LesenForschungRadarAnlageframework
Anmelden / Registrieren
AlleTechnikProdukteInvestierenBörseSonstiges

Lesen

2026-11-1298 Beiträge

Aktualisieren

Mondo Launches Beni Dual-Wheeled Following Robot: Starting at 4,999 Yuan, Nearly 8,000 Orders on Kickstarter

Thema · 妙动科技Beni机器人Wesentlich
2026-11-12 11:44 GMT+8

On November 10, Chinese startup Mondo (Miaodong Technology) officially launched its first dual-wheeled following robot, Beni, for global sale. Weighing less than 2 kg and smaller than a cat, the device starts at 4,999 yuan in the domestic market, placing it directly within the price range of DJI's Mini series drones and high-end gimbal cameras.

Beni's core selling point is its unique "dual-wheel leg" design. It can travel at speeds exceeding 28 km/h on paved roads, navigate irregular terrain, and even perform jumps and somersaults. This design allows it to automatically compress and balance its chassis like an off-road vehicle, smoothly traversing grass-paved paths or gravel roads. For filming, Beni features a mechanical gimbal and a 1/1.3-inch CMOS sensor, supporting 4K60P video recording and 2x electronic zoom.

In practical testing, Beni demonstrated strong pure-vision tracking capabilities. Testers could select a target in complex crowds without wearing a motion tracker, and Beni maintained lock-on even when passersby crossed or the target accelerated. After being tripped by a speed bump and righting itself, it could re-acquire and lock onto the out-of-frame target within 3-5 seconds by swinging its gimbal. However, current intelligent obstacle avoidance is limited at speeds above 5 m/s, and advanced actions like climbing stairs require manually enabling specific modes that reduce forward speed.

Mondo's team consists of former DJI executives and core algorithm engineers who previously worked on Tesla's Optimus humanoid robot project. They position Beni as "filling the gap in ground-based autonomous mobile imaging," primarily targeting low-angle dynamic following shots for family, pet recording, and urban sports like skateboarding and tennis. During pre-sales, Beni secured nearly 8,000 crowdfunding orders on Kickstarter, becoming one of the highest-funded single products in the consumer robotics category on the platform.

For the embodied AI industry, Beni's launch represents a significant Product-Market Fit (PMF) exploration. Instead of pursuing a general-purpose humanoid form, the company found a commercial entry point by solving a clear pain point in vertical niche scenarios: stable low-angle dynamic following shots that are difficult for human photographers to achieve. Although the first-generation product still has limitations in operational proficiency and all-terrain adaptability, its market feedback proves that consumers are willing to pay for this new form of hardware.

Quellen:GeekPark

Forschung

German Startup SE3 Raises $30M Series A to Advance Spatial AI for Military Drones

Thema · SE3无人机融资Wesentlich
2026-11-12 16:00 GMT+8

Munich-based AI startup SE3 has announced the completion of a $30 million Series A funding round, led by prominent venture capital firm Accel, with participation from Keen Venture Partners, Seedcamp, and Lakestar. The company specializes in developing "spatial intelligence" software for military drones and other unmanned systems, aiming to address the widespread issue of vendor lock-in in the defense tech sector.

SE3's core product fuses video and sensor data from multiple drones and ground vehicles to generate a unified 3D map of the battlefield. Soldiers can query this map using natural language, asking questions such as how to reach a target without being detected or what route is suitable for heavy vehicles. Additionally, SE3 develops AI systems embedded within the drones themselves, enabling them to navigate autonomously, detect objects, and coordinate with other drones even when GPS signals are jammed.

According to Fortune, SE3 co-founder and CEO Lukas Köstler stated that the most basic capability—GPS-free localization—has been delivered thousands of times and is "very robust," but the higher-level fleet command AI is still in early stages and requires further development over the next one to two years. Köstler previously worked on Tesla's Autopilot team and at Nvidia Research.

Unlike approaches relying solely on general-purpose large language models, SE3 uses classical computer vision algorithms to process sensor data, feeding it into a language model equipped with conventional software tools for querying. This architecture allows customers to swap in their own preferred frontier AI models as the system's "brains," thereby avoiding dependence on a single hardware vendor. Köstler noted that if drone manufacturers each control their own software stacks, customers face costly lock-in.

