SoftBank founder Masayoshi Son is doubling down on the company’s AI investments with his plan to create and list Roze in the U.S. According to the Financial Times, this startup will focus on AI and robotics and is designed to help build data centers. More importantly, Son said that he’s aiming for a $100 billion valuation for the company and plans to list it as early as this year.

The Japanese telecom and investment firm is one of the biggest investors in AI and tech. It partnered with OpenAI and Oracle in early 2025 to invest $500 billion in the Stargate project, although it has since faced issues, and OpenAI has reportedly abandoned first-party data centers in favor of leasing compute. It’s also building a 10-gigawatt data center in Ohio, powered by a $33 billion natural gas plant funded by the Japanese government.

Aside from its AI investments, the company also poured $2 billion into Intel more than a year after the chipmaker announced disastrous results. While this isn’t a huge amount in the context of semiconductor fabs, it’s a vote of confidence in the direction that the company is taking and a crucial injection of funds at a time when Intel was struggling to right the ship.

Son is known for taking huge risks on tech investments. For example, SoftBank acquired ARM Holdings in 2016 for $32 billion and still holds around 90% of the company despite listing it in 2023. Its current market cap is around $223 billion, meaning SoftBank made a windfall with this investment. In fact, the company used Arm shares as collateral for a $5 billion loan to invest in OpenAI. It also invested $20 million in Alibaba in the year 2000 and only exited the position in 2024, giving the company $8.5 billion — about 425 times its initial investment.

However, not all its bets were winners. Some of its large losses came from investments in WeWork and various tech startups, especially during the early years of its Vision Fund. In fact, the company lost 93% of its market value during the dotcom crash and was at risk of going bankrupt during that time. Despite that, the company is still going big on AI today, even though there were some concerns that it was turning into a bubble.

SoftBank’s massive investment in AI means that it’s prone to fluctuations in the fast-evolving industry. It showed a $50 billion loss in late 2025 when investors started asking about the profitability of the billions of dollars poured into AI infrastructure. Although the company has rebounded since then, it’s facing another challenge after news leaked that OpenAI missed internal targets for both revenue and active users.

As Tesla begins its sudden pivot from an EV firm to a robotics and AI company, it's building out the supply chain it will need to reach its lofty goals of affordable humanoid robots that are manufactured at a rapid pace. But the scale of its production ambitions means it will need to rely heavily on the supply of raw materials, components, and manufacturing labor from China, as SCMP reports.

Tesla has a strong standing relationship with China, with its own Shanghai production facility that employs 20,000 people. It also sources many of the batteries for its vehicles from Chinese suppliers, and uses lots of raw materials in vehicle production sourced from China. China is also a major buyer of its vehicles, with over a third of all its sales in 2025 coming from China.

But the United States and China have spent the last year erecting and tearing down trade barriers. From America's side, the economic muscle came from its access to cutting-edge graphics cards and other chips used for AI training and inference. China's strength came from its manufacturing industry and access to raw materials, often termed "rare earths."

Although both are necessary for the AI industry to flourish, the ratio is much more lopsided when it comes to robotics. The "brain" of a humanoid robot will always demand impressive processing power and benefit from advanced AI software; the robots themselves demand an awful lot of critical materials. Many of those materials are mostly found in China.

While Chinese manufacturing and supply dominance affects all industries that produce just about everything, the scale of critical materials and manufacturing know-how necessary for next-generation robotics makes companies like Tesla incredibly dependent on the country.

The long list of must haves

Robots are complicated in ways that silicon and semiconductors aren't. Where chips are intricate and minute, and need clean room fabrication and specialized design software to even conceive of how they could be made, robots are decidedly more physical. Sure, they're technically advanced, but they're also made with iron, titanium, nickel, chromium, copper, and manganese. They use actuators, motors, bearings, and lubricants.

In short, they need a lot of stuff to make them, and a lot of that stuff comes primarily from China. Although key materials like Beryllium and Boron have strong supply chains within the U.S., China dominates the global supply of critical materials like Gallium (94%!), Zinc, Neodymium, Molybdenum, Indium, and Praseodymium, to name just a few.

It's not just the raw materials under the ground that's important, though. It's the expertise to extract it, the processing facilities to make it usable, and the logistical infrastructure required to get it where it needs to go.

“With about 50 to 70 per cent of manufacturing and core component production expertise residing in China, we expect Chinese players to take on greater roles in the global humanoid robot supply chain,” said Cheng Xin, a partner at US consultancy Bain & Co. “In some core components … they accounted for at least 55 per cent of the global humanoid robot bill of materials (BOM).”

