Tech news today, October 28, 2026: Google, OpenAI, Oracle, Nvidia, Scale AI, SK Hynix & More. Daily Global Technology Digest | Wednesday, October 28, 2026 It's October 28, 2026, and the tech landscape is shifting in real time: new AI capacity deals, platform power plays. The global AI industry is entering Wednesday with another extraordinary financial milestone. South Korean chipmaker SK Hynix has reported record quarterly revenue of approximately $75.5 billion and operating profit approaching $63 billion, driven by relentless demand for advanced memory. The company says its production is effectively committed through 2029. Meanwhile, China's leading memory manufacturer is preparing to produce advanced AI chips despite manufacturing yields far below those of its South Korean competitors. The battle for computing infrastructure is also taking a new direction. OpenAI is reportedly planning to purchase Google's Tensor Processing Units, potentially deepening a relationship between two companies that compete directly in artificial intelligence. Nvidia and Oracle are expanding their collaboration around Vera Rubin processors, while AI infrastructure startup General Compute has raised $100 million to develop a cloud platform built around specialized inference hardware. Investors are continuing to make substantial commitments despite growing concerns about AI spending. Dallas-based investment firm Disruptive has announced a $10 billion fund targeting late-stage technology and defense companies. Financial technology startup Casap has secured $56 million to help banks investigate fraudulent payment disputes, while Swiss biotechnology startup Atlas Bio has emerged from stealth with funding to develop AI simulations of clinical trials. But the industry's expansion is creating new security and regulatory challenges. Privacy organization Noyb has filed complaints against Scale AI over alleged surveillance of workers using its Outlier platform. Researchers have also uncovered evidence that automated submissions accounted for 76% of responses to a British government consultation, demonstrating how technology can distort public participation in policymaking. Elsewhere, a major education software breach has exposed sensitive student information, Google Cloud and NASA are testing AI-controlled robots for future space missions, and European banks are pursuing locally governed AI infrastructure. The larger story is becoming clearer: the next phase of AI competition will depend on far more than model performance. Memory supplies, computing efficiency, software verification, regulatory accountability, and practical business applications are becoming increasingly important to the industry's long-term economics. From record semiconductor profits and multibillion-dollar investment funds to AI cybersecurity, biotechnology, robotics, and international regulation, here are today's top tech news stories. OpenAI reportedly plans to buy Google's AI chips as competition with Nvidia intensifies. OpenAI is reportedly planning to purchase artificial intelligence processors developed by Google, potentially creating another major commercial relationship between two of the industry's largest competitors. Reuters reported Wednesday, citing The Information, that the ChatGPT developer is considering buying Google's Tensor Processing Units, or TPUs. The report did not disclose the number of processors, a potential purchase price, or a delivery schedule. OpenAI declined to comment, while Google did not immediately respond to a request for comment. The proposed arrangement therefore remains a reported plan rather than a confirmed purchasing agreement. OpenAI already rents Google AI chips and recently announced plans to distribute its frontier models through Google Cloud. The development could have important implications for the AI semiconductor market. Nvidia remains the dominant supplier of advanced processors used to train and operate large models, but customers are increasingly interested in alternatives that could improve availability or reduce computing costs. Google developed TPUs primarily for its own computing requirements before expanding their availability to external customers. The company has also secured agreements with other AI developers, including Anthropic. For OpenAI, purchasing hardware from several suppliers could reduce dependence on a single processor architecture. The company is also developing custom AI chips with Broadcom, creating another potential source of computing capacity. The broader commercial relationship is notable because OpenAI and Google compete directly in AI models, consumer applications, enterprise services, and search-related products. Their cooperation illustrates how expensive computing infrastructure is encouraging partnerships between companies that remain competitors elsewhere. Why It Matters: OpenAI's reported TPU purchasing plans could strengthen Google's position in the AI chip market while increasing competitive pressure on Nvidia. Google Cloud and NASA test ai-controlled robot for future moon and Mars missions. Google Cloud and NASA are testing an AI-controlled robotic system designed to assist astronauts during future deep-space missions. The project combines robotics, computer vision, and a vision-language-action