Currently, SE3 has contracts with Germany's armed forces, the Bundeswehr, to develop related technologies, and its technology is already deployed in Ukraine. Following this funding round, SE3 plans to increase its headcount from approximately 40 to 120 employees and open offices in London, Berlin, the Nordics, the Baltics, and Poland. While defense will drive revenue for at least the next 12 months, the company also intends to begin exploring civilian robotics.

Quellen:Fortune

Forschung

Colombian Rental Fintech Startup Palomma Raises $10M Series A Led by Mexican VC Firm Cometa

Thema · Palomma融资Kurzfassung
2026-11-11 21:00 GMT+8

Medellín-based Colombian rental fintech startup Palomma has announced the completion of a $10 million Series A funding round led by Mexican venture capital firm Cometa, according to an exclusive report by Crunchbase News. Including $4 million in previously undisclosed funding, Palomma's total raised stands at $14 million. Other participants include Volpe Capital, Clocktower Technology Ventures, Tekton Ventures, Simma Capital, and Yango Ventures, alongside existing investors Y Combinator, 17Sigma, Nido Ventures, and Norte Ventures.

Palomma provides financial services for landlords, tenants, and property managers. Its business combines software that helps property managers handle leasing inquiries, customer service, and collections with a payments system for collecting rent and paying landlords. These relationships allow it to sell financing, rental guarantees, and other financial products to tenants and property owners. The company reports that its annual recurring revenue has grown thirtyfold in 18 months to about $3 million. It processes approximately $130 million in monthly rental payments, serving more than 200,000 tenants through over 500 property managers in Colombia.

Cipriano Echavarría, co-founder and CEO of Palomma, stated that the funding will be used to expand in Colombia, develop additional AI and financial products, and prepare to enter other markets. While considering Mexico, Brazil, and the Spanish-speaking U.S. market, Colombia remains the priority. Pepe Bolaños, managing partner at Cometa, noted that Palomma uses AI to automate operations and leverages proprietary payment data to improve underwriting, providing a competitive advantage that can strengthen as the company grows.

Quellen:Crunchbase News

Forschung

Investment Blogger Joachim Klement: Cost of Capital for Small European Businesses Exceeds That in the US, with the Gap Widening as Size Decreases

Wesentlich
2026-11-12 15:01 GMT+8

Klement admits his previous judgment was wrong. He had assumed that lower interest rates in Europe meant corporate borrowing costs should be cheaper than in the US; according to Pablo Fernandez’s annual survey, equity costs in Europe and the US were also roughly comparable. However, research by Bo Becker, Efraim Benmech, and Joao Monteiro provided an unexpected answer: the cost of capital for European firms is actually higher, and this gap widens the smaller the firms are.

The original question of this study was to explain why almost all of the world's largest trillion-dollar companies are in the US—there is not a single one in Europe. The researchers found that the gap does not stem from fundamentals, but rather that there are fewer new companies in Europe, newly listed firms have valuations lower than their US peers, and they grow more slowly after listing. The key variable is the size premium within the cost of capital: for small European firms below the median market capitalization, the size premium in their cost of capital is far higher than for similar US firms. In the latter half of the 2010s, this gap continued to widen, making it even harder for European small companies to obtain the capital necessary for growth.

Klement acknowledges that Becker et al.'s study only covered listed companies, but he believes the same issue applies to private companies that have not yet gone public. This inference itself is not a finding of the study, but rather the author's own judgment. However, if the cost mechanism indeed exists prior to listing, it can supplement the earlier OECD research findings on funding supply: the problem for European startups is not just that top-tier capital is hard to reach, but that every step of expansion from small to large faces higher financing costs than US counterparts.

For investors and policymakers, this means diagnostics need to be more precise. If the problem is merely insufficient funding supply, increasing fund sizes would solve it; but if the cost itself is the obstacle, the issue is not whether there is enough capital, but whether capital is cheap enough for small businesses. Klement argues that Europe and the UK should discuss how to make capital cheaper for small companies so they can invest in their own growth.

