Top Tech News Today, August 13, 2026: Anthropic, DeepMind, Google, Lenovo, Microsoft, SpaceXAI & More

It’s Thursday, August 13, 2026, and the biggest technology story today isn’t a single product launch or funding round. It’s the sheer scale of the infrastructure, capital, and security systems now forming around AI. A data center company is weighing a $100 billion IPO, AI startups are racing toward $40 billion valuations, India is landing one of its largest GPU deployments, and companies are experimenting with everything from offshore data centers to satellites built to watch other satellites in orbit.

At the same time, the risks are becoming harder to separate from the opportunity. Microsoft is recalibrating its presence in China, AI leaders are debating new oversight structures, and cyberattacks are spilling from corporate networks into warehouses and global supply chains. Meanwhile, Google is pushing Gemini deeper into smartphones, autonomous-driving startups are expanding across borders, and the race for faster storage, cheaper compute, and reliable energy is intensifying.

The common thread: AI is no longer just changing software. It is reshaping the physical, financial, geopolitical, and security foundations of the technology industry itself. Here are the top tech stories making waves today.

Microsoft Pulls Back in China While AI Keeps a Strategic Door Open

Microsoft has steadily reduced its physical presence in China after years of geopolitical tension, tighter regulation, and technology export controls, but the company has stopped short of abandoning the market entirely. Reuters reports that Microsoft has closed or withdrawn from at least 15 branches and joint ventures in recent years, while China’s contribution to the company’s worldwide revenue had fallen to about 1.5% by 2024. The company reportedly considered a more complete exit in 2023 before deciding to maintain selected operations.

What remains is increasingly tied to AI, cloud services, and Chinese companies expanding overseas. Microsoft has relocated some top researchers and developed alternative research hubs outside China as U.S. restrictions complicate advanced AI work and access to high-end computing. Yet Azure and Microsoft’s AI tools can still serve Chinese businesses pursuing customers and operations abroad, giving the company a commercial reason to preserve a carefully limited presence. The shift illustrates the difficult position facing U.S. technology companies: China remains a massive source of engineering talent and corporate demand, but national-security concerns and diverging technology regimes are making deep integration harder. Microsoft’s strategy increasingly looks like controlled exposure rather than the broad market expansion envisioned a decade ago.

Why It Matters: Microsoft’s shrinking China footprint shows how geopolitics is redrawing Big Tech’s operating map even as AI and cloud demand make a complete separation economically difficult.

Source: Reuters.

Google Unveils Pixel 11 Series with Tensor G6 and Expanded Gemini AI Features

Google launched its Pixel 11, Pixel 11 Pro, and Pixel 11 Pro XL smartphones on August 12, powered by the new Tensor G6 chip and featuring major camera upgrades, thinner designs, and deeper integration of Gemini AI tools such as Magic Capture for selecting optimal frames. The devices start at $899 for the base model with 256GB storage, up from previous base capacities, and promise seven years of software support. Pre-orders began immediately, with availability set for August 20. The lineup also includes a Pixel Tag competitor to AirTag and updates to the Pixel Watch 5. These phones emphasize on-device AI processing for features like sign-language-to-text translation and proactive assistance. In the broader ecosystem, the release intensifies competition in the Android flagship space against Samsung and Chinese rivals, while showcasing Google’s strategy to differentiate through AI rather than radical hardware redesigns. It matters for startups building on Gemini APIs and for the consumer hardware market as AI becomes a primary selling point in mid-to-premium devices.

Why It Matters: The Pixel 11 solidifies Google’s position as an AI-first hardware player, pressuring rivals and expanding opportunities for developers in on-device intelligence.

Source: TechCrunch.

SpaceXAI Releases Grok 4.6 Model Matching Top Frontier Benchmarks

SpaceXAI (formerly xAI) launched Grok 4.6 on August 12, a new flagship model optimized for long-running agents, coding, and multi-step interactive tasks. It matches GPT-5.6 Sol on the Artificial Analysis Intelligence Index and shows gains over Grok 4.5 across coding and knowledge benchmarks. Pricing remains competitive at $2 per million input tokens and $6 per million output tokens, with a faster variant available at double the cost.