China has enormous control over the supply of almost everything Tesla needs to build its next generation of humanoid robots. That puts it in an incredibly strong position, with the ability to curtail Tesla's ambitions whenever it likes.

China holds the cards

Unfortunately for Tesla, this has already happened. In April last year, Tesla CEO Elon Musk complained that a Chinese block on exporting "rare-earth magnets" had impacted production of the Tesla Optimus robot. He was forced to ask for a license to use them - perhaps an in-person visit would have helped.

China took all rare earth mineral resources within its borders under state ownership in the fall of 2024, so its leadership has the top-down control necessary to maintain tight controls on the supply of these kinds of critical materials well into the future. Although it has cut that supply outright to the U.S. several times during the ongoing trade negotiations with the Trump administration, it has since begun issuing limited export licenses to some companies.

The American government is aware of this chokepoint in its supply of such critical materials - many of which are also important for national security and the production of cutting-edge silicon - and is attempting to develop a new stockpile to provide a buffer against China's dominance. But even that $12 billion investment would only amount to a 60-day reserve, and its focus is on providing the necessary materials for civilian needs. In such a scenario where access to raw materials was cut off, Tesla would be forced to compete with many other firms and organizations that need those same materials.

Even beyond the risk of mercurial leadership threatening trade relations, China also has its own domestic humanoid robotics industry to consider. There are over 100 Chinese companies currently working on designs, and Chinese officials are making moves to centralize their development to accelerate progress and cut down on parallel research.

Chinese companies are already shipping consumer-grade robots, with over 13,000 deployed in 2025 alone. That's years of production ahead of the several hundred prototype Optimus models Tesla has made.

Just as BYD has overtaken Tesla in the EV space, China seems almost destined to get a huge head start in humanoid robotics over Western companies like Tesla. It already has more units in the world, its production is already higher, its access to key technologies and raw materials is better, and on tap.

Everything is close, efficient, and available. That's not something Tesla or other Western robotics firms can match.

And if China ever needed to slow down the competition, it could turn the tap off for those same supplies that are much harder to find anywhere else. Even if that didn't halt progression elsewhere, it would have a huge effect on pricing.

With Tesla hoping to get its Optimus down to $30,000 per unit in the future, the realization of that dream may be more down to China's whim than anything else.

Around 16,000 humanoid robots were installed globally in 2025, according to a report from Counterpoint Research (via SCMP), and of that total figure, almost 13,000 were deployed in China. Around 5,200 of them were shipped by the Shanghai-based Agibot Innovation, with a further 4,200 shipped from Hangzhou-based Unitree Robotics. In comparison, the top-performing Western company for humanoid robotics shipments was Tesla, but it secured less than five percent of the overall market and came in fifth in terms of overall sales.

This backs up similar data reports from Omdia, which found Chinese humanoid robotics companies took all the top spots in terms of sales for 2025. It was a bumper year for the technology, with sales increasing almost 500 percent year on year. Omdia predicts that by 2035, the market will grow to around 2.6 million units a year, while CounterPoint Research suggests that it will jump another 500 percent in 2026 alone, which would see close to 100,000 humanoid robots deployed if realized.

Without a major leap in production, sales, and deployment from Western companies, it seems only likely that the vast majority of those humanoid robots will come from Chinese companies.

This is extremely important, as the industry is currently finding its footing, deciding upon standards and norms. If Chinese firms set those, it could cement their dominance for some time to come.

“The next two years will see more humanoid enterprises commercialising the mass-production versions of robots, and their performance will largely determine the development pace of the whole industry,” Counterpoint said in its report, and this is something that Nvidia is also clearly thinking about, too.

If China can leverage its enormous manufacturing capabilities to advance humanoid robotics and deploy them at scale rapidly, it risks winning the AI race without needing to be the best or the smartest. It just needs to be the one that can get the software into the hardware and get it where it needs to be: In factories and homes.

World models are the next frontier

Although the top AI companies are still competing to develop the next best large language model for their chatbots, image generators, and smart assistants, for robotics, the frontier developments are “World Models.” These are neural network AIs designed for the physical world, trained on video and image data rather than simply raw text. This is evident in Nvidia's full-stack physical AI platform, which uses its Cosmos AI platform to train data for self-driving cars, as explained at CES 2026.

