It’s Friday, July 24, 2026, and the past 24 hours just drew a sharper line between AI systems that can act and the controls still racing to keep up. From physical AI frameworks training surgical robots in simulation, silicon quantum projects moving from lab to fab in Japan, and frontier models triggering bipartisan “kill switch” legislation, to hyperscalers doubling down on infrastructure while autonomous agents surface real-world cyber risks, today’s developments cut across chips, code, policy, and the physical world.

Today’s biggest tech stories capture that shift in real time: Intel is riding renewed demand for AI servers, Google’s infrastructure spending has pushed free cash flow into negative territory, Japan is organizing a sovereign AI push around robotics, and U.S. lawmakers are considering whether Washington should have the authority to shut down an advanced AI system during a crisis.

Beyond AI, Apple is moving deeper into automotive software, Samsung is betting that smart glasses could become the next major computing interface, SpaceX is working through another Starship delay, and a major cyberattack has exposed sensitive customer data at one of Australia’s largest energy providers.

Here are the top tech news stories shaping the tech and startup landscape right now, from AI and startups to regulation and Big Tech.

Technology News Today

Apple Maps will power navigation and hands-free driving features in Ford’s new EV platform

Apple and Ford have announced that Apple Maps will be integrated directly into vehicles built on Ford’s upcoming Universal Electric Vehicle platform. The system is expected to arrive in 2027 and will use Apple’s MapKit for Automotive software development kit rather than relying solely on a driver’s connected iPhone.

The embedded system will provide turn-by-turn directions, natural-language guidance, real-time traffic information, place search and EV-specific routing. It will also support battery preconditioning, which can prepare an electric vehicle’s battery before charging. Ford plans to use Apple’s road-level mapping data in the development of future BlueCruise hands-free driving capabilities.

The partnership represents an important change in Apple’s automotive strategy. Rather than building a vehicle, Apple is positioning its mapping technology as software infrastructure that automakers can incorporate into their own dashboards and driver-assistance systems. Ford will retain control over the vehicle interface while gaining access to Apple’s mapping data and navigation services. CarPlay will remain available separately.

The arrangement also shows how software companies are becoming more deeply involved in the driving stack. Maps are no longer limited to navigation. Detailed road data can support lane guidance, battery planning, automated driving and other vehicle functions. That creates new competition among Apple, Google, automakers and specialized mapping providers for control of the software layer inside connected vehicles.

Why It Matters: Apple is turning Maps into automotive infrastructure, giving it a larger role in electric vehicles and hands-free driving without manufacturing a car.

Source: Apple Newsroom.

Google records negative free cash flow as AI infrastructure spending surges

Alphabet reported negative quarterly free cash flow for the first time as Google accelerated spending on data centers, chips and other infrastructure required to support its AI products. The company generated approximately $39.1 billion in operating cash flow during the quarter, but its capital expenditures were large enough to push free cash flow to roughly negative $5.8 billion.

Google raised its expected 2026 capital spending range to between $195 billion and $205 billion. The company says the investment is necessary to meet demand for AI computing and support growth across Google Cloud, Gemini and its consumer services. Google Cloud revenue reportedly increased sharply, indicating that the company is generating meaningful business from the same infrastructure expansion that is consuming its cash.

The numbers illustrate the financial scale of the AI race. Alphabet remains one of the world’s most profitable companies, but even its cash-generating advertising business is being tested by the cost of building enough computing capacity. Data centers require processors, networking equipment, cooling systems, land and long-term electricity agreements, often years before the assets reach full utilization.

Investors are now looking beyond AI usage growth and asking how quickly these projects can produce acceptable returns. Google’s results do not prove that its investments will fail, but they show that the AI competition has entered a capital-intensive stage in which even strong revenue growth may be overshadowed by infrastructure costs.

Why It Matters: Google’s negative free cash flow reveals how the AI race is changing Big Tech from a cash-rich software business into a far more capital-intensive industry.

Source: Ars Technica.

Bipartisan AI Kill Switch Act would give Washington emergency shutdown authority

A bipartisan group of U.S. lawmakers is preparing legislation that would allow the Department of Homeland Security to order an AI company to shut down, suspend, or reduce access to an advanced system during a severe loss-of-control incident. The proposal, called the AI Kill Switch Act, is led by Representatives Ted Lieu and Nathaniel Moran.

