10 AI Companies That Are Working On 2030 Technology

Everybody knows what OpenAI is, but most people don't know that there are many AI companies out there that are silently going to be changing the world. In today's video, I'll be introducing you to 10 AI companies you probably never knew existed. So, one of the first disruptive companies that you might actually know is Neuralink. Now, this one is incredible because it lets people with paralysis, people who cannot move, use computers or phones with their mind. I know, I know it sounds sci-fi, but it is real. So, the surgery uses a special robot to put the tiny wires right in place, which is 10 times more accurate, just a million times more accurate than a human can do. And in the future, it actually might help people see, hear, or move again if they've lost those abilities. It's literally one of the first companies to make this kind of brain chip small, wireless, and ready for real people, not just experiments. In simple words, Neurolink is making a brain chip that lets people control devices with their thoughts, which could change how we live, even for people with disabilities. I don't know about you guys, but I think that's pretty incredible and disruptive. Another company I'd like to talk about is a company called Renovate Robotics. So, this is a US-based startup focused on automating roof work. So, Renovate Robotics was founded in 2021 and operates out of Brooklyn, New York. So, what does Rufus do exactly? So, Rufus is designed to automate the installation of asphalt shingles on residential roofs with future plans to handle additional roofing tasks. Its main goals are to increase productivity. Roofers can install asphalt shingles up to three times faster than a human roofer, effectively doubling or tripling your productivity for roofing crews. It can also improve safety by reducing the amount of time workers spend on dangerous sloped roof surfaces. Roofers helps lower the risk of injuries and fatalities, which are high in the roofing industry. This also addresses labor shortages. With a shortage of skilled roofing labor, Rufus enables contractors to complete more work with fewer people. It also enhances the quality. Rufus uses machine vision and AI to ensure shingles are installed accurately and consistently, improving installation quality and providing data for warranties. And now, Rufus uses machine vision. It uses cameras and AI powered computer vision to see the roof segment shingle locations and guarantee the precise placement of each shingle. There's also autonomous navigation. The robot navigates roof planes using a winchbased gantry system, moving X and Y axes and adjusting its path based on realtime AI analysis of the roof's layout and obstacles. The roof also records the installation data which can be shared with insurance and warranty providers to verify the quality and integrity of the roof. Now, if you're wondering how this is deployed, Renovate Robotics initially plans to use roofers as a subcontractor, completing roofing jobs with partner contractors. There's also robotics as a service. In the long term, Renovate plans to lease Rufus robots to roofing contractors, allowing them to increase productivity and safety without large upfront investments. Today, we're unveiling Roffus V1. This system is half the weight, much more reliable, faster, and more accurate than the system that we showed earlier this year. This robot will excel on big buildings. That's things like multif family apartment buildings, big houses, and other commercial buildings. It'll work on slopes up to a 1212. So, really helping contractors maximize their productivity on those big projects. One other advantage to this new system is that it's much more modular. We can quick change the endector. So, that lets us test different processes. So, right now we're just focused on shingle installation, but we'll be able to do shingle tear off, solar install, and other processes on the roof. If you're a contractor that's excited about the future of roofing, please reach out to us on our website or on LinkedIn. That's www. rennovate robotics. com. Another company I'd like to talk about is Shield Aai. So, what does Shield AI exactly do? So, Shield AI is a US-based defense technology company specializing in artificial intelligence and autonomous systems for military and government applications. Founded in 2015 and headquartered in San Diego, California, Shield AI's mission is to protect service members and civilians by developing intelligent autonomous systems that can operate in the most challenging and contested environments. Shield's flagship AI product is the Hivemind, an advanced autonomy and AI software stack. Hivemind enables aircraft, drones, and other vehicles to operate fully autonomously. Meaning they can navigate, make decisions, and execute missions, all without GPS, communications or direct human control. This technology is designed to work across a wide range of platforms from quadricepters to fighter jets. The company also has and develops a suite of autonomous aircraft. Now, Shields AI's Hivemine software can be integrated into both its own and thirdparty aircraft, enabling swarming operations, unmanned teaming. The software is designed to scale quickly across different airframes, supporting rapid deployment and integrations. Shield AI's technology