Data annotation and the training itself are incredibly intensive, costly, and lengthy processes, so it’s no surprise to learn that Nvidia is keen to be a part of it. Its Cosmos platform provides developers with foundational models to build upon, and it offers bespoke versions of its Blackwell-powered server designs specifically tailored for robotic development workloads.

Agibot released its World dataset in late 2024, including a range of foundational models for training different AI-driven robotics. It might not be chasing the superintelligence moonshot, but models like G0-1 are already showing impressive capabilities. It’s building ties in other strong manufacturing countries, too, like Malaysia.

Unitree making its own UnifoLM-WMA-0 world model open source will aid adoption, too. It currently uses it in its G1 humanoid robot, but with most of its revenue currently coming from the education sector, making its models open source gives a much greater chance of its standards becoming commonplace within the market, helping to cultivate a strong developer ecosystem. Nvidia's GR00T is also a foundational model for robotics, while it comes under a research and non-commercial license, it's not open source.

Sometimes you don’t need to be the best to become the standard. You just need to be the most accessible.

Consolidation and fears

One bright spot for humanoid robotics companies outside of China is that there are too many of them inside China. Part of the reason the country has made such great advances in humanoid robotics in recent years is the sheer scale of people working in the industry. At the start of 2025, there were fewer than 100 companies globally building and developing humanoid robotics. By the end of the year, there were over 150 in China alone. This has its advantages, but the diversification of talent and parallel developments could ultimately slow the industry and lead to a contraction there.

Forbes quotes Li Chao, the spokesperson for China’s National Development and Reform Commission (NDRC), as saying there are simply too many companies working on the technology, and this has the potential to overwhelm the industry. He believes that it would be beneficial to consolidate some of these efforts to prevent more duplicate developments. China has most recently made efforts to consolidate its tech stack, so it wouldn't be entirely surprising to see this move in robotics.

Americans working in space agree. Speaking on the TechFirst podcast, former head of NASA’s robotics and AI unit, Dr. Robert Ambrose, said that he thought the messier and “chaotic” American entrepreneurship model of encouraging development of the technology was more likely to result in a stable industry in the long run.

But it will need to put the financial investment in place to achieve that. With the Chinese government already using a state-backed ‘guidance fund’ to encourage what it estimates could be close to $140 billion in humanoid robotics over the next couple of decades, it is clearly banking on it as a future economic success story and strategically important technology. Although some of the companies receiving that investment may ultimately fold if an AI bubble pops, it’s still accelerative in ways that current Western approaches to humanoid robotics aren’t.

Even without the firms that kick-started the industry, the infrastructure, standards, and deployed hardware would still be there, building a strong foundation for future developments.

Working to their strengths

China might have a strong advantage in its established manufacturing base, increasingly close ties with other regional powerhouses, and it got out of the gate strong with Huawei despite export controls in the AI hardware race. But that doesn’t guarantee its win.

A major strength of American innovation over the decades has been its ties with its partners and allies around the world. Japan has incredibly experienced robotics industries and companies. Europe has the advanced machinery necessary to create chips for robotics at scale. The elements are all there, but whether or not they can deliver is another question entirely.

Over the next few turbulent years, China may have a golden opportunity to get there first, with open-source models easing adoption. Western companies might have flashy demos and the backing of the biggest firms in the world, but if China can cultivate relationships, deploy good-enough humanoid robotics safely and cheaply at scale, and do it first, that’s a compelling argument to make its standards, global standards.

A diverse group of researchers from Nvidia, Stanford, Caltech, and other institutions has introduced NitroGen. In a LinkedIn post on Friday, Jim Fan, Nvidia Director of AI & Distinguished Scientist, heralded NitroGen as “an open-source foundation model trained to play 1000+ games.” However, the implications are much wider, spilling from game worlds into the real world, with sizable benefits for simulations and robotics.

You could say, this research presents an attempt to distill a ‘GPT for actions.’ Thus, it is a kind of LLM breakthrough, applying this proven large-scale training tech beyond the fields of language and computer vision. Moreover, the pioneering building of “generally capable embodied agents that can operate in unknown environments has long been considered a holy grail of AI research,” asserts an introduction to the research paper.

Interestingly, NitroGen’s foundation is the GROOT N1.5 architecture, originally designed for robotics. And its application inside the world of gaming shows potential to circle back and bring great benefits to robots working in diverse or unpredictable environments, too.

NitroGen was adapted to play games packed with wildly different mechanics and physics – that’s the nature, and fun, of video games. The researchers used 40,000+ hours of public gameplay videos shared by streamers. Videos in which gamers overlaid their real-time gamepad interactions on the stream were particularly helpful.