The model is accessible via Cursor, Grok Build, the API, and partners including OpenRouter. Training involved extended runs with high-quality engineering data and improved optimizers. This release accelerates the agentic AI race among major labs, offering developers stronger tools for complex workflows at lower relative costs. For startups, it expands options beyond closed models from OpenAI and Anthropic, potentially lowering barriers for agent-based applications in enterprise software.

Why It Matters: Grok 4.6 strengthens open competition in frontier AI, giving startups affordable access to high-capability agents for production use.

Source: SiliconANGLE.

AI Data Center Giant Vantage Weighs $100 Billion IPO as Compute Boom Reshapes Infrastructure

Vantage Data Centers is exploring strategic options that could include an initial public offering valuing the company at roughly $100 billion, potentially turning the hyperscale infrastructure operator into one of the largest publicly traded pure-play data center companies. Reuters reported that Vantage, backed by Silver Lake and DigitalBridge, has held preliminary discussions with advisers about an IPO that could raise around $10 billion, although a sale is also among the options being considered. Any transaction remains at an early stage and could happen as soon as next year.

The numbers show how dramatically AI has changed the economics of physical computing infrastructure. Vantage has raised roughly $11 billion since late 2023 as cloud providers and AI companies race to secure electricity, land, cooling capacity, and high-density server space. The company is also involved in a Wisconsin data center campus connected to the OpenAI and Oracle Stargate buildout. A $100 billion valuation would signal that investors increasingly see data centers less as conventional real estate and more as critical technology infrastructure whose value is tied directly to AI demand. It would also provide a new public-market benchmark for a sector absorbing extraordinary amounts of private capital.

Why It Matters: A Vantage IPO near $100 billion would show that the AI investment boom is creating enormous value well beyond chips and models, extending deep into the physical infrastructure required to run them.

Source: Reuters.

Anthropic in Talks to Acquire Decart AI for About $6 Billion

Anthropic is negotiating to buy Israeli startup Decart AI, which specializes in real-time generative video, world models for simulated environments, and GPU optimization technology, in a deal valued around $6 billion. Reports emerged late on August 12, noting the talks remain early-stage and could still fall through. Decart’s tech could help Anthropic improve inference efficiency and absorb rising demand ahead of a potential IPO. The startup previously raised $300 million with Nvidia as an investor.

If completed, Decart’s team would integrate into Anthropic’s inference organization. This would mark one of Anthropic’s largest acquisitions and signal consolidation in specialized AI infrastructure as compute costs soar. For the startup ecosystem, it highlights the premium placed on efficiency tools that reduce training and inference expenses.

Why It Matters: The potential deal underscores how infrastructure optimization has become critical for AI labs scaling amid intense compute demand.

Source: Bloomberg.

Legal AI Startup Legora Seeks Funding at More Than $10 Billion Valuation

Swedish legal AI startup Legora is seeking fresh capital at a valuation above $10 billion, according to the Financial Times, barely four months after being valued at roughly $5.6 billion. Founded in 2023, Legora builds AI software for lawyers that assists with document review, drafting, due diligence, and regulatory work. Its customer roster includes major organizations such as Linklaters, Deloitte, and Heineken, placing it squarely in one of the fastest-growing enterprise AI categories.

The valuation discussions underline how quickly investors are repricing startups that can turn generative AI into software businesses with measurable professional-services demand. Legora’s annual recurring revenue reportedly rose about 50% to roughly $150 million during the second quarter, while the company plans to expand its workforce substantially. Rival Harvey has also been linked to funding discussions at a valuation near $15 billion, suggesting that legal AI is developing into a major competitive market rather than a niche application. The opportunity is significant because law firms spend heavily on skilled labor, research, document processing, and compliance. AI tools capable of reducing time spent on those tasks can command enterprise budgets that are far larger than typical consumer subscriptions.

Why It Matters: Legora’s potential $10 billion-plus valuation shows investors are betting that specialized AI software for high-cost professional work can become one of the strongest enterprise application markets.