is actively used by the US military and allied forces worldwide. Now, Shield AI uses AI with path planning, mapping, computer vision, and reinforcement learning to allow unmanned systems to carry out complex missions with minimal human oversight. But let's take a look at what they have to say. The key differentiator in our autonomy capabilities and the AI that we uh deliver on our systems is the ability for the systems to operate in a resilient intelligent manner. That is to really think for themselves, to adapt to varying conditions, to operate with or without human input, to be able to navigate conditions where communications may be degraded or denied, where external information may no longer be available. Uh, these systems still are able to operate even if they can't communicate with each other. And when they can communicate with each other, they can then progress and coordinate as a team. VBAT teaming enables operation in comms and GPS environments by using a state estimation framework that is built up without relying on GPS and instead uses shared sensor observations to build up that common operating picture and a collaborative planning framework that is built on top of that expects communication loss. One of the exciting things about VBAT teaming is also how we're deploying the capability. We're putting our AI pilot into a payload that can go onto any VBAT coming off of our production line. So, this transformational capability can be realized with a small modular payload. Every VBAT in the team gets its own AI pilot. This modular payload mounts inside of any of our latest VBA platforms. It contains an Xavier NX compute module with CPUs and GPUs to accelerate our autonomy algorithms, a carrier board to allow us to interface with the peripherals. Fundamentally, from the ground up, we've built our Hivemine architecture to scale for VBAT teams in the hundreds or thousands of VBATs. I could tell you as a former Little Bird pilot commander, Special Ops Aviation Regiment, VBAT teams are going to be the standard by which all other autonomous teams are going to be measured. So now we need to take a look at another company and I think this one may actually be the backbone of the economy because it's crazy what this company can do. This company is called Covariant and they are an American AI and robotics technology company founded in Emeryville, California and it was founded in 2017 and they are a leader in developing AI systems that enable robots to perform complex tasks in dynamic real world environments especially in warehouses and fulfillment centers. So what does coariant do? Well, Coariant builds large generalpurpose AI models, notably the RFM1, that give robots human-like reasoning abilities. And these models are trained on massive diverse data sets, including text, images, video, robot actions, and sensor data collected from real warehouse operations. Now, the company's flagship platform is called the covariant brain. It's a universal AI system that allows robots to see, reason, and act in the physical world. It enables robots to handle a wide range of tasks such as picking, sorting, depalitizing, kitting, and order induction, often outperforming humans in speed and reliability. Now, co-variance technology is widely deployed in warehouses where it powers robotic arms and systems that pick and sort items from bins, manage inventory, and adapt to changing SKUs and workflows. The AI allows robots to learn tasks quickly and share learnings across fleets, making automation more flexible and scalable for logistics, e-commerce, and retail operations. A co-variant partners with major automation and logistics companies such as ABB and NAP to integrate its AI into robotic systems used by global retailers and fulfillment providers. Their technology is trusted by top fulfillment companies to address labor shortages and fluctuating demand. Robotics foundation models have the potential to power billions of robots around the world. Covarian's RFM1 is a robotics foundation model that provides robots the ability to reason much like humans do. This model is setting a new frontier for what's possible by unlocking the next generation of flexible and reliable robotic systems. Representing the first time generative AI has successfully given commercial robots a deeper understanding of language and the physical world. For robots to reason about the physical world, they need a fine-trained understanding of the objects in their environment and how those objects react to robotic actions. Traditional computer-based simulations are limited to what can be hard-coded and programmed. Alternatively, today's generative AI models can generate text, images, and within certain parameters, videos. But they cannot generate videos of realistic robotic actions because they lack the necessary data set focused on robotics. The data needed for robotics foundation models to generate fine- grained simulations of object dynamics can only be found through the real world deployments of robotic systems. Since 2017, Covariant has developed the world's largest multimodal robotics data set by deploying AI powered robots in real world warehouse environments around the world. RFM1 is a

robotics foundation model trained on coariance real world data set which includes the wide range of information a robot needs to make decisions. RFM1's understanding of physics emerges from learning to generate videos. The model can predict via AI generated videos how objects will react to robotic actions. This video is generated by RFM1. It shows a robot picking a polybagged apparel item out of a bin and placing it onto a scan table before getting pushed onto a conveyor. This is a complex scene with several objects interacting with several components which are being predictively generated as video. Next to it, you can see the actual captured video. There are several parts of this robotic pick and place action that are generated by RFM1. Starting with the picking action, this video predictively shows an item being picked out of a bin. RFM1 predicts several key elements to simulate this pick, including interpreting the intended robot action and deciding how to pick the item. RFM1 can also predict how a scene will change after an object is manipulated. For example, RFM1 predicts what the bin will look like after this item is removed. Thanks to its multimodal anytoy setup, RFM1 is not just a video prediction model that simulates how the world would change every fraction of a second. It can also directly predict how the bin in front of the robot will change 3 seconds into the future as a result of that robot's action. Acting as a highle