It’s Friday, July 17, 2026, and the global technology race is entering a more consequential phase. The AI arms race has entered a sharper, more consequential phase. In the last 24 hours, a Chinese lab released one of the largest open frontier models yet, Apple launched a major lawsuit accusing OpenAI of stealing hardware trade secrets, Google fell further behind on its flagship model, and Europe moved to rein in what it calls addictive design at Meta. Meanwhile, the infrastructure powering all of this — from memory chips to data centers — is attracting record capital and regulatory attention across continents.

Today’s biggest stories capture that shift from every angle. China’s Moonshot AI is challenging leading U.S. models with Kimi K3; Meta is recruiting senior AWS talent as it weighs a larger cloud push; and Google is facing fresh scrutiny over delays to Gemini 3.5 Pro. At the same time, billions of dollars are pouring into AI inference and robotics, cyberattacks are disrupting real-world production, and regulators are moving against some of the technology’s most harmful uses. Here are the 15 stories shaping tech, startups, and AI today.

Here are the top technology news stories that matter most right now

Technology News Today

Google Falls Months Behind Schedule on Gemini 3.5 Pro as It Refines Coding and Agentic Capabilities

Alphabet’s Google has reportedly delayed the broader release of Gemini 3.5 Pro, its flagship frontier AI model, by several months after internal testing revealed the system fell short of the company’s expectations in coding performance and complex, long-horizon reasoning tasks. The model, which was previewed at Google I/O earlier in 2026 and had been expected to launch around June, remains in limited enterprise preview as engineers work to improve its capabilities. The delay underscores the intensifying competition among leading AI developers as Google seeks to position Gemini 3.5 Pro against frontier models from OpenAI and Anthropic. Shares of Alphabet declined following the report amid broader investor concerns over the company’s AI execution.

Alphabet shares fell more than 4% after reports of the delay surfaced, reflecting growing investor concern about Google’s position in the frontier-model race. The company is competing against increasingly capable systems from OpenAI, Anthropic, Meta, SpaceX, and Chinese developers, including Moonshot AI.

A delayed model release does not necessarily mean Google has lost its technical position. The company still controls a formidable AI stack spanning custom Tensor Processing Units, Google Cloud, DeepMind, Android, Search, YouTube, Workspace, and billions of consumer accounts. However, that scale creates its own challenge: Google must improve its models while ensuring they work reliably across the products people and businesses use every day.

The timing is especially difficult because competitors are releasing models at a faster pace. Kimi K3’s strong benchmark results this week further increased pressure on Google to demonstrate that Gemini can remain among the industry’s leading systems.

Why It Matters: Gemini’s delay shows that even Google’s vast computing resources cannot guarantee predictable progress at the frontier of AI development.

Source: Bloomberg.

Apple Overtakes Nvidia as the World’s Most Valuable Company

Apple surpassed Nvidia in market capitalization on Friday, reclaiming the position of the world’s most valuable publicly traded company. The shift followed a strong advance in Apple shares and a pullback across several semiconductor stocks after an extended AI-driven rally.

Apple’s market value is approaching the $5 trillion threshold, reflecting renewed investor confidence in its hardware ecosystem and AI strategy. Nvidia had taken the top position as demand for graphics processors pushed its valuation to unprecedented levels, making it the clearest market winner from the buildout of generative AI infrastructure.

The change does not necessarily signal a collapse in demand for Nvidia’s chips. Instead, it shows how quickly market leadership can shift as investors reassess the returns from data center spending and the value of companies that own direct consumer distribution.

Apple controls a global installed base of iPhones, Macs, watches, and services. Investors are betting that it can turn those devices into a major AI distribution network, even as the company faces delays, regulatory disputes, and a trade-secret lawsuit involving OpenAI.

Why It Matters: Apple’s return to the top shows that control of consumer devices may be valued as highly as control of the chips supporting the AI boom.

Source: Reuters.

China’s Moonshot AI Launches Kimi K3, Challenging Western Frontier Models

Beijing-based Moonshot AI has released Kimi K3, a new open-weight model that immediately ranks among the strongest AI systems available for coding and agent-based tasks. The Alibaba-backed startup says Kimi K3 was built for long-horizon workflows, including projects that require an AI system to plan, write code, test its work, and make corrections across multiple steps.