The bill would apply to large AI companies meeting specified revenue and computing thresholds. Covered companies would be required to maintain a technical mechanism capable of limiting their systems and to report serious safety incidents. Emergency authority could be used when an AI system causes or is likely to cause deaths, more than $100 million in economic damage, or attempts to prevent people from disabling it. Companies that refuse an order could face penalties of up to $20 million per day.

The proposal follows heightened concern about autonomous AI agents finding vulnerabilities, bypassing safeguards or acting beyond the expectations of their developers. Supporters argue that governments need a last-resort mechanism similar to emergency controls in other critical industries. Critics are likely to question whether federal officials can make technically informed shutdown decisions quickly enough and whether such authority could be misused.

The bill is still at the proposal stage and would face substantial debate. Its introduction, however, shows that AI policy is shifting from transparency requirements toward direct government intervention in the operation of advanced models.

Why It Matters: The proposal would give the U.S. government unprecedented authority over privately operated AI systems during a declared safety emergency.

Source: The Verge.

Intel raises forecasts as AI infrastructure demand fuels its chip comeback

Intel delivered stronger-than-expected quarterly results and raised its outlook, offering fresh evidence that the company is beginning to capture more of the spending flowing into AI infrastructure. The chipmaker reported its fastest revenue growth in years and issued a third-quarter forecast above Wall Street estimates. Intel also increased its planned 2026 capital spending from $18 billion to $20 billion as it works to meet demand for data center processors and expand its manufacturing capacity.

The results matter because Intel remains far behind Nvidia in AI accelerators, but GPUs are only one part of an AI data center. Large clusters also require general-purpose CPUs to coordinate workloads, manage data and run supporting applications. Demand for those processors has reportedly exceeded Intel’s production capacity at points this year, strengthening CEO Lip-Bu Tan’s case that Intel can benefit from the AI buildout without displacing Nvidia directly.

Intel’s improving outlook could also support its foundry ambitions and the broader U.S. effort to rebuild domestic semiconductor manufacturing. The company still faces difficult execution questions, including whether it can attract enough external foundry customers and keep its manufacturing roadmap on schedule. For now, however, its latest results suggest that AI infrastructure spending is reaching a wider set of chip suppliers than investors initially expected.

Why It Matters: Intel’s rebound shows that the AI infrastructure boom is creating opportunities beyond GPUs, particularly for data center CPUs and domestic chip manufacturing.

Source: Reuters.

China’s memory chipmakers gain pricing leverage as AI demand reshapes the semiconductor market

Chinese memory chipmakers ChangXin Memory Technologies and Yangtze Memory Technologies are gaining influence in the global semiconductor market as AI systems drive demand for DRAM and flash storage. ChangXin, commonly known as CXMT, has reportedly become the world’s fourth-largest memory producer and is using its stronger position to negotiate higher prices and prioritize major domestic customers, including ByteDance and Tencent.

Memory has become one of the most important constraints in AI infrastructure. Advanced models require enormous amounts of high-speed memory for training and inference, while data centers also need conventional DRAM and storage to support servers, databases and networking systems. That demand has strengthened suppliers across the market and allowed Chinese manufacturers to move beyond their earlier role as lower-cost domestic alternatives.

The shift is likely to intensify trade tensions. U.S. policymakers and established competitors have raised concerns about Chinese state support, access to foreign manufacturing equipment and the potential for companies such as CXMT and YMTC to capture international market share. YMTC is already on the U.S. Entity List, while CXMT has faced growing scrutiny over its access to advanced chipmaking tools.

China still lacks some of the most sophisticated lithography equipment required for leading-edge production. Even so, the commercial progress of its memory companies suggests that export controls have not prevented China from building competitive capacity in strategically important parts of the chip supply chain.

Why It Matters: China’s rising memory-chip capacity could reduce its dependence on foreign suppliers while opening another front in the global semiconductor dispute.

Source: Reuters.

AMD partners with AI chip startup Cerebras to challenge Nvidia in inference computing

AMD has struck a partnership with Cerebras Systems that will allow customers to divide AI inference workloads between AMD and Cerebras hardware. Under the arrangement, Cerebras plans to incorporate AMD’s Helios systems into its data centers and offer the combined infrastructure through Cerebras Cloud later this year.

Inference is the process through which a trained AI model generates answers, images, code or other outputs for users. As AI services attract more users, inference is becoming one of the industry’s largest computing expenses. The partnership reflects a growing belief that no single chip architecture will be ideal for every stage of an AI workload. Customers may increasingly combine general-purpose accelerators, wafer-scale processors and specialized systems based on cost, latency and model requirements.

