Author: Techtime

Enigma’s Interactive AI Robots go Online

Posted on by Techtime

photo above: Enigma founders, Jonathan Jacobi (right) and Gal Niv

Enigma, a physical AI company, emerged from stealth phase with a $71 million seed round led by Index Ventures and Ribbit Capital, with participation from Conviction Partners and leaders from OpenAI, Anthropic, DeepMind, xAI, Cognition, Wiz and others. Founded less than a year ago by Jonathan Jacobi and Gal Niv, Enigma builds AI models to make robots intelligent and effortless to use. The company trains AI models that bring intelligence to any robot, on any hardware, and builds novel user interfaces that make them simple to use.

Enigma has developed more efficient ways to train its foundation models for robotics, lowering the requirements for massive manual data collection while keeping systems reliable in physical environments. The company is building a unified software solution that pairs its models with the ability to adapt them across different robots and settings – removing much of the engineering complexity traditionally required to deploy intelligent robots.

While robots have grown more capable, they remain difficult to use. Enigma advances capability and usability together, developing AI models alongside novel user interfaces and robot-agnostic software that make robots intuitive, on any robot and for any task. The goal is robots that are not just powerful, but natural to use and interact with, for engineers, enterprises, and end users alike. “No matter how capable robots get, if they aren’t intuitive to use, most people never will,” said Jonathan Jacobi, co-Founder and CEO of Enigma.

“AI is moving beyond chatbots and screens”

To answer how to build intuitive interfaces with robots, Enigma is unveiling at www.robots.online a first-of-its-kind experience: 100 real AI-powered robots that anyone can use online, in real time. Using Enigma’s models and interfaces, people will have robots complete tasks and handle physical objects. At this scale, Enigma can learn how people instinctively approach robots – feeding directly back into better models and interfaces. Jacobi: “We believe AI’s next chapter is moving beyond chatbots and screens into systems that can understand, adapt to, and operate in the physical world. Our goal is to make intelligent robots as natural to work with as computers and smartphones are today.”

Enigma was founded 11 months ago by CEO Jonathan Jacobi and Gal Niv. The co-founders met in Israel’s elite Unit 8200, where they even shared a bunk bed. Jacobi started a computer science degree at 13 and finished it in high school, then became the youngest-ever employee at both Microsoft and Check Point at 17. Niv began hardware hacking at 10, joined a cybersecurity startup at 17, completed a four-year degree in a single year, and became the unit’s youngest cyber-operations manager. After years of working together across cybersecurity, scientific research and large-scale systems, they founded Enigma on the belief that physical AI is the next frontier.

Microchip to Acquire Hailo

Posted on by Techtime

photo above: Raspberry Pi 5 whith Hailo AI Accelerator

Microchip Technology has signed a definitive agreement to acquire Hailo, a Tel-aviv based provider of accelerated edge AI processors and comprehensive AI software flows. The transaction is expected to close towards the end of the current quarter ending September 30. The terms of the transaction are not being disclosed. The proposed acquisition is expected to expand Microchip’s portfolio for intelligent edge systems and strengthen its ability to deliver edge AI solutions for intelligent edge applications, including drones, robots, smart cameras, industrial automation and embedded AI systems.

With Hailo-8, Hailo-10 and Hailo-15 processors, Microchip will gain a portfolio that supports classic computer vision, while adding advanced camera, ISP, DSP, video encoding and AI video stream processing capabilities for intelligent edge systems. The Hailo acquisition brings multiple products, more than 100 current customers and an established developer community of more than 10,000 users. Its portfolio spans both edge AI accelerators and vision SoCs supporting workloads and capabilities including CNNs, transformers, LLM/VLM workloads, Image Signal Processing, DSP, H.264/H.265 encoding and AI video stream processing.