Quellen:klementoninvesting.substack.com

Forschung

AMD Zen Original Architect Mike Clark Explains Zen 6 Design Shift: Dense Cores Become Highest IPC Performance Core

Thema · AMD Zen 6架构Wesentlich
2026-11-12 06:22 GMT+8

Mike Clark, AMD's original Zen architect, explained a fundamental design shift during an AMD In-Depth interview with Chips and Cheese: starting with the Zen 6 generation, AMD's dense technology cores are no longer area-optimized products but serve as the highest IPC performance cores. He explicitly stated that while dense technology was previously just an area trade-off, this generation's team leveraged it to add extra IPC within the core. The classic design is now repositioned as the high-frequency version, while dense technology serves as the top-tier performance core because full L3 cache has been enabled in this implementation, providing 4MB per core, totaling 128MB for 32 cores, and delivering the highest achievable IPC.

The mechanism behind this reversal is not mysterious. Clark explained that because Zen 6c targets lower frequencies than the high-frequency cores, it can accommodate more capacity in its queues—the floating-point scheduler is just one example; other queues throughout the core are larger in Zen 6c, aiming to extract more IPC and gain leverage in SMT environments. He simultaneously emphasized a design principle: there should be no software-differentiating delta between the two architectures, meaning the same optimizations apply to both.

Regarding AI inference capabilities, Zen 6 adds two VNNI pipelines on the vector side. Clark admitted this was not the initial design goal—they did not use the term 'agentic' when they began designing. However, Zen 5 had already established a new foundation for AI workloads, and the team always believed the core had a place in AI: orchestration, data type manipulation, and small model inference. Observing the direction of inference workloads, they decided to add VNNI pipelines. Since these are long-cycle projects and the floor plan was already set, this was a relatively late decision, but the team ultimately succeeded in squeezing in the extra pipelines.

There is a counter-intuitive reduction in the execution engine: the number of integer multipliers decreased from three in Zen 5 to two. Clark explained that Zen 5 used a more unified scheduler, where adding a third multiplier to the scheduling group was simple and did not increase complexity. However, Zen 6 required returning to a more distributed scheduler to scale the architecture; that extra multiplier did not yield significant gains in Zen 5, so to optimize area and the scheduler, they reverted to two multipliers.

The frontend also saw improvements. Zen 5's dual-decode setup made it difficult for a single thread to synchronize instruction cache and operation cache fetches in larger workloads. Zen 6 spent considerable time optimizing this, improving efficiency from both delivery and power perspectives. Regarding branch prediction, misprediction recovery latency was reduced, and accuracy saw a slight improvement. Clark stated they do not intend to expand the branch prediction structure but rather layout TAGE tables more efficiently, while focusing each generation on restarting execution as quickly as possible after a misprediction.

Quellen:chipsandcheese.com

Forschung

South Korean Humanoid Robot Startup Robros Announces World Action Model, Targets 2027 Open-Source Release

Thema · Robros世界动作模型Kurzfassung
2026-11-12 14:18 GMT+8

South Korean humanoid robot startup Robros announced its latest technology roadmap on November 10 during the 'NVIDIA Physical AI Startup Connect' session at NVIDIA AI Day Seoul 2026. The presentation was led by Im Dae-gyu (임대규), the company's Chief AI Officer (CAIO). Key announcements included a 'Real-to-Sim-to-Real' training methodology and a proprietary 'World Action Model' (WAM).

According to the launch details, the 'Real-to-Sim-to-Real' approach involves creating a digital twin of actual work environments within a virtual space. Robots are trained in this simulation using NVIDIA's Isaac Sim, Isaac Lab, and Omniverse platforms, before the learned actions are transferred to physical robots for validation. This method aims to reduce the time and cost associated with repeated trial-and-error in the real world and mitigate dangerous operations. Robros has already reconstructed its laboratory environment using this method and validated box and manufacturing part handling tasks on its proprietary humanoid robot, IGRIS-C.

Additionally, Robros revealed plans to develop the WAM, an AI model designed to understand surrounding environments and generate corresponding robot actions. Currently, the model is undergoing pre-training using simulation data, videos of human operations, and data collected from actual robot deployments. The company explicitly stated that it targets releasing the WAM as an open-source project in 2027. If realized, this would provide global robot developers with a foundational action-generation model based on a simulation-first strategy.