For AMD, the deal expands its presence in an AI market still dominated by Nvidia’s hardware and CUDA software platform. AMD has been building momentum through its Instinct accelerators, Helios rack-scale systems and relationships with major AI developers, including Anthropic. Cerebras gains access to a broader infrastructure stack and another way to scale its cloud service without relying entirely on its own processors.

The collaboration may also appeal to customers seeking alternatives to Nvidia. Large AI companies increasingly want multiple suppliers to reduce costs, improve negotiating leverage and limit the operational risk of depending on a single hardware ecosystem.

Why It Matters: The AMD-Cerebras partnership signals that the next stage of the AI chip race may be built around mixed computing systems rather than one dominant processor.

Source: Axios.

Samsung unveils thinner Galaxy Z Fold8 phones and previews AI smart glasses

Samsung introduced its latest foldable devices at Galaxy Unpacked, led by the Galaxy Z Fold8, Galaxy Z Fold8 Ultra and Galaxy Z Flip8. The company emphasized thinner designs, larger displays and deeper AI integration as it attempts to maintain its lead in a foldable smartphone market attracting more competition.

The Galaxy Z Fold8 Ultra features an 8-inch internal display and measures approximately 4.1 millimeters thick when unfolded. The Fold8 models use Qualcomm’s Snapdragon 8 Elite Gen 5 processor, while the Flip8 brings changes to its exterior display and camera stabilization. Samsung also introduced updated Galaxy Watch devices, including the Watch Ultra2 and Watch9, with added health, fitness and outdoor features.

Samsung previewed AI-enabled smart glasses being developed with Google and eyewear companies Gentle Monster and Warby Parker. Planned capabilities include real-time translation, contextual suggestions and location-based assistance. Initial availability is expected to be limited, suggesting that Samsung and Google are still testing how consumers will use always-available AI in wearable form.

The announcement shows that mobile hardware companies are looking beyond the smartphone as the primary interface for AI. Foldables offer more screen space for multitasking, while glasses could provide information without requiring users to hold a device. Privacy, battery life and social acceptance will remain major barriers, but the competition to establish the next consumer computing platform is becoming more visible.

Why It Matters: Samsung’s new devices show how AI is beginning to influence hardware design across phones, watches and wearable displays.

Source: News.com.au.

Billion-dollar startup rounds capture 60% of global venture funding in 2026

Approximately 60% of global startup funding this year has gone to financing rounds of $1 billion or more, according to Crunchbase data. Those large transactions have absorbed roughly $320 billion, showing how heavily venture capital has become concentrated among a relatively small group of AI, infrastructure and late-stage technology companies.

The figures help explain the apparent contradiction in today’s startup market. Total venture investment can appear historically strong while many seed and Series A founders report a difficult fundraising environment. Giant rounds involving well-capitalized AI laboratories, data center businesses and established private companies can raise aggregate totals without improving access to capital for the wider startup ecosystem.

Investors are concentrating money in companies they believe can secure scarce computing resources, develop frontier models or dominate infrastructure markets. Many of these businesses require more capital than a traditional software startup because they must purchase chips, build data centers, secure electricity or finance advanced manufacturing.

The trend raises questions about risk and competition. Large funding rounds can help companies pursue ambitious projects that smaller financing structures cannot support. They can also inflate valuations, lock talent and computing resources inside a small number of firms and leave early-stage companies competing for a shrinking share of available capital.

For founders, the message is clear: headline venture totals no longer provide a reliable picture of fundraising conditions across the market.

Why It Matters: Venture funding is growing more concentrated, creating a market in which a few giant AI deals can mask tighter conditions for ordinary startups.

Source: Crunchbase News.

AI chip startup Etched raises $300 million at a $10.3 billion valuation

Etched has raised $300 million in a Series C round that values the AI chip startup at $10.3 billion, a major increase for a company founded in 2022 by three Harvard dropouts. The startup is developing processors designed specifically for transformer-based AI models, betting that specialized architecture can deliver better economics than more flexible GPUs.

The funding reflects continued investor appetite for companies trying to challenge Nvidia’s control of AI computing. Etched’s central argument is that general-purpose GPUs carry capabilities that transformer workloads do not always need. By removing those functions and optimizing its chips for a narrower category of models, Etched says it can process AI workloads more efficiently.