model that translates instructions into physical movements. In early testing at NASA's Johnson Space Center, the system was assigned a task to locate and move a medical bag. Google says the robot achieved an initial success rate of 85% under controlled laboratory conditions. Researchers used the open-source Pi0 model, developed by Physical Intelligence, and adapted it using Google Cloud computing infrastructure. The model contains approximately four billion parameters and was fine-tuned in about two and a half hours, according to Google's technical report. The research addresses a practical problem facing missions beyond Earth's immediate vicinity. Astronauts traveling to Mars may experience communication delays of many minutes, making continuous assistance from ground controllers impossible. Robots capable of performing physical tasks could help crews manage equipment, retrieve supplies, and respond to operational needs without waiting for instructions from Earth. Unlike software agents, physical robots must operate in environments where mistakes can damage equipment or threaten human safety. The researchers are therefore using simulated environments and controlled experiments to evaluate movements before considering more demanding applications. The current demonstration does not establish that the system is ready for a space mission. Its reported success rate reflects a specific laboratory task rather than general-purpose robotic reliability. Nevertheless, the project illustrates how advances in AI models and physical automation could eventually support exploration, remote operations, and other environments where human assistance is limited. Why It Matters: The Google-NASA collaboration demonstrates how physical AI could support future space missions while advancing robotic technologies useful in remote and hazardous environments on Earth. SK Hynix reports record $75.5 billion quarterly revenue as AI memory demand extends through 2029. South Korean semiconductor manufacturer SK Hynix reported record third-quarter revenue of 104.62 trillion won, about $75.5 billion, as demand for memory used in artificial intelligence systems continued to outstrip supply. Operating profit rose 632.1% from a year earlier to 83.33 trillion won, or about $63 billion. The company's operating margin reached 79.7%, an unusually high level for a semiconductor manufacturer. SK Hynix attributed the results to strong demand and higher prices for high-bandwidth memory and advanced server DRAM. The company also announced additional shareholder returns, bringing its planned dividends and share repurchases for 2026 to approximately 100 trillion won. The company's outlook also signals the pressure facing the global AI supply chain. SK Hynix says its DRAM and high-bandwidth memory production is effectively committed through 2029, based on signed agreements and customer demand under discussion. More than half of its 2027 and 2028 capacity is covered by signed agreements, although not every prospective order is contractually binding. SK Hynix expects its newer HBM4 products to surpass HBM3E in both shipments and revenue during the fourth quarter. The company is also preparing to expand manufacturing capacity, including production at its Yongin facility. For AI developers, the results highlight a growing constraint. Advanced processors require substantial amounts of specialized memory to operate efficiently, making memory availability increasingly important to the economics of computing infrastructure. Why It Matters: SK Hynix's record earnings demonstrate how the AI infrastructure boom is translating into extraordinary profits for memory suppliers while creating longer-term supply constraints for computing companies. Big Tech Earnings Week Tests Returns on Massive AI Infrastructure Spending Amazon, Microsoft, and Alphabet are expected to report accelerating cloud growth this week, but investors remain focused on the sustainability of AI capital expenditures that could reach as much as $600 billion combined across the three companies this year. Google Cloud is projected to grow roughly 95 percent in the September quarter, Azure around 48 percent, and AWS about 38 percent. Meta's recent results showed advertising strength offset by negative free cash flow and higher capex guidance. Markets have been turbulent amid concerns over financing costs and return timelines. Microsoft and Alphabet report on October 28; Amazon follows the next day. Analysts are watching cloud backlog, guidance on future spending, and any concrete productivity or revenue metrics tied to AI products. The results will influence broader sentiment toward the AI investment cycle. Why It Matters: This earnings cycle is a key litmus test for whether the largest AI infrastructure investments are translating into durable revenue growth or simply higher costs. China's CXMT prepares to manufacture advanced AI memory despite yields of just 20% to 30%. China's leading memory chipmaker, ChangXin Memory Technologies, is preparing to begin limited production of advanced HBM3E memory despite manufacturing yields substantially below those of its South Korean competitors. Reuters reported Wednesday that CXMT expects to start production within approximately two months, citing people familiar with the company's