The transaction would also bring a demand-generation engine for edge AI adoption. Hailo’s Raspberry Pi ecosystem, gated Developer Zone, GitHub activity and community forum create a self-sustaining funnel that converts developer engagement into qualified opportunities and customer pipeline. “Microchip’s customer reach, channel scale and broad technology portfolio would create a strong platform for bringing advanced vision processing and AI acceleration to broader range of intelligent edge applications,” said Hailo CEO Orr Danon.

A Unicorn That Lost Its Shine

Hailo was founded in 2017 by veterans of IDF elite Military Intelligence technology unit. The company was established around a fundamentally new computing architecture designed specifically for deep learning workloads, enabling real-time AI tasks such as object detection while significantly reducing power consumption, chip size, and cost. Since its inception, Hailo has raised approximately $340 million in funding.

In its latest financing round, completed in April 2024, the company secured $120 million. According to a Reuters report, the round valued Hailo at approximately $1.2 billion, earning it “unicorn” status. At its peak, Hailo employed around 250 people. In early June 2026, Hailo announced the layoff of approximately 110 employees – nearly half of its workforce – as part of a broad restructuring initiative aimed for “aligning the organization’s structure with the next phase of its growth.”

Xsight Labs Raises $300 Million at a $2.8 Billion Valuation

Posted on by Techtime

Israeli semiconductor startup Xsight Labs has raised more than $300 million in a new funding round valuing the company at $2.8 billion post-money, as it prepares to ramp production following a series of major customer wins.

The round was led by Fidelity Management & Research, with participation from both existing and new investors, including Intel Capital, Battery Ventures, T. Rowe Price, Valor Equity Partners, and Atreides Management.

According to the company, the new capital will accelerate development of its next-generation networking chips, expand its engineering workforce, and increase manufacturing capacity, supply-chain capabilities, and customer deliveries as it prepares for what it describes as large-volume orders from Tier-1 customers.

Xsight said its two flagship products—the X2 Ethernet switch and the E1 DPU—have already been selected by several global network operators and are currently being evaluated by leading hyperscale cloud providers.

One customer already disclosed publicly is SpaceX’s Starlink, which selected the X2 chip for its next-generation Starlink V3 satellites. According to previous company statements, the chip will serve as a core component of the satellites’ onboard networking infrastructure, handling traffic measured in terabits per second.

Founded in 2017, Xsight Labs was established by Israeli networking semiconductor veterans Gal Malach, Guy Koren, and Erez Shizaf. The company’s founding investor and chairman is Avigdor Willenz, one of the founders of Galileo and a key figure behind several successful semiconductor companies, including Annapurna Labs, Habana Labs, and Leaba Semiconductor.

The company now employs more than 250 people and has raised over $750 million since its inception.

Tackling One of AI’s Biggest Bottlenecks

The funding comes as networking has become one of the most critical bottlenecks in AI infrastructure.

As hyperscale data centers deploy ever-larger GPU clusters, performance increasingly depends not only on compute power but also on the ability to move enormous volumes of data between servers, accelerators, storage systems, and switches with extremely low latency and high energy efficiency.

Xsight’s flagship switching product, X2, is a 5nm Ethernet switch ASIC manufactured by TSMC. It delivers 12.8 Tbps of switching capacity, supports links of up to 800 Gbps, consumes less than 200 watts, and offers processing latency below 700 nanoseconds.

The company’s differentiation extends beyond raw performance. Xsight has developed a programmable instruction architecture called XISA, enabling customers to customize packet processing, introduce new protocols and networking functions, and modify the data plane even after deployment. The company positions this as an open and programmable alternative to more proprietary networking architectures.

Its second flagship product, the E1 DPU, is designed to offload infrastructure workloads—including networking, storage, virtualization, and security—from CPUs and GPUs. The processor delivers up to 800 Gbps of throughput and is built around 64 Arm Neoverse N2 cores.

Unlike architectures that route part of the traffic through slower processing paths, Xsight says all processing cores reside directly in the data path, allowing every packet traversing the chip to be processed at full speed.

Demand for programmable networking silicon continues to grow as hyperscalers and AI system builders seek alternatives to proprietary network architectures while improving utilization, power efficiency, and software flexibility.