world model, the model understands the prescribed robot action and can reason about the future outcome. Here are images showing the model predicting what the bin would look like if the prescribed robot actions were executed. This shows emerging capability of RFM1 having a notion of object perience, meaning it understands the objects that are not touched will not change. RFM1 can predict how an object will behave upon manipulation. Due to the complexity of the system and item physics, these scenarios would be nearly impossible to predict at scale using traditional simulation methods. RFM1 represents a new generation of flexible and reliable robotic systems that will address the growing shortage of workers willing to engage in highly repetitive dangerous tasks, lifting productivity and economic growth for decades to come. Another company you may have actually heard of is called Anderil. So Anderil Industries is an American defense technology company founded in 2017 by Palaki, the creator of the Oculus Rift. And Anderil specializes in developing advanced autonomous systems, artificial intelligence, and robotics for military and national security applications. And mission is to modernize and transform the United States and allied military capabilities by introducing Silicon Valley style innovation and speed into the traditionally slowm moving defense sector. The company is notable for investing its own capital into research and development, building finished products before seeking government contracts, and focusing on rapid results driven by deployment. So, what does Andre do? Andre builds autonomous unmanned aerial systems and counter UAV systems, autonomous surveillance towers, and sensors for persistent monitoring and force protection. They also have something called Latis OS. This is their core software platform. Latis is an AI powered command and control system that fuses data from thousands of sensors including Andrew's own and thirdparty devices to provide real-time situational awareness, target detection, and automated decision support for operators. Now, Andrew's systems are used to monitor borders, military bases, and critical infrastructure, providing automated detection, tracking, and classification of objects using advanced computer vision and sensor fusion. This is an incredible company that's backed with over $2 billion in venture capital with a recent valuation of $14 billion as of August 2024. And Andrew basically is open about its military focus, aiming to keep the United States and its allies technologically ahead of adversaries like Russia and China. And even recently, they unveiled their autonomous fighter jet, which is set to take flight very soon that could absolutely change the future of warfare. Well, certainly an innovative company. You should definitely be paying attention to what they do. Next, let's take a look at a company that could really change the chip side of things. This is a company that is actually trying to take on Nvidia. So, this is a company called Cerebra Systems. They are a pioneering technology company specializing in building the world's most powerful AI hardware and systems with a focus on accelerating artificial intelligence and deep learning workloads far beyond what traditional computing hardware can achieve. Founded in 2015 by Andrew Feldman and many others, this company is located in Sunnyville, California. So what Cerebras does is the wafer scale engine. It's best known for its largest computer chip ever built. Unlike traditional processors which are made from small separate chips, the WSC is constructed from an entire silicon wafer resulting in a single massive AI processor. The latest version, the WSC3 powers their CS3 system and shatters benchmarks for AI inference and training. Cerebrus packages its WSE chips into dedicated AI systems CS1, CS2, and CS3. now designed for high performance AI training and inference.

And these systems are used by organizations in fields such as medical research, cryptography, energy, and scientific computing to build on premise supercomputers or access AI power through cloud offerings. The company's hardware is engineered to dramatically speed up the training process and deployment of large AI models such as those used in natural language processing or computer vision. For example, the CS2 system powered by the WSC chip features 850,000 AI optimized cores and can train models with billions of parameters much faster than traditional GPU clusters. Cerebra systems offer supercomputer level performance in a space and power efficient package allowing customers to achieve the same computational output with fewer machines and lower energy consumption compared to conventional data centers. Now, in addition to hardware, Cerebrris provides a software platform called Seoft and cloud-based solutions, enabling researchers and enterprises to leverage their technology without needing to manage physical infrastructure. From a shared vision, the Cerebra's founders united to build the unimaginable. They believed they could build something better, to create the largest, fastest chip in the world. Wafer scale integration was once thought to be impossible. But Cerebras turned the impossible into reality. The largest chip of its kind. It stands as a testament to human ingenuity and ambition. A chip to power supercomputers and drive the forefront of AI capabilities. Wafer scale engines are the biggest AI chips now powering the biggest AI supercomputers for the most challenging AI workflows. Fueling advancements across industries, science, and global innovation. Cerebras isn't just shaping the future of AI. It's inspiring the next generation to continue to imagine, innovate, and inspire. Cerebras is powering the future of AI with our revolutionary compute. Designed in Silicon Valley, assembled in America, built for the world. Together, we can push the limits of what's possible. The Black Hornet is a series of ultra lightweight palmsiz military micro drones, nano unmanned