Source: Financial Times.

DeepMind’s Demis Hassabis Pushes Independent AI Oversight Body for AGI Safety

Google DeepMind leader Demis Hassabis has discussed creating an independent industry body that would establish common AI safety practices as developers move closer to more capable artificial intelligence systems. The Wall Street Journal reported that Hassabis has raised the proposal with executives at major AI labs and U.S. officials, including Treasury Secretary Scott Bessent and White House technology adviser Michael Kratsios. The envisioned organization would attempt to codify safety guardrails and shared practices around advanced AI development.

Hassabis has reportedly compared the idea conceptually with institutions such as the International Atomic Energy Agency, although overseeing software developed by competing private companies would pose very different technical and political challenges. The proposal arrives as frontier AI companies face pressure to move quickly while governments struggle to establish regulatory frameworks that can keep pace with model capabilities. An independent technical organization could potentially create standards that sit between voluntary company policies and formal government regulation. The unresolved question is whether competing labs would accept meaningful external oversight, especially when safety restrictions could influence development speed, product launches, or access to valuable models. The effort also reflects a broader shift in the AI debate from hypothetical long-term risks toward practical questions about who sets and verifies safety rules.

Why It Matters: A credible independent AI safety institution could become an important layer between self-regulation and government mandates as frontier models become more capable and economically important.

Source: Wall Street Journal.

OpenAI-Backed Startup Thrive Holdings Raises $2 Billion at $12 Billion Valuation

Thrive Holdings, an OpenAI-backed company focused on applying AI to traditional businesses, has raised $2 billion in new funding at a $12 billion valuation. TechCrunch reported that the financing includes participation from SoftBank, D1 Capital Partners, and Altimeter Capital, giving Thrive substantial capital to continue acquiring and building businesses where AI can change how services are delivered.

Thrive’s model represents a different branch of the AI investment boom. Instead of selling another general-purpose model or standalone chatbot, the company is using AI as an operating layer inside established industries. That thesis reflects growing investor interest in businesses that combine software economics with large service markets where labor remains a major cost. Accounting, IT services, compliance, and other operational categories can offer enormous addressable markets if AI meaningfully increases productivity without sacrificing reliability. The $12 billion valuation also shows how capital is moving beyond the foundation-model companies themselves toward businesses capable of translating AI capabilities into industry-specific revenue. The harder part will be execution: integrating AI into existing workflows, businesses, and customer relationships can be slower and more operationally demanding than shipping pure software.

Why It Matters: Thrive’s $2 billion raise reflects a growing investor thesis that some of AI’s largest companies may emerge by rebuilding traditional service industries rather than creating new foundation models.

Source: TechCrunch.

AI Coding Startup Cognition Discusses Funding at $40 Billion-Plus Valuation

AI coding startup Cognition is in early discussions with investors about raising another round at a valuation of at least $40 billion, according to Bloomberg reporting cited by multiple outlets. Such a valuation would represent an increase of more than 50% from its recent level and comes as the company behind the Devin coding agent reportedly approaches an annualized revenue run rate near $1 billion. The discussions remain preliminary, meaning the final terms could change or no transaction could ultimately occur.

Cognition’s trajectory captures the extraordinary amount of money flowing into AI developer tools. Coding has emerged as one of the clearest commercial uses for generative AI because software development produces structured work that can be measured by completed tasks, fixed bugs, shipped features, and developer time saved. The market is becoming increasingly crowded, however, with model providers and startups competing through coding agents, IDE integrations, command-line tools, and autonomous software engineering systems. A $40 billion valuation would set an exceptionally high expectation for future revenue growth and defensibility. It would also suggest investors believe coding agents can capture a meaningful portion of the hundreds of billions of dollars businesses spend each year on software engineering rather than remaining supplemental developer assistants.

Why It Matters: Cognition’s potential valuation shows that AI coding has moved from an experimental productivity feature into one of the most aggressively financed software markets.

Source: Bloomberg.

