Cape Town runner plans to run marathon to raise funds for adults with neurodivergence

Most robotics companies focus on collecting vast amounts of real-world data, but Machenta is taking the opposite approach. The company successfully raised $40 million in a seed round to train robots without needing any data from the real environment.

The round was led by Creandum and Union Square Ventures. This amount represents a significant seed check for the London-based startup, which has only 11 employees, and poses a bold challenge to the conventional approach in the robotics industry.

Machenta was founded in 2025 by Edward Jones, a professor at Imperial College London, and his former graduate student Vitalis Vosylyus. Edward Jones serves as CEO. The company's idea originated about six years ago when Jones began questioning whether robots could perform physical tasks as quickly as humans.

This question spurred years of laboratory work at Imperial College, which ultimately led to the creation of Machenta. Early investors included Hoxton Ventures, and British documents point to an earlier, smaller round of around £4.3 million in 2025. Now, the company has much greater financial resources thanks to the $40 million raised.

Machenta's model trains exclusively within computer-generated environments. These simulations mimic how objects move, slide, and collide with each other. Jones told Business Insider that the model was trained without a single pixel of real-world data.

The result of this approach is the robot's ability to master a new task after just one demonstration, without further training. In demonstration examples, the robots performed table cleaning and placed items into bins on an assembly line.

Despite the successes, the system is not yet perfect; precise and complex movements remain a challenge. Early lab versions also required about 30 minutes of additional data collection for each task, which Machenta aims to eliminate entirely.

Machenta's competitors are major players in the market. The physical intelligence market is valued at approximately $11 billion. Reports indicate that FieldAI is raising funds at a $10 billion valuation, and Skild AI exceeds $14 billion and already sells similar technologies, generating $100 million in annual recurring revenue from over 60 clients.

Many of these competitors heavily rely on real-world data, but Jones believes this is a trap. He argues that simply accumulating more data will not solve the problem on its own, as no one knows exactly how much data generalized robots will require. If simulation can bridge this gap, the robotics economy could change rapidly.

Collecting real-world robot data is a slow and expensive process, requiring equipment, personnel, and endless hours of recording. Simulation, however, scales through computational power, which is becoming increasingly cheaper, potentially allowing small teams to train capable robots.

This is why this funding round stands out. The funds will be used to hire staff, acquire more computational power, and generate more simulated training data to tackle more complex and precise tasks.

However, the team is focused on research, and there are no immediate plans for product commercialization. Although early access to the model is not yet open, robot developers can already register their interest. The prospects for manufacturing workshops and warehouses are huge: imagine retraining a robot manipulator in seconds instead of weeks. For investors, this signals that Europe can still compete in the field of physical AI.

The press service of the Fund for the Development of Culture and Arts of Uzbekistan announced that the Samarkand Marathon Series 2026 will take place in Samarkand on November 7 and 8.

The series will open with the Samarkand Marathon Expo, and the following day will feature races of 2 km and 5 km, as well as special events, including a 2 km race with Nordic walking, an inclusive run, and a Kids Run for children.

Participants in the main marathon will be able to run a distance of 42.2 km, as well as take part in a half marathon of 21.1 km and a 10 km race. The start of all races is scheduled for Registan Square, where participants will pass through the iconic places of the historic city.

The organizers of the Samarkand Marathon series will launch an environmental initiative called 'Avvalgidan tozaroq' ('Cleaner than before'). This program involves reducing plastic use, transitioning to reusable materials, implementing a separate waste collection and recycling system, and increasing environmental literacy through an eco-point system.

One element of this environmental campaign will be plogging, which will be held for the third time, allowing runners to collect trash simultaneously. Additionally, during the days of the sporting event, the Samarkand Marathon lecture hall, morning running excursions around Samarkand, and an evening concert are planned.

As is traditional, the marathon will have a charitable nature, and all collected funds will be directed towards building children's sports playgrounds in the regions where the series starts.

Mirkhan Sagitov, Head of the Marathon Implementation Department and Curator of the Samarkand Marathon series, reported that interest in the event from abroad has significantly increased. Over fifteen hundred people from 63 different countries have registered.

According to him, this indicates that the marathon is becoming increasingly known outside Uzbekistan and attracts runners who wish not only to participate in the competition but also to familiarize themselves with the rich cultural heritage of Samarkand. The organizers aim to develop the Samarkand Marathon as a major international sporting event, striving to make Samarkand the main marathon center of Central Asia.

The Samarkand Marathon series also includes events such as Bukhara Night Race, Sky Camp Bo’stonliq, Zaamin Ultra, and Aral Ultra.