Robros indicated that it will expand proof-of-concept (PoC) initiatives in manufacturing and logistics while simultaneously advancing its own AI model development and field data accumulation. Im emphasized that deploying humanoid robots in industrial settings requires data reflecting diverse working environments and validation technologies, stating that the company will enhance robot operational capabilities through its proprietary platform and simulation-based training.

Quellen:스타트업 생태계의 모든 것 \'플래텀(Platum)\'

Forschung

Tesla CEO Musk Extends Cybertruck Production Through End of Summer 2027 to Clear Backlog

Wesentlich
2026-11-12 02:52 GMT+8

Tesla CEO Elon Musk said on Wednesday via X that the company will extend Cybertruck production "through the end of summer 2027" to work through more of the vehicle's order backlog. This decision reverses a previous plan to cease production around mid-2027.

Five days earlier, Musk had announced that Tesla would wind down Cybertruck production at its Gigafactory Texas around the middle of next year to free up assembly capacity for the steering-wheel-less Cybercab robotaxi and the Optimus humanoid robot. The extension implies that Tesla must maintain the complex Cybertruck supply chain and production line longer than initially intended, potentially slowing its pivot toward AI-driven mobility and robotics.

According to Cox Automotive data, Tesla sold just 76,000 Cybertrucks over the past three years, significantly underperforming sales expectations. Despite low volumes, Musk noted that a substantial number of unfulfilled orders remain, forcing the company to adjust its capacity allocation strategy which was originally designed to prioritize new business lines.

Quellen:Business Insider

Forschung

Google Cloud Releases 2027 Security Trends: Proposes 'Beyond Zero' Architecture to Manage AI Agent Risks

Wesentlich
2026-11-12 08:00 GMT+8

U.S. cloud provider Google Cloud released its 2027 enterprise security trend forecast on November 12, 2026, stating that artificial intelligence will further reshape the cybersecurity landscape. On the defensive side, AI agents are expected to assist Security Operations Centers (SOC) in automatically investigating and handling incidents; however, on the offensive side, software supply chain contamination, data extortion, and abuse of AI models and cloud computing resources will become primary threats.

To address risks arising from excessive permissions granted to AI agents, Google Cloud proposed the "Beyond Zero" security architecture. This approach argues that enterprises cannot merely verify whether a user or AI agent has permission to access an application; they must also restrict accessible resources and executable operations based on security policies and real-time risk. This shift aims to prevent attackers from using techniques like prompt injection to induce high-permission AI agents to execute malicious actions and move laterally across other enterprise resources.

The report also highlighted management blind spots caused by "Shadow AI Agents." As employees or departments build AI agents without integrating them into IT security management, these agents may unknowingly access sensitive data or consume cloud resources. Google Cloud recommends limiting tool usage permissions for such agents and tracking their compute usage and costs. Additionally, Mandiant, a subsidiary of Google, disclosed "LLMjacking" attack cases, showing that attackers use stolen API keys to impersonate enterprises and call large model services, with some victims accumulating up to $1 million in unauthorized fees.

Quellen:iThome

Forschung

South Korean Tech Giant Naver Signs ~2 Trillion KRW AI Compute Deal, Plans Independent AI Agent App and Logistics Investment

Thema · Naver AI战略Wesentlich
2026-11-12 14:19 GMT+8

South Korean internet giant Naver disclosed major strategic developments during its third-quarter earnings conference call on November 12, 2026. CEO Choi Soo-yeon stated that the company has signed a computing service supply contract worth approximately 2 trillion KRW (about $1.4 billion) with a US-based AI inference company. The deal covers about 20 megawatts (MW) of computing power over four to five years, with initial revenue expected to begin in 2027. This order marks the first large-scale external customer for Naver’s “AI Factory” business, validating its strategy to convert internal infrastructure capabilities into B2B services.

The contract includes Nvidia B300 chips and the next-generation Vera Rubin platform. Naver emphasized that it has proactively secured most domestic data center capacities equipped with high-density power and liquid cooling facilities, noting that contracted capacity already exceeds its 2028 target of 200 MW. CFO Kim Hee-chul revealed that negotiations are underway with at least ten other potential clients, with further announcements expected. Naver previously reduced capital expenditure pressure through an investment agreement with Brookfield Asset Management and participation in Nvidia’s equity increase, adopting a model where an asset-holding company purchases GPUs while Naver leads operations.