Japanese Self-Driving Startup Turing Eyes U.S. Expansion and Possible $10 Billion IPO

Tokyo-based autonomous-driving startup Turing plans to establish a U.S. office within the next year as it expands research and begins positioning itself for a possible American market entry. The company could eventually pursue a U.S. public listing within roughly five years at a valuation around $10 billion, according to the Wall Street Journal. Founded in 2021, Turing is developing autonomous-driving software and has relationships with Japanese automotive groups including Subaru and Denso.

The U.S. move matters because autonomous driving is increasingly becoming a global contest involving automakers, chipmakers, AI companies, and highly capitalized startups. Turing is targeting mass production of advanced driver-assistance technology by 2030 and has been testing AMD processors as it explores alternatives to Nvidia hardware. AMD Ventures is also an investor. Expanding into the United States could give Turing access to a deeper pool of AI engineers, investors, computing resources, and automotive partners, but it would also place the startup closer to formidable competitors including Waymo and Tesla. Autonomous driving remains one of AI’s most technically demanding commercial applications because mistakes happen in the physical world, making validation, regulation, hardware reliability, and enormous training datasets crucial.

Why It Matters: Turing’s U.S. expansion shows that the autonomous-driving race is widening beyond American and Chinese leaders as Japanese startups seek a larger role in AI-powered mobility.

Source: Wall Street Journal.

Google Pixel 11 Puts Proactive Gemini AI at the Center of Its Smartphone Strategy

Google unveiled its Pixel 11 smartphone family with a stronger focus on proactive Gemini AI, upgraded cameras, and its new Tensor G6 processor. The company says the latest devices are intended to make Pixel more personal by allowing Gemini-powered features to anticipate useful actions and surface relevant assistance across everyday tasks. Preorders opened with pricing beginning at $899, as Google continues moving the Pixel line further into the premium smartphone segment.

The bigger story is Google’s attempt to turn AI into a reason to upgrade hardware rather than merely another app installed on a phone. Smartphones already contain years of user context through photos, messages, calendars, location history, and communications, giving on-device and tightly integrated AI a potential advantage over standalone assistants. Google’s ownership of Android, Gemini, Tensor silicon, and Pixel hardware allows it to integrate those layers more tightly than most competitors. That also increases pressure on Apple and Samsung to prove that their own AI features can deliver tangible benefits rather than feature checklists. As AI models become widely accessible, smartphone competition may increasingly shift toward who can combine models, operating systems, chips, and personal context most effectively while maintaining security and privacy.

Why It Matters: Pixel 11 shows Google betting that the next smartphone upgrade cycle will be driven less by raw hardware improvements and more by deeply integrated personal AI.

Source: Google.

Lenovo Posts Record $26.9 Billion Quarter as AI Revenue Jumps 60%

Lenovo reported its strongest first quarter on record, with quarterly revenue reaching $26.9 billion, up 43% year over year. The company said all of its major business groups recorded first-quarter highs for both revenue and operating profit, while revenue tied to AI-related products and services jumped approximately 60%. AI now accounts for roughly 35% of Lenovo’s total revenue, according to figures released alongside the results.

The performance illustrates how the AI spending cycle is spreading across the hardware industry. Lenovo remains best known as a PC manufacturer, but its exposure increasingly spans AI PCs, servers, infrastructure, and enterprise technology. That diversification gives the company several ways to participate as businesses upgrade both employee devices and back-end computing systems. It also suggests that the AI hardware cycle is broadening beyond Nvidia’s GPUs and the hyperscale data centers that have dominated investor attention. PC manufacturers are betting that local AI processing will eventually become a standard feature across business and consumer devices, while enterprises are simultaneously adding server capacity for larger workloads. Lenovo still has to prove how durable that demand will be after the initial upgrade wave, but the latest quarter indicates customers are spending now.

Why It Matters: Lenovo’s record quarter offers fresh evidence that AI spending is reaching PCs, servers, and enterprise hardware rather than remaining concentrated among a handful of cloud giants.

Source: Lenovo Investor Relations.