plans. Only 20% to 30% of its manufactured HBM3E chips currently meet the required specifications, compared with estimated yields of around 70% at SK Hynix. CXMT has already provided samples to several Chinese AI processor developers. The company did not respond to Reuters' request for comment, and it has not publicly confirmed its production schedule. The decision illustrates the pressure on China's semiconductor industry as American export restrictions limit access to advanced foreign technology. High-bandwidth memory is essential to powerful AI processors because it allows enormous amounts of data to move quickly between memory and computing components. Low manufacturing yields create substantial financial challenges. If most chips cannot be sold, manufacturers must spread production costs across fewer usable components. Nevertheless, domestic supply may have strategic value even when initial production is relatively inefficient. Chinese AI hardware developers need access to memory components to support alternatives to Nvidia processors. CXMT also faces technological challenges beyond manufacturing yields. According to Reuters, its HBM3E products remain behind leading South Korean offerings in some performance evaluations. The announcement underscores the difference between demonstrating that advanced semiconductor technology can be produced and manufacturing it economically at commercial scale. Why It Matters: CXMT's manufacturing plans show how semiconductor restrictions are accelerating China's domestic AI hardware development, even when production costs and technical performance remain challenging. Nvidia and Oracle expand Vera Rubin AI cloud infrastructure with focus on GPU efficiency. Nvidia and Oracle are expanding their cloud infrastructure partnership to support Nvidia's Vera Rubin computing architecture, including Vera CPUs and integrated AI systems. The companies outlined the plans Tuesday as businesses increasingly question whether expensive AI processors are being used efficiently. Oracle intends to make Vera processors available through bare-metal servers and virtual machines, allowing customers to operate CPU-based services separately from GPU infrastructure. Nvidia is also promoting integrated systems designed to support increasingly demanding AI workloads. The announcement builds on the companies' collaboration around large computing installations and next-generation data-center architecture. The focus on CPUs reflects an important change in how AI infrastructure is being designed. Large AI models depend heavily on GPUs, but autonomous agents also perform tasks involving software execution, database operations, networking, and other activities that rely on conventional processors. If those supporting activities become bottlenecks, purchasing additional GPUs may not improve overall performance. VentureBeat's recent survey research found that 79% of 218 respondents reported GPU utilization of 50% or less. The findings are self-reported and do not establish how much capacity is genuinely wasted. Nvidia and Oracle are attempting to address part of that problem by offering computing systems better suited to complete AI workflows rather than isolated model execution. However, hardware improvements alone cannot solve problems caused by poorly designed applications, limited customer demand, or inefficient software. For enterprises, the relevant question is whether new infrastructure reduces the total cost of completing useful work. Why It Matters: Nvidia and Oracle's expansion reflects growing demand for AI infrastructure that improves computing efficiency rather than simply adding more processors. Disruptive announces $10 billion fund to back AI, defense, and late-stage technology companies. Dallas-based investment firm Disruptive has announced a $10 billion investment fund targeting late-stage companies in artificial intelligence, defense technology, and other capital-intensive industries. The firm plans to invest in at least 10 businesses, with individual commitments potentially reaching $1 billion. Disruptive was founded in 2012 by investor Alex Davis and has previously backed companies including Palantir, Airbnb, Spotify, Groq, and ElevenLabs. Unlike its earlier approach of raising capital for individual transactions, the new vehicle represents the firm's first traditional blind-pool fund, allowing managers to allocate committed capital across multiple investments. TechCrunch reported the announcement Tuesday The fund represents another substantial commitment to technology companies that require large amounts of money before reaching public markets. Traditional venture capital firms often spread investments across dozens or hundreds of startups, expecting a relatively small number to generate most of the returns. Disruptive's strategy concentrates significant capital in a smaller group of established businesses. That approach could provide late-stage companies with access to financing without immediately pursuing initial public offerings. It could also create substantial exposure if valuations decline or portfolio companies struggle to reach profitability. The focus on AI and defense reflects growing investor interest in businesses developing strategically important technologies with expensive research and