In tests, NitroGen was successful in games as diverse as “RPG, platformer, battle royale, racing, 2D, 3D, you name it!” enthuses Fan. While results are promising, the Nvidia scientist says that this is just the start, with a big hill left to climb.

This first version of NitroGen is intentionally focused on fast motor control, or ‘gamer instinct,’ as Fan calls it. According to the shared research, the new LLM also has a “strong competence across diverse domains,” and the model works in procedurally generated worlds as well as unseen games with a “52% relative improvement in task success rates over models trained from scratch.”

All the research into NitroGen so far has been open sourced, and those interested in gaming, robotics, and LLMs are encouraged to tinker. Tweaks to the pretrained model weights, the entire action dataset, and code are all open to your flights of fancy and fiddly digits.

The GTC 2018 Slideshow

Nvidia's yearly GPU Technology Conference (GTC) was overflowing with demos, presentations, tutorials, and sessions. As expected, AI and deep learning topics dominated the show. And much to our disappointment--even though we pretty much knew it wasn't happening--Nvidia didn't announce new cards for gaming. The company did announce a new Tesla-powered Quadro professional graphics card and what it bills as "The World's Largest GPU," which is actually sixteen Tesla GV100 GPUs crammed into a single chassis that draws a mind-bending 10kW of power.

We'll cover all that, and more, as we take you on a trip through Nvidia's bustling tech showcase.

Jensen Huang's Keynote

Jensen Huang, Nvidia's CEO, took to the stage in a freewheeling keynote to announce several new products, including the "world's largest GPU," new photo-realistic real-time ray-tracing advances, the Volta-powered Quadro GV100, and the company's latest advances in autonomous driving and robotics.

Huang's keynote came just days after an autonomous Uber vehicle was involved in a fatal accident with a pedestrian. The accident has sent shockwaves through the industry as Uber pieces together the details of the incident. Huang struck a somber note while he discussed the company's approach to autonomous vehicle safety, explaining that it designs its autonomous driving platforms with both redundancy and diversity to ensure the highest level of safety. Huang later revealed that the company had suspended portions of its autonomous driving program even though Uber's vehicle didn't use Nvidia's DRIVE technology.

Nvidia Unveils "The World's Largest GPU"

Nvidia's new DGX-2 server, which Huang bills as the "World's Largest GPU," comes bearing 16 Tesla V100s wielding a total of 512GB of HBM2. Nvidia's new NVSwitch interconnect fabric ties it all together into a cohesive unit with a unified memory space. The DGX-2 server also features two Intel Xeon Platinum CPUs, 1.5TB of memory, and up to 60TB of NVMe storage all crammed into a chassis weighing 350 pounds.

The system pulls 10 kW of power and provides twice the performance of the previous-generation DGX-1. The 16 NVSwitches provide up to 2.4TB/s of throughput between the GPUs, thus powering the system up to two petaFLOPS (via the tensor cores). External connectivity comes in the form of eight EDR InfiniBand or eight 100GbE NICs that provide up to 1,600Gb/s of bandwidth.

Nvidia claims the beast can replace 300 Intel servers that occupy 15 racks. And more importantly, the system is purportedly 18 times more power efficient than those 300 servers. Bold claims indeed. You can get your own DGX-2 to test out Nvidia's claims, but it'll set you back $400,000.

Boosting The Tesla V100

The Tesla V100 is a formidable beast: the 815mm² die is the size of a full reticle and wields 21 billion transistors. Here we see the massive die on Nvidia's new SXM3 package found in the DGX-2 system. The updated package runs at 350W (a 50W increase over the older model) and comes with 32GB of HBM2.

Nvidia announced that it would bump all new Tesla V100's up to 32GB of HBM2 memory, effective immediately. The nagging memory shortage has been a key factor in the current GPU shortage, so the Tesla V100's increased memory allocation could exacerbate the shortage. Nvidia is also taking the unusual step of suspending sales of the 16GB model, so all new models will feature the increased memory capacity.

Several blue-chip OEMs (like Dell EMC, HPE, and IBM) have announced that new systems with the beefed-up card will appear in Q2. So we expect the units will fly off shelves as the data center continues its transition to AI-centric architectures.

Hitting The NVSwitches

Nvidia's new NVSwitch is a breakthrough achievement for the company that allows it to tie 16 GPUs together in the DGX-2 server. The system features 12 NVSwitches that deliver an aggregate of 14.4TB/s of bandwidth.