Xsight competes in a market dominated by Broadcom, Nvidia (Mellanox), and Marvell, but aims to differentiate itself through a combination of high performance, energy efficiency, and greater programmability.

In that context, the latest financing represents more than another funding round. It is intended to finance Xsight’s transition from product development and technology validation to high-volume manufacturing and commercial deployment. If the company succeeds in converting hyperscaler evaluations into production orders, it could emerge as a significant new player in one of the most strategic layers of AI infrastructure.

“Don’t Build an Entire Weapon System—Build a Component Defense Primes Can Easily Integrate”

Posted on by Techtime

[Photo: DefenseTech Haifa conference. Credit: HiCenter Ventures]

As investment in Israel’s defense-tech sector continues to accelerate, one of the strongest messages emerging from this week’s DefenseTech conference in Haifa was strategic rather than technological: for most startups, the fastest route to market is not developing a complete weapon system, but building a specialized component that can be integrated into the platforms of major defense contractors.

According to data published by JNS, startups working with Israel’s Ministry of Defense have raised nearly $3 billion since the beginning of 2026, compared with roughly $1 billion during all of 2025. The report attributed the surge to growing private investment driven by rising global demand for defense technologies following recent conflicts.

That message was at the center of the conference, organized by HiCenter Ventures, Gornitzky GNY, and EY, which brought together entrepreneurs, investors, representatives of Israel’s defense establishment, and executives from leading defense companies.

Ilana Abrakin, Head of Defense Tech at HiCenter Ventures, argued that early-stage companies should avoid competing directly with established defense manufacturers such as Rafael, Elbit Systems, Israel Aerospace Industries, or global defense primes.

“We encourage founders to design their products as independent, flexible, and platform-agnostic components that companies such as Lockheed Martin, RTX, Israel Aerospace Industries, Elbit Systems, or Rafael can easily integrate into their existing platforms through a plug-and-play approach,” she said.

According to Abrakin, many startups fall into the trap of trying to build complete end-to-end systems, even though their commercial prospects are significantly stronger as technology suppliers to established defense contractors. In line with this strategy, HiCenter Ventures announced plans to invest in ten defense-tech startups this year while supporting them with fundraising, pilot programs with defense companies, export compliance, and long-term strategic partnerships.

Rafael also signaled its intention to deepen collaboration with startups. Dr. Moshe Shuker, Senior Vice President of R&D at Rafael, said the company is adapting its innovation strategy to the rapidly evolving defense-tech ecosystem and is looking to expand cooperation with young companies—from integrating individual technologies and subsystems to collaborating on Rafael’s flagship defense platforms.

“The combination of agile startups and established defense organizations can dramatically shorten the path from an idea to a battlefield-ready capability,” he said.

Another trend highlighted during the conference was a shift in investor priorities. According to HiCenter Ventures CEO Lior Hanuka, investors are increasingly focused on the software and artificial intelligence layers that power defense systems, rather than on the hardware platforms themselves.

Key areas of interest include battlefield operating systems, mission-management software for drones and autonomous robots, secure military communications networks, and AI applications capable of supporting real-time decision-making. At the same time, lessons learned from the conflicts in Gaza, Lebanon, and Ukraine have made electronic warfare, GPS denial, spectrum dominance, and counter-drone technologies some of the sector’s fastest-growing investment areas.

The overall picture emerging from the conference is of a defense-tech ecosystem that is rapidly maturing: less emphasis on building complete platforms and greater focus on specialized technologies that can be integrated into existing defense systems—a model that enables startups to reach the market faster while becoming part of the supply chain of major defense contractors.

Nanovel Secures €2.5M EU Grant for AI Harvesting Robots

Posted on by Techtime

Israeli ag-tech startup Nanovel has been awarded a €2.5 million non-dilutive grant through the European Innovation Council (EIC) Accelerator, part of the European Union’s Horizon Europe program. In addition, the EIC has selected the company as a candidate for an equity investment of up to €4.3 million, which could serve as an anchor investment in Nanovel’s planned €8 million funding round.