infrastructure requirements. However, a large fund does not guarantee attractive investment returns. Performance will depend on entry valuations, company execution, liquidity opportunities, and the timing of eventual exits. Why It Matters: Disruptive's $10 billion fund highlights continued institutional appetite for large investments in AI and defense despite growing scrutiny of technology valuations. General Compute raises $100 million at $800 million valuation to build specialized AI inference cloud. AI infrastructure startup General Compute has raised $100 million in Series A funding at an $800 million post-money valuation, betting that specialized processors can make AI inference more economical. Blackbird and Abstract Ventures led the financing, with participation from Fuse Venture Partners, Village Global, and other investors. General Compute operates in the emerging market for specialized AI cloud services, where companies compete to provide computing capacity outside the largest cloud platforms. Rather than focusing exclusively on Nvidia GPUs, the startup is developing infrastructure that can match workloads with different computing architectures, including specialized systems from companies such as Cerebras, SambaNova, and Positron. The investment comes during a difficult period for AI infrastructure startups. Investors are questioning whether every company building GPU capacity can generate sufficient demand and attractive margins to justify its spending. General Compute is pursuing a different strategy by concentrating on inference, the process of running trained AI models to generate responses and complete tasks. Inference requirements can differ substantially depending on model size, response speed, memory needs, and the number of simultaneous users. A processor optimized for one workload may perform poorly on another. Specialized computing services could help customers reduce costs by letting them select hardware that closely matches their requirements. However, operating a profitable cloud business still requires reliable infrastructure, effective software integration, sufficient customer utilization, and disciplined capital spending. The funding shows investors remain willing to back differentiated AI infrastructure providers, even as concerns grow about companies building largely interchangeable computing capacity. Why It Matters: General Compute's financing suggests investors are becoming more interested in AI infrastructure businesses that compete through computing efficiency rather than hardware capacity alone. Scale AI faces five European privacy complaints over Outlier worker surveillance. Privacy organization noyb has filed five complaints against Scale AI over alleged violations of European data protection law involving workers using its Outlier platform. The complaints were submitted Tuesday in France, Italy, the Netherlands, and Austria. Outlier connects freelance contributors with tasks used to evaluate and improve AI systems, including reviewing model responses and preparing training information. According to noyb, the platform has more than 900,000 freelance workers. The organization alleges that Scale AI requires invasive biometric identity verification and allows third-party tracking technologies to collect information about workers' online activity. The complaints seek investigations by European data protection authorities. Regulatory proceedings have not established the allegations as legal violations. The complaints highlight an often-overlooked part of the AI industry: the human labor required to prepare training data and evaluate model behavior. Advanced AI systems depend on contributions from specialists who review answers, assess accuracy, and provide examples used during model development. According to noyb, Outlier uses tracking technologies associated with companies including TikTok, LinkedIn, Facebook, and Datadog. The organization also objects to identity verification procedures involving facial recordings and government identification documents. Companies have legitimate reasons to prevent fraud and verify workers' identities, but European privacy rules impose requirements on necessity, proportionality, consent, and handling sensitive information. For AI developers, the outcome could influence how training platforms recruit contributors, monitor work, and verify identities. The dispute also shows that regulatory scrutiny of AI increasingly extends beyond models themselves to the employment arrangements and data-processing systems that support their development. Why It Matters: The complaints could establish important precedents for privacy protections and worker monitoring across the AI data-labeling industry. AI bots generated 76% of responses to UK government consultation, new research finds. Automated submissions accounted for approximately 76% of responses to a British government consultation on tobacco and disposable vaping products, according to research disclosed Tuesday. Researchers examining the 2023 consultation found that bots generated 90,835 of 118,756 responses. Almost all the automated submissions opposed restrictions on disposable vaping products. Many contained repeated material and originated from a relatively small number of internet addresses. Britain's Department