The model reached the top position on Arena.ai’s Frontend Code Arena, recording a 76% pairwise win rate in head-to-head tests. It finished ahead of Anthropic’s Claude Fable 5 and OpenAI’s GPT-5.6 Sol on that benchmark. Kimi K3 also scored 88.3 on Terminal Bench 2.1, narrowly trailing GPT-5.6 Sol’s 88.8. Its broader Text Arena performance placed it ninth overall, a major improvement from the previous Kimi generation.

The launch is another sign that Chinese AI labs are closing the capability gap with leading U.S. developers. Open-weight availability could make Kimi especially attractive to companies that want to run models on their own infrastructure, modify them for specialized tasks, or avoid dependence on a single American provider.

Why It Matters: Kimi K3 strengthens China’s position in frontier AI while giving global developers another high-performing open-weight alternative.

Source: Fundpluse via Moonshot AI.

Meta Hires Senior AWS Executive as Its AI Infrastructure Ambitions Grow

Meta is hiring Dave Brown, one of Amazon Web Services’ most senior computing executives, as the Facebook parent accelerates its data center expansion and considers a larger role in cloud infrastructure. Brown spent nearly two decades at Amazon and most recently held responsibility for major parts of AWS’s compute, AI, and platform operations.

Brown is expected to report to Meta’s infrastructure chief, Santosh Janardhan, and help oversee the company’s growing network of AI data centers. The appointment comes as Meta plans to spend between $125 billion and $145 billion in capital this year, with much of that money directed toward servers, networking systems, power capacity, and facilities for training and operating AI models.

The recruitment also raises questions about whether Meta could eventually offer AI computing services to outside companies. Chief Executive Mark Zuckerberg has acknowledged interest from businesses seeking access to Meta’s models and excess computing capacity. A commercial cloud offering would place Meta in more direct competition with Amazon, Microsoft, Google, and Oracle.

The move shows how the AI race is expanding beyond model developers. Control over chips, data centers, energy, networking, and cloud distribution is becoming just as important as the underlying algorithms.

Why It Matters: Meta’s latest hire suggests it is building the leadership and infrastructure needed to become a larger force in AI cloud computing.

Source: The Wall Street Journal.

Microsoft Prepares Multi-Model AI Security Tool to Compete With Anthropic

Microsoft is preparing to release an AI cybersecurity product internally known as Project Perception, according to people familiar with the company’s plans. The system is expected to use models from Microsoft, OpenAI, and Anthropic, selecting among them based on the security task being performed.

The product would give customers access to advanced AI security capabilities without requiring them to commit to one model provider. A routing layer could send vulnerability analysis, code review, threat detection, and incident response work to the model that offers the best balance of performance, speed, and cost. Microsoft reportedly plans to introduce the product this month.

Project Perception appears aimed partly at Anthropic’s growing influence in cybersecurity. Anthropic’s most capable models have demonstrated advanced abilities in finding software flaws, analyzing malicious code, and assisting with defensive security operations. Those capabilities have attracted government attention but can be expensive for companies to deploy at scale.

Microsoft has a major distribution advantage through Azure, Windows, GitHub, Microsoft 365, Defender, and its enterprise sales organization. By supporting several model families, it can position itself as the platform that manages security workflows regardless of which AI developer leads individual benchmarks.

Why It Matters: Microsoft is turning model competition into a platform opportunity by offering businesses one AI security layer across several leading providers.

Source: The Information.

AI Backlash Forces Tech Companies to Increase Executive Security

Leading AI companies are increasing security for executives and employees following threats, attempted attacks, and growing public anger over automation, job losses, data centers, and the social effects of artificial intelligence.

Recent incidents reportedly include an attempted firebombing at OpenAI Chief Executive Sam Altman’s home and a person entering Anthropic’s offices to warn of a possible assassination threat. AI companies have also documented violent online posts and direct threats aimed at executives and staff.

The growing hostility is forcing companies to rethink public events, office security, employee travel, and corporate branding. Some executives now travel with bodyguards, while workers have been advised against wearing company logos in public. Palantir, Oracle, Salesforce, and other technology companies have increased spending on executive protection.

The backlash reflects fears that AI will eliminate jobs, concentrate wealth, weaken creative industries, and raise household energy costs through data center construction. While threats and violence are indefensible, the anger signals a widening trust gap between technology companies and the public.

