Alibaba-Backed Ninja Van Faces $1B Valuation Reset in Brutal Southeast Asia Funding Winter
Warning! GuruFocus has detected 2 Warning Signs with BABA.
The recalibration mirrors broader pressures across the region's tech landscape. Grab Holdings Ltd. and GoTo Group have underperformed relative to U.S. Big Tech as they juggle growth ambitions with the need to shore up profitability. For logistics players, the challenge is even steeper. Competition from J&T Global Express Ltd. and Sea Ltd.'s SPX Express is keeping margins thin, even as e-commerce adoption accelerates in a market of more than 650 million people. Investors are increasingly focused on operational discipline, and valuation resets are becoming part of the playbook for companies looking to stay competitive.
Founded in 2014, Ninja Van operates in six Southeast Asian markets, delivering about 2 million parcels a day, according to its website. Its last major raise$578 million in a 2021 Series E from Alibaba, B Capital, and otherspushed its valuation well north of $1 billion. That momentum has since cooled, with management shelving IPO ambitions until profitability improves. For now, the company's near-term story could be less about chasing market share and more about finding the right cost structure to make the eventual public market debut worth investors' wait.
This article first appeared on GuruFocus.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Geek Wire
28 minutes ago
- Geek Wire
On a mission to ‘kill Google Chrome,' UW students join Y Combinator to launch AI-powered browser
GeekWire's startup coverage documents the Pacific Northwest entrepreneurial scene. Sign up for our weekly startup newsletter , and check out the GeekWire funding tracker and venture capital directory . Meteor co-founders and Farhan Khan, left, and Pranav Madhukar during a Y Combinator demonstration. (Photo courtesy of Meteor) Two computer science students on leave from the University of Washington are building what they call the 'world's most intelligent web browser,' using agentic AI to complete tasks like a personal assistant. Pranav Madhukar and Farhan Khan are co-founders of Meteor, a browser they launched this week out of Y Combinator, the Bay Area startup accelerator where the two students are participating in the summer 2025 batch. 'Essentially, we're on a mission to kill Chrome,' Madhukar said in reference to Google's market dominating internet browser. 'Or acquire it, for only $34.5 billion?' he quipped, off the news this week that AI startup Perplexity offered that much to purchase Chrome. Perplexity, with its Comet Browser, along with Meteor and several other startups, are all vying to change the way users browse and perform myriad tasks online — with help from generative AI. Madhukar, Meteor's CEO, believes the majority of time we spend in a browser each day is boring and repetitive work. 'AI is changing the game in basically every space, and we want to bring that to the browser, to make all of our lives a lot more efficient, a lot more streamlined,' he said. Meteor's agents can perform all sorts of tasks with simple prompts, ranging from adding a calendar item to applying for an internship to purchasing groceries or flight tickets. Further blurring the lines around what constitutes 'cheating' in the AI age, the Meteor browser will also complete homework assignments, as Madhukar shows in the demo video below. Meteor's purple cursor can be seen taking over tasks as it's prompted via a text window on the right. Madhukar and Khan also want to build in infrastructure such as buttons that provide suggestions to users, making the browser more proactive. Voice prompts are also on the horizon. 'The vision is to build your browser into your personal assistant,' Madhukar said. Madhukar and Khan met during orientation week at UW in Seattle. They were set to be incoming juniors this year before the plan for Meteor launched them on a different path. Madhukar previously built a couple viral projects that grabbed the attention of other students, including uwgeoguessr, a game that shows photos from areas on campus and users have to guess where those areas are. He also built a website that helped identify who the professor would be for a specific class at UW. It attracted around 2,500 active users before it was taken down by university administrators, Madhukar said on LinkedIn. Khan, who is Meteor's CTO, previously built an integrated development environment for hardware design that he said compiles 60 times faster than the industry standard. Meteor is built on Chromium, the open-source codebase that powers Google Chrome. Just a few weeks into Y Combinator, Madhukar and Khan made the decision to 'fork' — or create their own version of — Chromium's 30 million lines of code, rebuilding their product at the time from scratch. So far, Y Combinator has taught the pair the importance of moving fast and staying true to their vision. 'A lot of what we're trying to do is hold the convictions that we have,' Madhukar said. 'Two years from now, everyone will probably be using browsers that look nothing like the ones that we use today. Starting from that, we were like, 'How can we be the people to build that?'' Whether they return to Seattle from the Bay Area to build the browser and AI startup of the future remains to be seen. The accelerator runs into September, and after that everything's up in the air. 'We're not planning to return to UW for the foreseeable future,' Madhukar said. 'We're excited to make Meteor go big.'