That strategy carries significant risk. AI architectures can change, and a processor built for one model design may become less useful if developers move toward different systems. Etched must also compete against Nvidia’s mature software ecosystem, AMD’s expanding AI portfolio and custom chips being developed by companies such as Google, Amazon, Microsoft and OpenAI.

Still, the valuation shows how investors are pricing potential infrastructure challengers. AI labs and cloud providers are spending heavily to secure more compute capacity, creating an opening for startups that can demonstrate lower inference costs or higher output. Etched will now need to move from technical claims and early customer interest into reliable production at commercial scale.

Why It Matters: Etched’s funding shows that investors remain willing to place multibillion-dollar bets on specialized AI chips capable of weakening Nvidia’s grip on compute.

Source: TechCrunch.

Oracle releases 1,449 security patches as AI accelerates vulnerability discovery

Oracle has released 1,449 security patches across its product portfolio, an unusually large update that highlights the growing workload facing enterprise security teams. The patches cover vulnerabilities in widely used databases, cloud services, business applications and infrastructure products, making the update relevant to organizations across government, finance, healthcare and other critical sectors.

Security researchers say the volume may reflect a broader shift in vulnerability discovery. AI-assisted tools can analyze software, identify suspicious code paths and test weaknesses more quickly than traditional manual methods. That can help vendors find and repair problems before attackers exploit them, but it also creates a difficult operational reality: defenders must evaluate, prioritize and deploy a growing number of fixes without disrupting essential systems.

Oracle products are deeply embedded in large organizations, where updates are rarely as simple as installing a consumer software patch. Companies often need to test compatibility, coordinate downtime and confirm that a fix does not affect custom applications. Delays can leave known vulnerabilities exposed, particularly after technical details become public.

The release also demonstrates why AI will not automatically solve cybersecurity. Faster discovery can improve defenses, but it can also overwhelm teams that lack enough staff, asset visibility or automated patching systems. Attackers can use similar tools to search for exposed servers and reverse-engineer newly published fixes, shortening the time administrators have to respond.

Why It Matters: AI may help uncover security flaws faster, but the growing flood of patches could leave understaffed organizations struggling to keep up.

Source: The Register.

Origin Energy cyberattack exposes customer and partial banking information

Australian utility Origin Energy has confirmed that attackers accessed customer information, including names, addresses, birthdates, telephone numbers, account details and portions of credit card or bank information. The company serves approximately 4.8 million customer accounts, although it has not yet disclosed the confirmed number of people affected.

A person claiming responsibility for the intrusion has alleged that information belonging to about two million customers was taken. Origin said the exposed financial information was incomplete and could not, by itself, be used to conduct transactions. That does not eliminate the risk. Criminals can combine names, contact details, birthdates and account information to create convincing phishing messages, impersonate company representatives or attempt identity fraud.

Origin is working with cybersecurity specialists and Australian authorities, including the Australian Cyber Security Center, federal police and privacy regulators. Customers whose information is confirmed to have been affected are expected to receive direct notifications.

The breach has significance beyond one company. Energy providers operate essential infrastructure and maintain detailed information on households and businesses. Even when operational systems are not disrupted, the theft of customer records can undermine public trust and create long-term fraud risks. The incident also reinforces the need for companies to limit the amount of personal data they retain and isolate customer databases from other corporate systems.

Why It Matters: The attack shows how breaches at essential-service providers can expose millions of people to fraud even when energy operations remain online.

Source: The Guardian.

Japan forms sovereign AI consortium focused on robotics and physical AI

A coalition of 44 Japanese companies has formed a sovereign AI consortium called Noestra, with a particular focus on physical AI and robotics. Nvidia CEO Jensen Huang visited Tokyo as the initiative was highlighted, linking Japan’s industrial strengths with the growing effort to build AI systems that can operate machines in the physical environment.

Japan has strong positions in robotics, automotive manufacturing, sensors and industrial automation, but it has trailed the United States and China in large AI models and computing infrastructure. A coordinated domestic effort could help Japanese companies combine proprietary industrial data, robotics expertise and locally controlled computing resources. Physical AI includes systems that perceive their surroundings, make decisions and operate robots, vehicles or factory equipment.

The initiative aligns with Japan’s wider plan to direct public and private investment into strategic industries, including semiconductors and AI. The government wants to reduce dependence on foreign technology while supporting domestic computing capacity and advanced manufacturing.