of Health and Social Care identified the suspicious submissions and excluded them from its decision-making process. The proposed restrictions subsequently became law. The findings were published in the scientific journal Tobacco Control and reported by The Guardian. The study raises concerns about how automated technology can distort public participation in government decisions. Public consultations are intended to help policymakers understand how proposed rules might affect citizens, businesses, and organizations. If automated systems can generate thousands of submissions, relatively small groups may be able to create the appearance of widespread public opposition or support. The researchers described the incident as an early documented example of large-scale automated interference in formal public health policymaking. The investigation does not establish who organized the campaign, and automated submissions should not automatically be attributed to modern large language models. Nevertheless, increasingly capable generative AI could make similar campaigns more convincing and harder to detect. Governments face a difficult balance: stronger identity verification and automated screening may reduce manipulation, but overly restrictive participation requirements could discourage legitimate public contributions. The findings suggest that protecting consultation processes will become an increasingly important part of digital governance. Why It Matters: The research demonstrates how automated content generation can distort public consultation processes and threaten confidence in technology-assisted policymaking. Harmonic expands Aristotle AI into formal verification for C software and Verilog chip designs. AI startup Harmonic has expanded its Aristotle system beyond mathematical research into formal verification for software and semiconductor designs. The company announced Tuesday that Aristotle now supports C programming and Verilog, a language used to describe digital hardware. Formal verification uses mathematical methods to establish whether a system satisfies precisely defined properties. Unlike conventional testing, which examines selected inputs and operating conditions, formal methods can evaluate whether specified requirements hold across all relevant cases within a defined model. Harmonic says its system can propose properties, generate mathematical proofs, and identify assumptions that engineers should review before accepting the results. The development addresses a growing problem as AI systems generate more software code. Faster code production can create new bottlenecks in reviewing, testing, and verifying software behavior. Harmonic reports that Aristotle identified bugs as deep as 110,000 cycles into a digital design developed by Tensordyne. In another demonstration involving the LibTomMath numerical library, the company says the system fixed three bugs and improved modular exponentiation performance by 31% to 38%. These are company-reported results, not evidence that the technology will achieve comparable outcomes across every project. Formal verification also has limits. A mathematical proof establishes that specified properties hold under stated assumptions; it does not guarantee that every possible security or design problem has been identified. Nevertheless, expanding these methods beyond specialized research teams could improve software reliability, semiconductor development, and security testing. Why It Matters: Harmonic's expansion demonstrates how AI could make rigorous software and hardware verification more accessible as automated code generation increases development speed. Frontline Education confirms student data exposed in cyberattack affecting school software systems. Education technology company Frontline Education confirmed that a cyberattack compromised sensitive student information in systems used by American schools. The company disclosed that attackers exploited a vulnerability involving third-party software during a two-day intrusion in August. Frontline initially notified customers about exposure involving employee information, but subsequent investigation identified student records among the affected data. Potentially compromised information includes names, addresses, birth dates, Social Security numbers, government identification details, Medicaid identifiers, and student identification numbers. The specific information exposed varies by individual. Frontline has begun notifying affected people and says it has corrected the underlying vulnerability. The incident matters because Frontline provides administrative software to more than 10,000 educational organizations, supporting systems associated with about 50 million students. However, that customer reach must not be confused with the number of people affected by the breach. The company has not established a complete public count of compromised student and employee records. School districts use centralized platforms to manage staffing, student services, health information, and operational requirements. A vulnerability affecting a shared provider can therefore expose information belonging to multiple institutions. Student information requires particular care because children may not be able to monitor identity misuse or take protective measures