Nvidia has developed its NVLink fabric to sidestep PCIe's bandwidth limitations, but scalability is limited due to the six NVLink ports available per GPU. The new NVSwitch bypasses those limitations with 18 8-bit links per switch that deliver a total of 300GB/s of bandwidth between each GPU. That's 10 times the speed of standard PCIe 3.0 connections. Huang equated this to having the equivalent of 200 NICs connecting the DGX-2's top and bottom trays.

The switches feature two billion transistors fabbed on TSMC's 12nm FinFET process. The resulting fabric uses memory semantics to avoid disrupting the existing GPU programming models. The fabric is a non-blocking crossbar, meaning it delivers the full 300GB/s of bandwidth to each GPU regardless of how many are actively transmitting data, and provides a unified memory space that allows for larger AI models. As with all switches and fabric topologies, there will be a latency penalty for traversing the interconnect, but Nvidia hasn't shared specifics yet.

Let Me Do That For You

Nvidia's new Isaac SDK is a suite of libraries, drivers, APIs and other tools designed to allow manufacturers to develop next-generation robots that leverage AI for perception, navigation, and manipulation. 

Nvidia's demo showcased a 7DoF robot from ABB opening and closing cabinet drawers in real time. The system uses a simple Xbox Kinect sensor to detect its environment, highlighting how AI can enable the use of cheaper components for next-generation robots. Nvidia also debuted its new Isaac Sim, which is a testing suite that allows developers to take their new robots for a spin in a simulated testing environment, thus saving us humans from costly accidents--or robot uprisings.

Revolutionizing The Warehouse

If you're wondering where all of the GeForce GTX 1060's have gone, we've got your answer. Kinema Systems brought its rather large robot to the show to demonstrate how it can detect and unload assorted items from a pallet, and it's all powered by that popular graphics card.

Multi-SKU picking involves sorting and moving packages of various sizes and weights from a randomly-packed pallet. In the past, training robots for these types of tasks was almost impossible because they couldn't sense the difference between different boxes.

AI changes that. Kinema trains deep-learning models to detect boxes and identify them automatically with information the system gathers online. The system then employs a video camera above the robot and runs an inference workload to analyze the video output, thus allowing the robot to detect the size, shape, and weight of the randomly-placed boxes. The company uses a GTX 1060 to power the robot. Nvidia recently changed its EULA for consumer devices to forbid their use in data centers. But we aren't aware of any restrictions on industrial use, so more of our gaming cards could be destined for the local robot-equipped warehouse.

Step Aboard The Holodeck

Jensen Huang debuted the photo-realistic Holodeck during last year's GTC keynote. This year's GTC found the company opening the new technology up to the public. Warner Brothers' Ready Player One movie debuts this week at theaters nationwide, but Nvidia provided a sneak peek with a VR rendition of "Aech's Basement," which is an escape room that requires teams of four to interact in virtual reality to solve various puzzles that eventually allow them to escape. Nvidia also had Koenigsegg Automotive AB's Regera Hypercar demo open to the public.

Quadro GV100 GPU, Yours For $9,000

Nvidia also debuted its new Quadro GV100 GPU at this year's GTC. The latest Quadro brings the Volta architecture to Nvidia's line of professional graphics cards. The GV100 comes packing 32GB of HBM2 memory, 5,120 CUDA cores, and 640 Tensor cores. Nvidia claims the beastly card delivers up to 7.4 TeraFLOPS of double-precision and 14.8 TeraFLOPS of single precision. Nvidia says the card can also provide up to 118.5 TeraFLOPS of deep learning performance (a measure of Tensor Core performance).

The company plans for workstation users to leverage its RTX technology. This will open up real-time raytracing to broader use cases, such as media creation.

If one GV100 isn't enough, Nvidia has also infused the new cards with NVLink 2, which means you can combine two cards into a single unit (much like SLI). Pairing two of the cards will cost you, though. A single card retails for $9,000.

First We Map The Site, Then We Build

No, that's not an Nvidia-branded bulldozer. AI pioneers foresee automation eventually breaking free of the bounds of standard vehicles and taking over construction equipment. To the relief of construction workers everywhere, that future is still off on the distant horizon. For now, enterprising companies like Skycatch are using intelligent drones (pictured to the left) to map construction sites using Nvidia's Jetson development kit.

The DRIVE PX Pegasus

Nvidia had its beastly DRIVE PX Pegasus module on display. Nvidia designed the supercomputer to power the latest Level 5 autonomous vehicles. The system comes outfitted with dual Tesla V100's and Xavier SoCs. The Pegasus crunches through 320 trillion operations per second to analyze the crushing amount of data flowing from the 360-degree cameras and LIDAR.