aerial vehicles or UAVs designed for covert reconnaissance and situational awareness on the battlefield. Originally developed by Prox Dynamics AS of Norway, the Black Hornet line is now produced and further advanced by Teley Fleer Defense, a subsidiary of Teley Technologies. The latest models, particularly the Black Hornet 4, leverage AI to enhance autonomy, navigation, and mission effectiveness. So, if you're wondering how this actually uses AI, the latest models, particularly the Black Hornet 4, leverage artificial intelligence to enhance autonomy, navigation, and mission effectiveness. The drone uses AI powered systems for pre-programmed missions, allowing it to follow specific routes, avoid obstacles, and adapt to changing environments without constant operator input. There is also collision avoidance. AI enables real-time obstacle detection and avoidance even in complex urban or indoor environments using onboard visual and infrared cameras. There's also GPS denied operation. The Black For Hornet can navigate and complete missions in environment where GPS is unavailable or jammed, relying on AI and onboard mapping. Now, there's also the reduced cognitive load. AI integration allows the drone to handle complex tasks autonomously, minimizing the need for manual control and letting soldiers focus on their mission. There was also real-time data analysis. The drone can process and transmit live video, thermal imagery, and snapshots, often using AI to enhance image quality and target identification. So, now another company that might not change the world, but I wanted to include this one because I thought it was rather interesting, is this company called Caper. Essentially, they have this AI enabled tech that allows you to shop more efficiently. This is Caper Cart, Instacart's AI powered smart cart that customers love. All they have to do is grab an item, drop it in, and Caper instantly adds it to their total on screen. Caper Cart is equipped with smart sensors and computer vision cameras that automatically recognize when a customer has added a product. And the built-in weights and measure certified scales even know how much it weighs. Yep, even produce. It's easy for customers to add items, remove them, and even add them back in. E-commerce integrations let your customers build a shopping list on your app and easily shop it from the cart. Personalized features like location-based deals and your items on sale are available for retailers to tailor every shopping experience to the customer. Even ads on Caper Cart can be personalized to make them relevant to your customers

experience. Surprise and delight your customers with gamification like Spin to Win. Cut down lines with customers paying right on their cart. When it's time to check out, everyone wins. Your customers will even get to save with your store's loyalty program. Caper integrates into your store's infrastructure and POSOS system, too. Conveniently access cart monitoring tools to help prevent shrink. And nesting charging gives caper carts enough juice to run all day. When every trip feels like an adventure, customers increase their basket size and keep coming back for more. Drive more revenue, more loyalty, and bigger baskets with Capercart. So, another company is called Physical Intelligence. Now, this is a robotics and AI startup company founded only in 2024 and based in once again San Francisco, California. Now this company is composed of engineers, scientists, roboticists and company builders and has quickly gained attention for its ambitious mission and significant financial backing from leaders such as Bezos, OpenAI and Sequoia Capital. So, so this company actually focuses on developing generalpurpose models that specifically designed to control a wide range of robot and physically actuated devices. Their core innovation is the creation of foundation models for robotics large versatile AI models trained on diverse data from multiple robots and tasks. Basically, they build LLMs but for physical actions. Their first generalist policy, PI0, is a foundational model capable of controlling various robots to perform complex real world tasks based on simple text instructions. Unlike traditional robot software, which is typically tailored for specific tasks or hardware, PIZ is designed to generalize across different robots and environments. It can handle tasks such as folding laundry, sorting trash, cleaning up rooms, and assembling boxes that require nuanced real-time decision-making and dexterity. Now, one of their most recent innovations is PI 0. 5. And right now, most robots work in factories because everything they do in there stays the same. But if we want them in our homes or offices, we need them to be able to generalize, which is to handle chaos. And that's where the new model 0. 5 comes in. It builds on the older model called Pi 0. But this one is better at doing tasks in new environments it hasn't seen before, like cleaning random houses. So PI 0. 5 doesn't just try to do new tricks. It tries adapting to what it's learned from different robots, videos, and even humans, giving it step-by-step instructions. It learns by mixing all kinds of data, pictures, actions, words, and robot demonstrations. Some data comes from robots with wheels, some from robots with no arms, and some just from the internet. The point is, lots of variety equals better generalizations. They tested this robot in homes they had never seen before. Tasks included things like putting away dishes, cleaning spills, or making a bed. And it didn't always get the right the first try, but eventually it figured out what the human might want to do. So overall, this is a company that aims to create a universal brain for robots that can perform any task in any environment. And this approach could eliminate the need for bespoke software for each robot application, making robotics more flexible, scalable, and accessible across industries such as manufacturing, logistics, healthcare, and home automation.