AI developers have largely framed adoption in terms of productivity and economic growth. The security concerns suggest that companies will also need clearer answers about employment, accountability, community costs, and who benefits financially from automation.

Why It Matters: Rising threats against AI leaders reveal that public opposition is becoming a material operational and reputational risk for the technology industry.

Source: The Wall Street Journal.

Xi Jinping Champions Open-Source AI and Support for Global South at World AI Conference

At China’s World AI Conference, President Xi Jinping promoted open-source AI development, pledged assistance to developing nations in building AI capabilities, and described unequal access to AI as an “injustice.” The remarks position China as a leader in inclusive global AI governance amid ongoing U.S.-China tech tensions.

The speech underscores China’s strategic push to shape international AI norms and expand influence through technology sharing. It has implications for global standards, export controls, and collaboration, potentially affecting how Western firms and governments approach partnerships and regulation in emerging markets.

Why It Matters: Xi’s comments signal China’s intent to lead in open and accessible AI frameworks, influencing global policy debates and opportunities for cross-border tech cooperation or competition.

Source: Fundpluse via News.cn, CNBC, Reuters.

OpenAI Launches GPT-Live Full-Duplex Voice Models for Real-Time, Natural Conversations

OpenAI rolled out GPT-Live voice models, enabling simultaneous listening and speaking in ChatGPT, delivering more fluid, human-like interactions without the previous turn-based limitations. The release advances voice AI toward more natural, real-time applications in assistants, customer service, and accessibility tools. It positions OpenAI competitively in the growing voice interface market.

Why It Matters: GPT-Live marks meaningful progress toward seamless voice AI, with broad implications for how people interact with technology in daily life and enterprise settings.

Source: Reuters.

UK Robotics Startup Humanoid Raises $150 Million at a $1.2 Billion Valuation

London-based robotics startup Humanoid has raised $150 million in the first tranche of a Series A round, valuing the company at $1.2 billion prior to the investment. The startup is reportedly seeking another $80 million to $100 million by September, which would make the financing one of Europe’s largest early-stage robotics rounds.

Humanoid is developing general-purpose humanoid robots for commercial and industrial work. The company joins a crowded field that includes Tesla, Figure AI, Apptronik, Agility Robotics, 1X, and several Chinese manufacturers attempting to move humanoid machines from laboratory demonstrations into warehouses and factories.

Investor interest in humanoid robotics has surged as advances in vision models, reinforcement learning, simulation, and robotic foundation models improve machines’ ability to learn physical tasks. Labor shortages and rising manufacturing costs are also pushing companies to investigate automation for jobs that remain difficult to perform with conventional fixed-purpose robots.

The central challenge is no longer building a machine that can walk across a stage. Robotics companies must show that their systems can operate safely for long periods, perform useful work consistently, and generate savings that justify their purchase and maintenance costs.

Why It Matters: Humanoid’s financing shows that investors increasingly view general-purpose robotics as one of AI’s largest potential markets beyond software.

Source: The Information.

Apple Issues Legal Warnings to Approximately 40 Former Employees Now Working at OpenAI

Apple has sent personal legal warnings to dozens of ex-employees at OpenAI, instructing them to preserve documents and to prepare to meet with company lawyers amid the escalating trade-secrets dispute. The warnings escalate Apple’s efforts to protect intellectual property as OpenAI builds its hardware team. They reflect heightened caution around employee transitions between competing tech giants.

Why It Matters: Apple’s aggressive legal posture toward former OpenAI staff underscores the high stakes of talent mobility and IP protection in the converging AI and hardware sectors.

Source: Financial Times.

Suno Hack Exposes Source Code and Details of AI Music Training Data

AI music startup Suno suffered a security breach that exposed the company’s source code and information showing how the platform gathered music, lyrics, and audio for its training systems. A hacker provided 404 Media with files that reportedly document scraping from YouTube Music, Deezer, Genius, podcast feeds, stock-music libraries, and other online sources.

The stolen material also allegedly included some customer contact information and payment-related data. Suno said the breach originated in November and maintained that sensitive user information was not compromised, but the newly disclosed files have intensified questions about the scope of the incident and why users were not notified.