independently. Frontline says it is offering monitoring services to eligible affected individuals. School administrators are also reviewing notifications and determining whether additional steps are necessary under applicable privacy laws. The incident reinforces the importance of evaluating cybersecurity protections across third-party software providers serving public institutions. Why It Matters: The Frontline breach illustrates how security weaknesses at a shared education technology provider can expose sensitive student records across multiple school districts. Fintech startup Casap raises $56 million to help banks fight fraudulent payment disputes. Financial technology startup Casap has raised $56 million in Series B funding to expand its AI platform for investigating payment disputes and suspected first-party fraud. Base10 Partners led the financing, with strategic participation from financial services companies Synchrony and Experian. The round brings Casap's total funding to approximately $90 million since its founding in 2023. The startup develops software that helps banks and payment providers examine customer disputes, evaluate evidence, and process chargebacks. Casap says its customers have reduced certain first-party fraud losses by as much as 51% and achieved chargeback success rates of up to 96%. Those figures are company-reported and should not be interpreted as independently verified performance across all financial institutions. The investment addresses a costly challenge facing banks and payment companies. Consumer protection rules require financial institutions to investigate disputed transactions, but handling those investigations can be expensive. Institutions may decide that resolving a small claim in a customer's favor costs less than conducting a detailed review. That creates opportunities for dishonest claims, including cases where customers dispute legitimate purchases. Casap uses AI to organize relevant information and support investigations, potentially reducing the time required to evaluate individual cases. However, automated fraud systems must also protect legitimate consumers. Incorrectly rejecting genuine disputes could create financial harm, regulatory exposure, and reputational problems. For banks, success will depend on the accuracy of decisions, investigation costs, customer outcomes, and compliance with applicable consumer protection requirements. Synchrony and Experian's participation provides strategic relationships that could support commercial adoption, although the company did not disclose a new valuation. Why It Matters: Casap's funding highlights growing demand for specialized financial AI applications that reduce operating costs while helping institutions investigate increasingly complicated payment disputes. Swiss AI biotech startup Atlas Bio raises $12.5 million to simulate clinical trials using human genetic data. Swiss biotechnology startup Atlas Bio has emerged from stealth with $12.5 million in total funding to develop AI models that simulate how patients might respond to experimental medicines. The financing includes a newly announced $10 million seed round led by Creandum and an earlier $2.5 million pre-seed round backed by First Momentum Ventures and LEA Partners. Atlas Bio is developing a foundation model trained on approximately two million RNA measurements collected from human tissue. The company wants to use that biological information to predict differences in treatment response across simulated patient populations. Its initial focus is helping pharmaceutical researchers evaluate drug candidates before advancing into expensive clinical studies. The approach targets one of the pharmaceutical industry's most expensive problems: identifying treatments that appear promising in early research but fail during human testing. Phase II clinical trials assess whether experimental treatments show enough evidence of effectiveness to justify larger studies. Failure at that stage can result in substantial financial losses. Atlas Bio hopes that AI-generated patient simulations will help researchers identify ineffective approaches earlier, improve study designs, and understand which populations may respond to particular treatments. However, its technology remains under development, and simulated biological responses are not equivalent to evidence collected from actual patients. Even highly accurate predictions would require scientific validation before they could influence regulatory decisions or replace established testing requirements. The company also faces challenges involving biological data quality, differences between patient populations, and the complexity of human diseases. The investment reflects growing interest in biotechnology companies developing specialized AI systems rather than general-purpose research assistants. Why It Matters: Atlas Bio's financing shows how investors are backing AI applications intended to improve drug development economics by identifying potential treatment failures before costly clinical studies. Event technology startup Easol secures $30 million and launches AI agents for festival promoters. London-based event technology startup Easol has secured $30 million in additional financing and introduced AI agents designed to automate tasks for