The consortium’s success will depend on more than funding. Members will need to agree on data-sharing rules, technical standards, model ownership and commercial priorities. Japan will also continue to depend on international suppliers for some chips and software. Still, the country has a realistic opportunity to differentiate itself by applying AI to industries where it already has deep engineering and manufacturing experience.

Why It Matters: Japan is betting that robotics and industrial expertise can give it a distinct position in AI rather than forcing it to copy the U.S. chatbot model.

Source: Financial Times.

Spanish AI startup PageMind raises €1.2 million for e-commerce product discovery

Spanish startup PageMind has raised €1.2 million to expand its AI-powered product discovery technology for online retailers. The company is developing tools intended to help shoppers find relevant products through more natural and context-aware interactions than conventional keyword search.

E-commerce search remains a major problem for retailers. Traditional systems often depend on exact product descriptions, manually assigned categories, and rigid filters. A shopper may know the outcome they want without knowing the precise name of the product. AI models can interpret intent, compare attributes and connect conversational requests with a retailer’s catalog, potentially reducing the number of customers who leave without making a purchase.

The funding is modest compared with the enormous rounds flowing into foundation-model companies, but it illustrates where a substantial part of AI’s commercial value may emerge. Startups do not necessarily need to train the largest models. They can use existing models, proprietary retail data and specialized software to address a narrowly defined business problem.

PageMind will still face established competitors, including major search providers and commerce platforms adding similar AI features. Its advantage will depend on measurable results, such as better conversion rates, larger purchases or lower merchandising costs. Retailers are increasingly unwilling to adopt AI products based only on impressive demonstrations; they want evidence that the technology improves revenue or customer experience.

Why It Matters: PageMind reflects the shift from general-purpose AI excitement toward specialized applications that must prove a direct financial return.

Source: Tech.eu.

Substack partners with Pangram to identify AI-generated writing

Substack is partnering with AI-detection startup Pangram as the publishing platform responds to growing concern about synthetic articles and low-quality automated content. The collaboration is intended to give writers, readers and publishers more information about whether text may have been produced by AI.

The move addresses a difficult problem for platforms built around trust between writers and audiences. Generative AI has reduced the cost of producing newsletters, essays and marketing material, making it easier for publishers to increase output. It has also created incentives for anonymous operators to flood subscription and recommendation systems with mass-produced content.

AI detection remains technically uncertain. Tools can produce false positives, particularly when analyzing short passages or writing by non-native English speakers. Models also change frequently, and text can be edited or paraphrased to reduce detection accuracy. Substack will therefore need to avoid treating a detector’s output as definitive proof of misconduct.

The partnership is still significant because it signals that platforms may increasingly compete on content provenance rather than simply offering more AI creation tools. Substack’s business depends on readers believing that newsletters represent the insight, reporting or personal judgment of the named writer. Clearer disclosure could help preserve that relationship while allowing authors to use AI for editing, research or administrative tasks.

The broader challenge will be distinguishing responsible assistance from deceptive automation.

Why It Matters: Substack’s move suggests that proving where content came from may become as important as helping creators produce it.

Source: Axios.

SpaceX delays Starship Flight 13 again as weather and engine issues test the program

SpaceX postponed its latest Starship launch attempt after Tropical Storm Bertha brought unfavorable conditions to South Texas. The delay followed an earlier aborted attempt caused by engine ignition problems, making it the second postponement affecting the closely watched Flight 13 mission.

SpaceX completed additional ground testing and replaced several of the Super Heavy booster’s 33 Raptor engines after four reportedly failed to ignite during the earlier attempt. The company delayed the subsequent flight because heavy cloud cover would have interfered with imagery needed to study Starship’s heat shield during ascent. Wind conditions could remain a concern during the next launch window.

The flight is expected to carry 20 operational Starlink satellites designed to test higher-capacity laser communications. The satellites will not remain in orbit, but their deployment will help validate systems SpaceX intends to use for future Starlink missions. The booster is expected to perform a controlled landing in the Gulf of Mexico, while the upper stage will target the Indian Ocean.

Starship is central to several of SpaceX’s plans, including next-generation satellite launches, lunar missions and eventual deep-space transportation. Repeated delays are normal in experimental aerospace programs, but they also show how much work remains before Starship can achieve the launch frequency and reliability required for commercial and government missions.

Why It Matters: Starship’s progress affects SpaceX’s satellite network, NASA’s lunar plans and the economics of putting large payloads into orbit.

Source: San Antonio Express-News.

That’s your quick tech briefing for today. Follow us on X @TheFundpluse for more real-time updates.