The breach could have consequences beyond cybersecurity. Record labels and artists have accused generative AI companies of using copyrighted music without permission or compensation. Internal source code and data-collection records could become important evidence in current or future litigation over how music-generation models were trained.

Suno has become one of the most prominent AI music platforms, allowing users to generate complete songs from text prompts. Its growth has made it a test case for whether generative music can develop into a licensed creative industry or remain locked in costly disputes with rights holders.

Why It Matters: The Suno breach connects two major AI risks—weak data security and unresolved questions about the use of copyrighted training material.

Source: 404 Media.

Google Adds Instacart, Canva, and YouTube Music Integrations to AI Mode

Google is expanding AI Mode in Search by allowing U.S. users to connect selected third-party apps, including Instacart, Canva, and YouTube Music. Once an app is linked, users can ask Google’s AI to retrieve information or initiate supported actions without leaving the search experience.

The integrations could allow someone to find a recipe and add ingredients to an Instacart shopping workflow, create visual materials in Canva, or surface music from YouTube Music. Google says users will control which applications are connected and can remove access through their account settings.

The update represents a shift in Search from answering questions to coordinating tasks. Google is attempting to make AI Mode an operating layer that sits between users and a growing range of online services. That puts it in competition with standalone AI assistants and agent platforms being developed by OpenAI, Microsoft, Amazon, and startups.

For app developers, integration with Google could provide valuable distribution. It could also create new dependence on a platform that already controls a large share of online discovery and advertising. Regulators will likely watch whether Google gives its own services preferential placement over competing applications.

Why It Matters: Google is turning AI Search into an action platform that could reshape how consumers discover and use online services.

Source: TechCrunch.

Anthropic Pushes States to Adopt Stronger Rules for Frontier AI

Anthropic is supporting tougher regulation of advanced AI developers, arguing that existing state transparency laws may not be sufficient to address the risks posed by increasingly capable models. The company is calling for independent safety audits and stronger government authority to examine systems that could threaten critical infrastructure or public safety.

Anthropic’s preferred approach would focus obligations on the largest AI laboratories rather than smaller startups. Company representatives have said the most powerful developers should face additional scrutiny because their models require enormous financial and computing resources and can produce capabilities unavailable in ordinary software.

The company’s position has drawn criticism from advocates of lighter regulation, who argue that large incumbents could use compliance requirements to raise competitors’ costs. Anthropic maintains that its proposals would apply only to companies above substantial investment or revenue thresholds.

The debate reflects a broader split within the AI industry. Some executives warn that restrictive state rules could fragment the U.S. market, while others believe federal action is moving too slowly to address frontier-model risks. States, including California and New York, have increasingly stepped into that policy gap.

Why It Matters: Anthropic’s campaign could influence whether U.S. AI oversight is shaped by state governments, federal agencies, or the industry’s largest developers.

Source: WIRED.

San Francisco Orders Apple and Google to Remove AI “Nudify” Apps

San Francisco’s City Attorney has sent cease-and-desist letters to Apple and Google demanding the removal of 13 AI-powered face-swap and “nudify” applications from their app stores. Officials say the apps are overwhelmingly used to create nonconsensual sexual images targeting women and girls.

The city argues that Apple and Google are profiting from applications that facilitate image-based abuse, despite app-store rules that prohibit harmful, deceptive, or sexually exploitative content. The letters seek immediate removal of the named services and information about the companies’ review, ranking, and revenue-sharing practices.

Generative AI has made it easier to create convincing fake images with limited technical skill and few source photographs. Victims can face harassment, reputational damage, and difficulties getting fabricated material removed once it spreads across social platforms and messaging services.

The action also targets the distribution layer rather than pursuing individual developers one at a time. Apple and Google operate the primary gateways for mobile software and can restrict access to millions of users. Regulators are increasingly arguing that this control brings responsibility for policing applications that enable predictable harm.

Why It Matters: The case could establish stronger legal expectations for app stores that distribute AI tools used to create nonconsensual sexual imagery.

Source: WIRED.

Fairlife Halts U.S. Production After Cyberattack Disrupts Systems

Coca-Cola-owned dairy company Fairlife has temporarily stopped production in the United States after an unauthorized third party gained access to parts of its computer environment. The affected systems included technology connected to manufacturing operations, prompting the company to suspend production while investigators assess and restore its network.