According to the company, the grant will be used to complete development of the commercial version of its autonomous harvesting robot, finance field trials in European citrus orchards, and accelerate its transition from Technology Readiness Level (TRL) 6 to TRL 9, ahead of a planned commercial launch in 2028. Nanovel is already conducting field trials in Spain and Italy, following validation trials in California carried out in collaboration with the Citrus Research Board (CRB).

Nanovel is addressing one of agriculture’s most pressing challenges: the growing shortage of manual labor for fruit harvesting. According to the company, manual picking accounts for roughly 50% of production costs for citrus growers, making automation a potentially significant driver of profitability while helping alleviate labor shortages.

Founded in 2018 by Itzik Mazor, former founder and CEO of Jordan Valley Semiconductor, which was acquired by Bruker, Nanovel is developing an autonomous harvesting robot powered by artificial intelligence, computer vision, and advanced robotics. The system is designed to tackle one of the industry’s most difficult technical challenges: harvesting fresh fruit concealed within dense foliage without causing damage.

At the core of the platform is a combination of deep learning models for fruit detection, an advanced computer vision system, and multiple telescopic robotic arms operating simultaneously. Unlike conventional harvesting robots that simply pull fruit from the tree, Nanovel has developed a proprietary Grip & Trim mechanism that gently grasps and cradles each fruit before precisely cutting its stem. The approach is designed to preserve fruit quality—an essential requirement in the fresh produce market, where even minor damage to the peel can significantly reduce commercial value.

The platform’s central computer coordinates the operation of all robotic arms and enables remote fleet management of multiple harvesting robots. Beyond harvesting, the system continuously collects agronomic data and supports selective picking based on fruit size and ripeness.

Nanovel is currently focused on harvesting oranges and lemons, but its technology roadmap includes expanding the platform to additional crops such as mangoes, peaches, and nectarines. The company’s long-term vision extends beyond replacing individual workers: it aims to provide a fully autonomous harvesting platform capable of operating around the clock in existing orchards, without requiring growers to redesign or modify their orchards to accommodate robotic harvesting.

Photo credit: Tal Bardak

The AI Price War: Meta and xAI Take Aim at OpenAI and Anthropic’s Moat

Posted on by Techtime

[Image caption: A satirical illustration depicts Mark Zuckerberg, Elon Musk, Sam Altman, and Dario Amodei as market vendors loudly advertising the prices of their AI models]

By Yohai Schwiger

The launches of Meta’s Muse Spark 1.1 and xAI’s Grok 4.5—announced just one day apart—may signal a new phase in the AI race. For the past two years, competition has largely centered on one question: who could build the most capable model? Now, a second question is rapidly emerging: who can afford to sell intelligence at the lowest price?

The target is a market still dominated by OpenAI and Anthropic. Both companies have established themselves as the leading providers of commercial AI models, serving a broad base of enterprise customers and developers. As a result, Meta’s and xAI’s aggressive pricing strategies represent more than a commercial decision—they appear to be an attempt to challenge the current balance of power and capture market share from the industry’s two leaders.

The pricing gap is striking. Muse Spark 1.1 costs $1.25 per million input tokens and $4.25 per million output tokens. Grok 4.5 is priced at $2 and $6, respectively. By comparison, OpenAI’s GPT-5.6 Sol costs $5 for input and $30 for output, while Anthropic’s Claude Opus is priced at $5 and $25. In other words, Meta’s and xAI’s flagship models are priced dramatically below the premium offerings of the two market leaders.

The obvious question is: why?

One possibility is that both companies have achieved technological breakthroughs that significantly reduced inference costs. Yet neither company emphasized that message in its product launch. xAI highlighted improved token efficiency, while Meta explicitly promoted what it described as aggressive pricing. At least for now, the story appears to be less about technology than about competitive strategy.

