating large AI models. Limited access could help Chinese technology companies address immediate shortages while Huawei, Cambricon, MetaX, and other domestic suppliers increase production of their own accelerators and computing systems.
For Beijing, the decision involves a difficult balance. Blocking Nvidia products can accelerate demand for Chinese chips and reduce dependence on American suppliers. But restricting access too aggressively could slow domestic AI development at a time when companies in the United States continue building larger clusters.
The policy also does not resolve whether Washington will approve all proposed shipments or impose additional licensing conditions. Semiconductor controls have changed repeatedly as officials attempt to limit military and strategic applications without completely cutting American companies out of China’s commercial market.
Why It Matters: Even limited access to H200 could alleviate near-term computing constraints for Chinese AI developers as the country builds a more independent semiconductor industry.
Source: South China Morning Post.
Asia startup funding reaches a multiyear high as China and AI deals rebound
Venture investment in Asia-based startups climbed to a multiyear high during the second quarter of 2026, according to an analysis of disclosed funding rounds. The increase was driven by larger transactions involving Chinese companies and by continued investor interest in artificial intelligence, semiconductors, robotics, and related infrastructure.
The recovery represents a marked change from the more cautious environment that followed China’s technology-sector crackdown, weak public markets, and strained relations with the United States. Large rounds are once again attracting sovereign funds, corporate investors, and global venture firms, although capital remains concentrated among a relatively small number of companies.
The figures suggest that investors are becoming more willing to separate geopolitical risk from the commercial opportunity presented by Asia’s engineering base, manufacturing capacity, and consumer markets. China’s progress in open AI models and domestic chip systems has also created investment opportunities that do not depend entirely on U.S. technology.
Still, the rebound does not mean funding conditions have normalized for every founder. Early-stage companies outside AI and other favored sectors continue to face longer fundraising cycles, and exit options remain uncertain. The headline increase is being shaped disproportionately by a group of large, capital-intensive deals.
Why It Matters: The funding rebound shows that global investors are returning to Asian technology companies, but the recovery remains heavily concentrated in AI and strategic infrastructure.
Source: Crunchbase News.
Fintech funding rises 23% even as the number of startup deals declines
Global fintech startup funding increased 23% during the first half of 2026 compared with the same period last year, but the total number of completed deals fell. Investors directed a larger share of their capital toward established companies and startups building AI tools, payments infrastructure, compliance software, fraud prevention, and financial data systems.
The divergence between funding value and deal count suggests that the market is recovering unevenly. Large rounds can lift aggregate totals even while younger companies struggle to secure seed and Series A financing. Investors appear more interested in businesses with measurable revenue, regulated market access, or infrastructure that can be sold to banks and other large financial institutions.
AI is influencing the category in two directions. Financial companies are buying automation software to reduce operating costs and improve underwriting, customer service, and compliance. At the same time, criminals are using AI to produce more convincing identity fraud and social-engineering attacks, increasing demand for defensive products.
The result is a fintech market less focused on consumer applications and speculative growth than during the previous investment cycle. Capital is moving toward companies that sit deeper inside the financial system and can solve expensive operational problems.
Why It Matters: Rising fintech investment masks a selective market in which fewer companies are receiving larger checks, particularly in AI, fraud prevention, and financial infrastructure.
Source: Crunchbase News.
Sheetz begins moving 11,000 virtual machines away from VMware
U.S. convenience-store chain Sheetz is migrating approximately 11,000 virtual machines across 838 locations away from VMware following years of uncertainty over licensing, product strategy, and costs under Broadcom’s ownership. The retailer plans to use StorMagic software as part of the transition.
The migration is notable because VMware has long been a foundational platform for enterprise data centers and distributed computing environments. Moving thousands of virtual machines across stores is a complex operational project that involves point-of-sale systems, inventory tools, back-office applications, security controls, and business continuity planning.
Broadcom has reorganized VMware’s product portfolio and shifted many customers toward bundled subscription licenses since completing its acquisition. The changes have prompted some businesses to consider alternatives from Nutanix, Microsoft, Red Hat, public-cloud providers, and smaller virtualization companies.
Sheetz’s decision does not mean VMware is losing every large customer, but it shows that organizations are willing to undertake difficult migrations when pricing and supplier predictability become serious concerns. Competitors now have an opening to gain customers who previously considered VMware too entrenched to replace.
Why It Matters: A large real-world migration suggests that Broadcom’s VMware strategy is creating meaningful opportunities for smaller infrastructure providers.
Source: Ars Technica.
AI industry leaders converge on regulation as frontier systems become more capable
The leaders of OpenAI, Google DeepMind, and Anthropic have each called for stronger oversight of the most capable AI systems, creating an unusual area of agreement among companies competing for leadership in frontier models. Their proposals generally support evaluations, reporting requirements, cybersecurity safeguards, and government involvement when systems cross defined capability thresholds.
The executives differ on how regulators should be structured and how much authority governments should exercise over model releases. They also disagree about the treatment of open-weight systems and whether rules should apply primarily to companies, computing resources, or individual models. Even so, the convergence marks a shift from the industry’s earlier emphasis on voluntary commitments and internal safety programs.
Support for regulation also carries competitive implications. Large AI labs have the legal teams, computing infrastructure, and compliance budgets needed to meet complex requirements. Smaller startups, academic groups, and open-source developers may find the same rules more burdensome.
Governments therefore face the challenge of addressing genuine national-security and public-safety concerns without creating regulations that permanently protect today’s largest companies from competition. Clear thresholds and narrowly defined obligations will be critical.
Why It Matters: Agreement among leading AI executives increases the likelihood of formal rules governing frontier models, but poorly structured regulation could strengthen incumbents at the expense of smaller competitors.
Source: Axios.
That’s your quick tech briefing for today. Follow us on X @TheFundpluse for more real-time updates.