Fairlife said it activated incident-response and business-continuity procedures, brought in outside cybersecurity specialists, and notified law enforcement. The company said there was no indication that the quality or safety of its products had been compromised. Canadian operations were unaffected.

The disruption shows how cyberattacks can extend beyond stolen records and interrupted office systems to affect physical production. Modern manufacturing plants depend on connected software for scheduling, inventory, quality control, machinery, logistics, and supplier coordination. When those systems are compromised, companies may stop operations even when production equipment itself has not been damaged.

Food and beverage businesses face additional pressure because any uncertainty involving production systems can quickly become a consumer-safety and supply-chain concern. Extended downtime could affect inventories, retailers, suppliers, and transportation partners.

Why It Matters: Fairlife’s shutdown demonstrates how a cyber intrusion can disrupt physical manufacturing and create immediate supply chain consequences.

Source: Reuters.

AI Infrastructure Startup Fireworks Raises $1.5 Billion at a $17.5 Billion Valuation

Fireworks AI has raised $1.5 billion in a Series D financing round, valuing the inference infrastructure startup at $17.5 billion. The deal places Fireworks among the most highly valued private companies supplying the software layer that runs AI models in production.

Fireworks helps developers deploy and operate open and proprietary models while managing speed, reliability, hardware utilization, and inference costs. These functions have become increasingly important as companies move from experimenting with generative AI to serving large numbers of paying customers.

Training frontier models attracts much of the industry’s attention, but inference—the process of generating answers and executing tasks after a model has been trained—can become the larger recurring expense. Startups that reduce those operating costs could capture a meaningful share of enterprise AI spending without having to build their own frontier models.

The valuation reflects investor confidence that enterprises will use multiple AI models rather than relying on a single vendor. In that environment, independent infrastructure providers can help customers route workloads across chips, clouds, and model families while avoiding technical lock-in.

Why It Matters: Fireworks’ financing highlights the growing value of the infrastructure required to run AI economically at production scale.

Source: Axios.

Walden Robotics Raises $300 Million as Physical AI Funding Accelerates

Walden Robotics has raised $300 million in new funding as investors continue to back startups developing intelligent machines for warehouses, factories, logistics, and other physical environments. The financing arrives amid a wider surge of capital into robotics companies attempting to combine modern AI models with increasingly capable hardware.

Details of Walden’s commercial rollout remain limited, but the size of the round signals significant investor expectations. Robotics startups typically require more capital than software companies because they must develop mechanical systems, purchase components, build prototypes, collect real-world training data, and establish manufacturing and maintenance operations.

Advances in multimodal AI are giving robots better ways to interpret visual information, follow natural-language instructions, and adapt to changes in their surroundings. However, physical deployment remains difficult. A robot that succeeds in a controlled demonstration must still prove it can work safely and reliably through thousands of hours in unpredictable environments.

Funding is focusing on companies that can connect AI to commercially useful machines. The eventual winners will need more than impressive models; they will require dependable hardware, affordable manufacturing, customer support, and measurable economic returns.

Why It Matters: Walden’s round shows that venture capital is increasingly moving from AI software into machines capable of performing work in the physical economy.

Source: Axios.

German Robotics Startup Microagi Raises $55 Million to Train Factory Robots

Munich-based robotics startup microagi has raised $55 million in what its backers describe as the largest seed financing completed by a German startup. The company was founded by former Red Bull Formula One engineer Bercan Kilic and is building AI systems intended to help existing industrial robots learn a wider variety of factory tasks.

Microagi’s strategy centers on training data. Its separate data-collection operation, called Shift, pays people to record themselves performing physical activities using cameras and sensor-equipped gloves. The business reportedly operates in 15 countries and has recruited more than 20,000 participants.

The resulting footage and motion data can be used to teach robotic models how humans manipulate objects, use tools, and complete multi-step jobs. This addresses one of the largest obstacles in physical AI: internet-scale text and image datasets exist for language models, but high-quality data describing real-world movement is much harder to collect.

Microagi plans to apply the technology across automotive manufacturing, logistics, food production, and other industries facing labor shortages. Its approach could also help factories improve existing robotic equipment rather than replacing every machine with a new humanoid platform.

Why It Matters: Microagi is targeting the data bottleneck that must be solved before adaptable factory robots can be deployed at large scale.

Source: Business Insider.

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