The real difference lies in their business models.

OpenAI and Anthropic are, first and foremost, AI research companies. Their core product is the model itself, and a significant share of their economic value depends on monetizing access through APIs, subscriptions, and enterprise offerings.

Meta operates under a fundamentally different model. Most of its revenue comes from advertising, meaning it does not need to recover its AI investment from every API call. Lower-priced models can accelerate adoption of Meta AI, strengthen Facebook, Instagram, and WhatsApp, attract developers to its ecosystem, and ultimately reinforce its core advertising business.

xAI also enters the competition from a different position. Although it is not a traditional hyperscaler, it benefits from access to Elon Musk’s broader ecosystem, including substantial capital, computing infrastructure, and strategic distribution channels. For xAI, expanding Grok’s user base may be just as important as maximizing revenue from each API request.

That gives both Meta and xAI greater flexibility to wage an aggressive pricing battle. Meta generates enormous cash flow from its advertising business, while xAI enjoys access to significant financial and infrastructure resources. OpenAI and Anthropic, by contrast, remain much more dependent on generating direct revenue from their AI products.

If that is indeed the strategy, the immediate goal is not necessarily higher profits but faster adoption, greater developer engagement, and larger market share. In other words, the objective is not simply to compete with OpenAI and Anthropic, but to pull them into a pricing battle where Meta and xAI may enjoy structural advantages.

OpenAI and Anthropic have already begun responding. Over recent months, both companies have expanded their portfolios with lower-cost models aimed at customers who do not require their most powerful systems for every task. The emergence of these new pricing tiers suggests that both recognize the competitive landscape is no longer defined solely by model quality.

Behind the pricing battle lies a more fundamental question: do frontier AI models still possess a sustainable competitive moat?

If comparable performance becomes available at significantly lower prices, competitive advantage may shift away from the models themselves and toward distribution, infrastructure, proprietary data, and customer-facing applications. In that scenario, the greatest value will no longer be created where models are trained, but where customers actually use them.

It is still too early to know whether Muse Spark and Grok 4.5 will materially reshape the competitive landscape. One thing, however, is already becoming clear: the next battle in AI will not be fought solely over who builds the smartest model, but over who determines the price of intelligence. For OpenAI and Anthropic—both widely expected to pursue public listings in the coming years—that could prove to be one of the defining strategic questions of their future.

Razor Labs Appoints Erez Egozi as CFO

Posted on by Techtime

Razor Labs, an AI company developing artificial intelligence solutions for the mining industry, has appointed Erez Egozi as its new Chief Financial Officer. Egozi succeeds Meital Cohen, who will step down following the publication of the company’s financial results in August 2026 after deciding to pursue new professional opportunities.

Egozi brings more than two decades of financial leadership experience in international technology companies. Over the past seven years, he served as Chief Financial Officer of GigaSpaces Technologies, where he was responsible for the company’s financial strategy, capital raising, financial planning and control, risk management, and commercial transactions with enterprise customers worldwide. Earlier in his career, he served as CFO of Pluristem and previously held a series of senior management positions at Verint.

The appointment comes as Razor Labs continues to expand its presence in the global mining market. According to the company, it recently launched a new analytics platform designed to monitor mining trucks manufactured by companies including Caterpillar and Komatsu. At the same time, the company is pursuing new commercial agreements while continuing its transition toward a recurring revenue (ARR) business model.

Meital Cohen, who has served as CFO since December 2023, led the company’s finance organization during a period in which Razor Labs signed strategic agreements worth tens of millions of dollars with a major international mining corporation. She also oversaw the company’s transition to a recurring revenue business model.

Razor Labs Co-founder and CEO Raz Roditi said Egozi is joining the company “at a critical stage of accelerated growth” and thanked Cohen for her contribution to strengthening the company’s business position.

Egozi said he is joining Razor Labs ahead of “a significant new chapter in the company’s development” and plans to support its continued international expansion and long-term growth strategy.