Bloomberg
30 minutes ago
- Bloomberg
BofA's Soft Spot for Data Centers and Singapore
Hi, it's Elffie Chew in Singapore, where I've been speaking with BofA investment banker Martin Siah about the hot topic of data centers. Elsewhere today, bankers are risking a US backlash over the China deals boom in Hong Kong. Today's top stories
Forbes
30 minutes ago
- Forbes
Robocar Startup 'Tensor' Unveils Luxury Self-Drive Car For 2026
Tensor is a Silicon Valley based startup that plans to be the first company to sell a true robocar (Link goes live at 10am PDT) with 'eyes off' self-driving ability and a steering wheel that folds away and is replaced by a screen. While most entrants have worked to first make a robotaxi to provide ride service, Tensor aims to start with a high-end electric vehicle consumers can buy, and either drive themselves, or allow it to do the driving. Able to operate without the cloud, they promise luxury, an agentic experience that does your bidding, and unlike so many vendors, a promise of privacy from being tracked. The vehicle is large and the design is sleek, though adorned with the largest number of sensors I've yet seen. This includes 37 cameras, 5 custom-designed lidars, 11 custom radars, and an array of microphones, ultrasonics, collision sensors, water detectors, data radios and more. Most of the sensors have built in cleaning methods and left with clear sightlines for good visibility. The vehicle was designed from the ground up to be a personal robocar, but will be made by Vietnamese automaker Vinfast. Tensor is a rebranding of the U.S. portion of AutoX, a Chinese/American robotaxi project which launched self-driving service in the suburbs of Shenzhen several years ago with a large fleet, showing off video of over 1,000 robotaxis, but has generated minimal news for 2 years. Tensor says it is an 'evolution' of AutoX aimed at personal cars. AutoX was one of the earliest companies to deploy vehicles with no safety driver in China, but had faced controversy over allegations of operation without active safety intervention controls in the USA. Tensor's design focus has been to spare few expenses to get out there first and do what was necessary to maximize safety--which is needed to be early. Tensor/AutoX began in 2016, and has had California test permits (including for unmanned operation ) since 2000. Their latest software approach is machine learning centric but not end-to-end, rather relying on a collected series of ML-based tools and their own transformer based 'Tensor Foundation Model.' Taxi vs. Personal Car They currently decline to comment on pricing, which is frustrating for a luxury product, but it will be more than existing luxury EVs like the Lucid Air, which do not self-drive. They have also decided to focus on sales to consumers rather than taxi service. Most self-driving teams doing cars, including AutoX, have started with ride-hail. There are a number of strong reasons for that. A taxi need only drive one city to be a workable business, and it comes home every night, remaining under the close watch and control of its maker. A consumer car must provide strong value in most major cities, or the market is too small. With no depot to visit the car or its owner must handle everything. Tensor will not be able to drive every street upon release, but plans to handle a worthwhile subset, such as highways and arterials in non-snowy places. Owners can drive the car conventionally on roads it doesn't self-drive on. Tesla also has ambitions to sell a personal self-driving car, but they recently decided to try a taxi first, in a limited area. Tesla's offerings can't self-drive at present, and instead operate in driver assist mode with a supervising human. For the last 8 years, Tesla has predicted their car will self-drive unsupervised within a short time, and this prediction was made again next week, but skepticism is high after the many failed predictions. Many other companies have showed concept personally owned self-driving cars, including features like fold-away steering wheels and reversible seats, but nothing has been announced for production. Tensor hopes to be the first, claiming they will ship in the 2nd half of 2026. Privacy An interesting, and sadly unusual feature of the Tensor is an effort to protect driver privacy. All other robocar designs are constantly in touch with the cloud, relaying all the car's activities to it. The cars have interior cameras which can feed video back to HQ (though Waymo and others say this is only turned on in unusual situations.) The Tensor also talks to the cloud for the usual reasons of updating maps and software and fetching traffic and weather, and it has the ablity for remote operations assistance that all robocars have, but the owner can turn that off, and give the car any 'assist' from the controls. In self-drive mode, the yoke wheel folds into the dashboard, and the center screen slides over to cover it; the pedals retract into the footwell. The interior of the vehicle is spacious and uncluttered. Tensor says in this mode, even though the vehicle can self-drive, the owner's data stays with the car and is not uploaded. This is a refreshing take. It would be nice to see more attention to privacy not just from consumer cars but from robotaxis. Today's transportation world is mostly private and anonymous. Even taxis used to be summoned with a street hail and paid with cash. That world is vanishing, replaced by a world where all your travels are recorded. Of course, robotaxis need a way to know if a passenger soils or damages the car and to get payment if that happens (or simply tell you if you left your Stradavarius behind) but there are ways to still do that with intermediate escrow that protect privacy and erase information when it's no longer needed. Interesting Hardware The vehicle is entirely drive-by-wire, with multiple redundant systems, including 3 braking systems, to allow this to meet automotive safety standards. Most cars have some physical link as a backup, but when self-driving that doesn't help. Some of the cameras are 17 megapixel, higher resolution than is typical. Radars have 1.1 degree imaging resolution which is fairly good. The car has cameras under the vehicle, which would presumably detect anything or anybody underneath the vehicle, a failure which sunk Cruise. Water sensors will detect if the car drives in water that's too deep, something we've seen a few other robocars fail on. Tensor's LIDAR claims very dense resolution, though I did not get a chance to personally examine its results. It includes a spinning 360 degree 1500nm unit on top, and a variety of smaller short range units on the sides for full visibility. Also present on the front bumpers are small screens at the corners which are used to show pedestrians that the vehicle sees them, and any intentions the vehicle has (like doors about to open.) While a robocars absolutely must see every pedestrian around it in every dirction at all times--or its badly broken--many people don't know that and so are comforted with a reminder. Software The Tensor's hardware suite is impressive but expensive. Tensor says it improves safety and this will help them come to market first. No matter how good the hardware is, however, the software is the key. I rode in the vehicle in a parking lot, but this showed only simple operations and doesn't allow judgment of the quality of the self-driving stack. Tensor built their own, on top of tools and hardware from their partner down the street, Nvidia. But most cars have this. We'll need to see millions of miles of test data to judge if the self-driving system is safe enough to bet your life on. Tensor didn't say, but presumably the code is not based on the AutoX stack, as new regulations starting in 2027 forbid Chinese software in self-driving systems.