Beyond infrastructure, Naver plans to launch an independent AI agent application by the end of 2026. This app will not only integrate generative AI features within search but also serve as a standalone service capable of performing actual tasks for users, such as bookings and payments. To adapt to the AI agent era, Naver’s advertising billing model will shift from “pay-per-click” to “value-based” charging tied to actual transaction outcomes, with a pilot program scheduled for the first half of next year.

In the e-commerce sector, Naver announced the establishment of a dedicated logistics subsidiary, planning to invest nearly 1 trillion KRW by 2030 to enhance delivery competitiveness. Data shows that free shipping and return policies increased member purchase frequency by 27%, and Q3 N Delivery transaction volume grew 38% year-on-year. By building its own logistics infrastructure and using shopping data to predict demand, Naver aims to shorten delivery times and optimize inventory distribution, consolidating its advantage against competitors like Coupang.

Quellen:바이라인네트워크

Forschung

Brian Lum, Fund Manager at Baillie Gifford Shin Nippon: Seiko’s Growth Logic Is Premiumization, Not Sales Volume

Thema · 精工高端化投资Wesentlich
2026-11-04

When managing Baillie Gifford Shin Nippon, Brian Lum’s inclusion of Seiko in the portfolio raised a specific question: how can a watchmaker with a century-and-a-half history coexist with other fast-growing technology small-caps in the fund? Lum’s answer is that this is not the same type of growth. Most companies in Shin Nippon are driven by rapid revenue expansion and disruption of the status quo; Seiko is different. Its growth is built on sustained premiumization—shifting the company’s focus from mass-market quartz watches to luxury brands, using accumulated brand equity to push profit margins from single digits toward double digits.

The underlying logic of this mechanism is a pyramid structure: consumers enter the brand ecosystem with entry-level casual watches and gradually upgrade to become enthusiasts and high-end collectors. Grand Seiko is the core vehicle of this structure, currently accounting for approximately 20% of Seiko’s sales, but with production volumes far below one-tenth of Rolex’s output. Lum believes that if the brand narrative and craftsmanship perception continue to accumulate, Grand Seiko could long-term reach the top five watch brands. In his view, the shift in profit margins from single digits to double digits is a natural outcome of this process.

The historical basis comes from the quartz crisis itself. Around Christmas 1968, Seiko launched the Quartz Astron 35SQ, and within a decade, two-thirds of jobs in the Swiss watchmaking industry disappeared. Lum cites this history not to prove that Seiko will again disrupt Switzerland, but to show that the company possesses the technical capability and cultural capital to challenge the existing landscape. He states directly: 'Seiko almost killed the Swiss watch industry before; it is not entirely impossible that it threatens its luxury watch sector again.'

The constraints are equally clear. Seiko needs to reverse the perception that it makes cheap watches, and its brand storytelling capability has not yet reached the level of top Swiss luxury brands. Lum acknowledges that Seiko is not yet the best storyteller among Swiss peers, but believes the raw ingredients are in place. Furthermore, the move from single-digit to double-digit profit margins is his estimate, not an achieved result. The family-operated philosophy of 'Don't hurry, don't stop' means this is a process measured in decades, not something for the next quarter.

The most powerful counterpoint is that this is a fund manager stating his thesis for a holding. Margin improvement is an expectation, brand ranking is a judgment, and historical analogy is rhetoric rather than causal proof. The article provides no independent verification data such as Seiko’s actual margin trajectory, Grand Seiko sales growth rates, or brand awareness surveys to confirm that premiumization is occurring. Readers should understand this as an investment argument with a specific mechanism, rather than a proven track record.

Quellen:bailliegifford.com

Forschung

RedBlue Capital Closes First Close of $50M Second Fund Amid Emerging Manager Squeeze

Thema · 红蓝资本第二只基金Wesentlich
2026-11-12 04:04 GMT+8

The US venture capital market is experiencing unprecedented concentration. According to PitchBook data, firms raising their fourth fund or later captured 88.2% of total US VC fundraising dollars in the first three quarters of 2026, while first-time funds secured a mere 4.5%. Only 81 first-time funds closed in the first nine months of the year, making 2026 potentially the worst year for new managers in at least a decade.