festival organizers and experience businesses. The funding consists of $20 million in equity and a $10 million debt facility. Equity investors include Y Combinator, Tiger Global, Foundation Capital, FMZ Ventures, Notion Capital, and Eight Roads, while Viola Credit provided the debt financing. Easol says its platform has supported bookings by approximately 2.8 million guests across more than 135 countries during 2026. The company has also added customers tied to major events, including Coachella, Stagecoach, and Camp Flog Gnaw. Its new AI product, called Teammates, is intended to help businesses handle sales, customer support, and operational activities. The financing illustrates how AI agents are moving into industries where operators often manage complex transactions with relatively small teams. Festival and experience businesses must coordinate ticketing, accommodation, equipment rentals, customer communications, and payment processing. Easol wants to reduce administrative work by giving AI agents access to relevant business information and workflows. The company cited examples involving an agent identifying approximately $50,000 in potential equipment-rental sales and another finding $28,000 in recoverable payments. Those are company-reported examples, not evidence of guaranteed financial results. The distinction between potential revenue and completed transactions matters when evaluating AI automation claims. Easol will also need to manage sensitive customer information, payment permissions, and the reliability of automated communications. For event operators, adoption will depend on whether the technology improves conversion rates, reduces operating expenses, and provides consistent customer service without introducing costly mistakes. Why It Matters: Easol's funding demonstrates growing investor interest in AI agents that automate revenue-generating and administrative workflows for businesses outside traditional enterprise software markets. Spain's Banco Sabadell Partners with Deutsche Telekom to deploy sovereign AI agents across banking operations. Spanish banking group Banco Sabadell has entered into an agreement with T-Systems, Deutsche Telekom's technology services division, to develop infrastructure for deploying AI agents across its operations. Announced Wednesday, the partnership focuses on building a secure, interoperable architecture that lets the bank use advanced AI systems without depending on a single model provider. Banco Sabadell, Spain's fourth-largest banking group, serves approximately seven million customers. The bank wants to expand AI applications across internal operations and customer-facing services while maintaining control over sensitive information and technology decisions. The companies did not disclose the agreement's financial terms or a full deployment timetable. The initiative reflects growing demand for digital sovereignty within Europe's financial services industry. Banks operate under strict requirements governing customer information, operational resilience, cybersecurity, and the outsourcing of essential technology functions. Using AI agents introduces additional challenges because autonomous systems may interact with multiple databases, software applications, and business processes. Banco Sabadell and T-Systems intend to establish technical and governance controls that support broader AI adoption while preserving flexibility to use different models. That approach could help reduce dependence on individual AI suppliers as technology evolves and commercial terms change. However, maintaining architectural independence does not eliminate all vendor dependence. Customers must still evaluate infrastructure providers, software integration requirements, and ongoing security obligations. The partnership also creates an opportunity for European cloud and technology service providers to compete with American hyperscalers for regulated enterprise customers. For banks, the commercial challenge will be translating new AI capabilities into measurable improvements without weakening customer protections or operational control. Why It Matters: Banco Sabadell's agreement demonstrates how European financial institutions are prioritizing security, regulatory control, and supplier independence as they expand enterprise AI adoption. That's your quick tech briefing for today. Follow us on X @TheTechStartups for more real-time updates. - Advertisement -

Blackbird Ventures leads $100m round for general Compute. General Compute, a US-based startup that rents AI computing capacity, has raised US$100 million in a series A round led by Blackbird Ventures and Abstract Ventures. The round values the company at US$800 million post-money, up from US$60 million after its seed round in May. General Compute buys inference chips from Nvidia competitors. It previously struck a US$400 million debt-financed deal to purchase chips from SambaNova. The startup has also added Positron and Cerebras as suppliers, showing its strategy of building computing capacity with alternatives to Nvidia hardware. Stay updated on the go with our mobile app. Get latest insights with smoother, more personalized experience through TIA mobile app. How would you feel if you could no longer use Tech in Asia? Share, tag us, and land on our Wall of!