India Lands Major AI Infrastructure Build as L&T Wins Together AI Order Worth Up to $1.57 Billion

Indian engineering conglomerate Larsen & Toubro has secured an AI infrastructure contract valued at roughly ₹10,000 crore to ₹15,000 crore, or as much as about $1.57 billion, tied to a large deployment for U.S.-based AI company Together AI. The project will involve around 10,000 Nvidia B300 GPUs at an L&T data center campus in Chennai and is being positioned as India’s largest single-cluster AI computing installation.

The project gives India something policymakers and technology companies have increasingly argued it needs: large domestic computing capacity capable of supporting advanced AI training and inference. India’s enormous developer base and fast-growing digital economy have made the country a major consumer of cloud services, but frontier AI infrastructure remains heavily concentrated in the United States and a handful of other markets. Building a cluster at this scale could support local startups, enterprises, researchers, and international AI companies seeking additional capacity. It also shows how Nvidia’s newest accelerator generations are driving infrastructure spending far beyond U.S. hyperscalers. For L&T, traditionally associated with engineering and industrial projects, AI data centers represent another convergence between heavy infrastructure and the technology sector as electricity, cooling, networking, construction, and semiconductor supply become central parts of the AI economy.

Why It Matters: A 10,000-GPU deployment in Chennai could materially expand India’s domestic AI capacity and deepen its role in the global compute supply chain.

Source: The Economic Times.

AI Storage Race Accelerates as Kioxia and Sandisk Unveil 2Tb QLC 3D NAND

Kioxia and Sandisk have unveiled a new ninth-generation 2-terabit QLC 3D flash memory technology aimed partly at AI infrastructure and other high-capacity computing workloads. The companies say the technology uses their CMOS directly bonded to Array architecture and a six-plane design to increase parallelism, while a NAND interface operating at up to 4.8 gigabits per second delivers a 33% improvement over the previous generation.

GPUs receive most of the attention in the AI infrastructure race, but storage is becoming another important constraint as training datasets, model checkpoints, embeddings, and inference workloads expand. QLC flash stores more bits in each memory cell, making it attractive where capacity and cost per terabyte matter, although manufacturers must carefully manage performance and endurance. Improvements in bandwidth and density can help data center operators move larger amounts of information while controlling space and energy requirements. That becomes increasingly important as AI clusters grow from thousands to tens of thousands of accelerators. Faster processors are of limited value if storage and networking cannot feed them efficiently. The announcement is another reminder that the AI buildout is forcing innovation across the entire semiconductor stack, from GPUs and high-bandwidth memory to networking chips, SSD controllers, and NAND flash.

Why It Matters: AI’s infrastructure bottlenecks extend far beyond GPUs, and higher-density, faster flash could become increasingly important as model datasets and inference workloads keep growing.

Source: Sandisk.

Naver Backs Wave-Powered AI Data Center Startup Panthalassa

South Korean internet giant Naver has invested in U.S. startup Panthalassa, which is developing offshore AI data centers that combine computing infrastructure with wave-generated electricity and seawater cooling. Yonhap News Agency reported the investment Thursday as Naver continues expanding its exposure to AI infrastructure. Panthalassa, founded in Oregon in 2016, is pursuing floating computing systems that could process AI workloads offshore and transmit results back through communications networks.

The concept addresses two problems becoming increasingly difficult for the AI industry: finding enough electricity and finding suitable land for enormous data centers. Offshore systems could theoretically tap local renewable energy and use surrounding seawater for cooling while reducing dependence on congested terrestrial grids. The engineering challenges are considerable, including corrosion, maintenance, connectivity, weather, equipment reliability, and the economics of operating sophisticated computing hardware at sea. Naver’s investment is nevertheless notable because major technology companies are exploring increasingly unconventional ways to secure compute. The company has also been expanding its domestic AI infrastructure, including plans connected to its GAK Sejong data center. If floating systems eventually prove economical, they could add another option alongside nuclear-powered facilities, renewable-energy campuses, and other emerging approaches to the AI industry’s growing energy appetite.

Why It Matters: Panthalassa’s approach shows how AI’s electricity and cooling demands are pushing data center design into places that would have sounded unconventional only a few years ago.