Against this backdrop, Olaf Sakkers’ firm, RedBlue Capital, announced the first close of its second fund, targeting a total size of at least $50 million. Sakkers studied anthropology at Princeton and served in the Israel Defense Forces, where he became interested in autonomous vehicles. He joined Maniv Mobility as an early investor before founding RedBlue in 2021 to focus on the 'real economy,' including robotics, logistics, manufacturing, and energy.

RedBlue’s strategy relies on differentiation and operational efficiency. The fund writes small checks of $50,000 to $100,000, typically joining deals led by other investors rather than leading them. This approach allows it to participate in more transactions and provide practical support, such as introductions to lead investors like Andreessen Horowitz. Sakkers notes that institutional limited partners (LPs) have risk models built to 'avoid getting a zero,' favoring established brands and making trust difficult for new managers to earn.

By his account, RedBlue’s first fund, approximately $5 million, has invested in 28 companies, with 15 marked up and a net internal rate of return (IRR) 'significantly above 50%.' However, there has been only one exit—an acqui-hire—meaning most gains remain unrealized paper profits. Public records confirm that HappyRobot, an early RedBlue investment, raised a $150 million Series C at a $1.2 billion valuation in August, and Westmag emerged with an $11 million seed round led by a16z in June.

Notably, Carta data indicates that smaller VC funds between $1 million and $10 million consistently outperform larger funds of $100 million or more at the top end of net TVPI (total value to paid-in capital). This suggests that small, specialized emerging managers may find unique viability in the current structure. RedBlue also leverages AI tools, such as Claude Code, to build custom databases for managing relationships and meeting notes, enhancing asynchronous work efficiency.

Quellen:TechCrunch

Forschung

AI ROI Measurement Firm Nebuly Raises $21M Series A Led by Planven

Thema · NebulyKurzfassung
2026-11-12 13:00 GMT+8

According to a report by Tech Funding News, Italian startup Nebuly has closed a $21 million Series A funding round led by Planven. Other participants included Exor Ventures, Tenacity Capital, TGVG Capital, and Fastweb+Vodafone via Founders Factory. CDP Venture Capital SGR also participated through its AI Fund and Digital Transition Fund.

Nebuly provides a platform for measuring the return on investment (ROI) of AI agents for enterprise clients. As companies increase spending on generative AI, proving the tangible value of these expenditures has become a critical challenge. CEO Francesco Signorato stated that the platform aims to answer boardroom questions about what AI is actually worth to the organization.

Founded in 2022 with offices in Turin and New York, Nebuly quantifies the value of AI tasks by analyzing employee usage of AI tools and the revenue impact of customer-facing assistants. Contracts typically start at $100,000 per year, targeting large enterprises with revenues exceeding $1 billion. Current clients include Oura, Iveco Group, Fastweb+Vodafone, CNH, and Zeiss.

The new capital will be used to grow the team from 40 to 60 employees within two months and hire sales personnel in New York, London, and across Europe.

Quellen:Tech Funding News

Forschung
Weiter →
Anmelden / Registrieren
LesenForschungRadarAnlageframework

Vector loads increased from two in Zen 5 to four (256-bit). Since Zen 5 could already perform two 512-bit loads, the 1,024 wires were already present; splitting their usage was simply more complex than using them contiguously. The L2 cache bandwidth issue was also resolved: Zen 5 prefetchers would sometimes preempt L2 bandwidth, whereas Zen 6 ensures prefetches and demand requests can run in parallel continuously, delivering the full 64 bytes per cycle.

Interviewer George Cozma reported a measured result: Zen 6c matched AMD's fastest Zen 5 server CPU in SPECint 2026, despite the frequency disadvantage. This provides a concrete anchor for Clark's IPC argument, but remains a vendor-side observation based on a single benchmark test and has not yet undergone independent verification.

All technical details stem from first-party statements by AMD architects. Clark also acknowledged that some content constitutes undisclosed secrets—for instance, the specific techniques for maintaining bandwidth within the L3 mesh. Readers should view these as explanations of design intent rather than independently verified performance guarantees.