Source: Yonhap News Agency.

Uber Freight Investigates Cyber Incident After Hackers Claim Million-File Data Leak

Uber Freight is investigating a cybersecurity incident after unauthorized access was detected in part of its systems and a hacking group claimed to have obtained roughly one million files. The company told The Next Web that the incident had been identified, contained, and remediated, and that its freight operations remain secure and fully operational. Uber Freight has also contacted federal law enforcement. The company has not authenticated the hackers’ claims about the alleged stolen data, making the full scope of the incident unclear.

The alleged attackers have been linked by Google researchers to a group tracked as UNC6671, which has used voice phishing and other social-engineering techniques against companies. The incident is particularly significant because freight platforms sit inside broad supply chains connecting carriers, shippers, warehouses, and corporate customers. A compromise can therefore create data risks well beyond a single consumer account. Uber’s core ride-hailing and delivery operations were not affected, according to the report. The event also highlights the growing importance of distinguishing between a cybercriminal’s claims and independently verified breach details. Attackers routinely use public leak claims to pressure victims, while companies must determine what systems were accessed and whether customer information was actually removed before issuing definitive disclosures.

Why It Matters: Cyberattacks against logistics technology can expose sensitive commercial data and potentially disrupt supply chains, making freight platforms increasingly important cybersecurity targets.

Source: The Next Web.

CEVA Logistics Cyberattack Disrupts Eight European Warehouses

Global logistics company CEVA Logistics is continuing to restore systems after a cyberattack disrupted operations at eight warehouses across Europe, according to SecurityWeek. The initial incident occurred on July 29, but new details released to affected customers show that some facilities remained unable to ship stored goods as recovery work continued. Companies reportedly affected include retailers and organizations such as Bol, De Bijenkorf, ING, Ace & Tate, Ajax, and Valve.

CEVA has also warned that some personal information may have been affected. Notifications cited by SecurityWeek referenced data including names, addresses, email addresses, telephone numbers, and order information, although one affected retailer said payment-card details, banking information, usernames, and passwords were not involved in its case. CEVA has more than 1,700 facilities across roughly 170 countries and operates as part of shipping giant CMA CGM, giving the incident broader significance than an isolated warehouse outage. Logistics companies are attractive cyber targets because technology systems increasingly coordinate inventory, transportation, customer records, and delivery schedules across interconnected organizations. Even when physical goods remain safe, disabling warehouse software can slow shipments and generate expensive knock-on effects for customers that depend on precise fulfillment timelines.

Why It Matters: CEVA’s continuing disruption shows how a cyberattack on digital logistics infrastructure can quickly turn into a real-world operational problem across multiple companies and countries.

Source: SecurityWeek.

Space Tech Startup Astranis Unveils Perceptor Satellites to Watch Activity in Geostationary Orbit

Satellite startup Astranis has introduced Perceptor, a new line of spacecraft intended to monitor other satellites and activity in geostationary orbit. The region sits approximately 22,236 miles above Earth and hosts strategically important communications, weather, reconnaissance, and military satellites. According to Space.com, Astranis is positioning Perceptor as a space-domain-awareness system capable of observing activity in an orbital neighborhood that has become increasingly important to national security.

Monitoring geostationary orbit has become more important as governments deploy maneuverable spacecraft and seek better visibility into what other satellites are doing thousands of miles above Earth. Unlike low Earth orbit, where thousands of commercial satellites now operate, GEO contains a smaller but exceptionally valuable population of spacecraft that can remain over roughly the same part of Earth. Astranis already operates satellites in GEO and says its existing systems support both connectivity and U.S. government missions. Perceptor expands the company’s role from communications infrastructure into sensing and surveillance, reflecting a broader convergence between commercial space startups and defense technology. As orbital activity increases, governments are likely to demand better tools for detecting unexpected maneuvers, characterizing nearby spacecraft, and protecting critical communications assets.

Why It Matters: Perceptor highlights the growing market for commercial space-security technology as strategically important orbits become more crowded and more contested.

Source: Space.com.