The system only consumes 500W of power, which is important because power consumption equates directly to mileage in electric vehicles. Pegasus is still in development. The final version will feature two of Nvidia's unnamed next-generation GPUs instead of the GV100 modules. Nvidia claims Pegasus will be the first Level 5 system to attain ASIL-D certification, which is the automotive industry's most stringent integrity certification.

The DRIVE Xavier

DRIVE Xavier is Pegasus's smaller brother. This system features Nvidia's Xavier SoC that comes packing 7 billion transistors built on the 16nm FinFET process. The entire system sips only 30W of power and features a custom eight-core ARM64 CPU and a Volta GPU with 512 cores. It also comes outfitted with computer vision accelerators and 8K HDR video processors.

Ray-Traced Audio Is A Thing

Real-time ray-traced video was all the rage at GDC and GTC 2018, but ray tracing can also be used for acoustics. This technique uses GPU acceleration to map out how sound travels in various rooms, such as large open spaces or restrictive small offices, to deliver a more realistic audio experience. Unlike previous audio solutions, ray tracing allows the compute engine to conduct these operations in real time without using pre-computed filters.

Isaac Powers Carter

As we covered earlier, Nvidia announced its Isaac SDK at the show. The company also had its new Carter reference design on the show floor. This diminutive little robot is designed to cart around objects in various environments, so it essentially serves as a delivery robot. Nvidia envisions small robots like these undertaking various tasks in busy environments. Carter isn't quite ready to navigate the overflowing show floor though. This was a static demo.

Truckers, Beware

Our picture of Einride's autonomous T-Pod delivery vehicle doesn't do justice to its sheer size. The massive vehicle is designed from the ground up to be an autonomous delivery truck; it doesn't even have a cab.

The electric truck can only travel 124 miles between charges for now, much to the relief of long-haul truckers everywhere. The truck can drive by itself on highways, but requires human assistance (via remote control) on city streets. The company plans to begin selling the new trucks this fall and hopes to have 200 trucks on the road in Sweden soon.

GPU Processing Comes To RAID

Nyriad came to the show with an interesting demo that highlights how GPUs can be used for unique applications. In this case, the company uses Nvidia's GPUs to calculate RAID parity.

Parity calculations allow RAID arrays to protect data in the event of a device failure, but the parity calculations result in decreased performance until the failed device is replaced and the RAID controller reconstructs the data. Nyriad's solution uses the massive parallel computing power of a GPU to provide an incredible amount of resilience during drive failures, all while delivering the same amount of performance.

Things Get AIRI

Pure Storage had its new AI-Ready Infrastructure (AIRI) demo on the show floor. The system features four of Nvidia's DGX-1 systems, with eight Tesla V100's apiece, fused together with Pure Storage's FlashBlade. Merging any form of storage with remote compute always incurs some type of latency penalty, but Pure uses Nvidia's GPUDirect RDMA and 100GbE Arista switches to reduce the impact. The system uses Nvidia's GPU Cloud deep learning stack and Pure's AIRI scaling toolkit to provide up to four PetaFLOPS of performance in a single rack.

Exploring Big Basin

Facebook's Big Basin is an open-source design that the company shares through the OCP (Open Compute) initiative. The primary box, called a JBOG (Just-A-Bunch-of-GPUs), houses eight V100's connected via NVLink in a hybrid mesh similar to the DGX-1. The JBOG slides into the main housing and is installed into a server rack.

The unit to the right houses two Xeon processors and all of the normal components of a server. The flexible PCIe 3.0 x8 cables connect the server to the JBOG. This design allows Facebook to upgrade the CPUs and GPUs independently of each other, thus providing upgrade flexibility as more powerful models come to market.

Holographic Data Storage

Holographic Data Storage

Holographic data storage involves a different approach compared to today's standard disk- and flash-based processes, recording data throughout the volume of a thick, photosensitive optical material. Once perfected, this method could allow for recording multiple images--perhaps thousands--in the same space by placing the laser at different angles.

For recording, the laser is split into a reference and a data-carrying beam. The latter essentially picks up information arranged in a checkerboard pattern as it passes through a spatial light modulator. It then interferes with the reference beam through the material, storing the data. A good explanation of how holographic data is manipulated can be found here, at Lucent Technologies' Bell Laboratories Web site.

Volumetric 3D Display

Volumetric 3D Display

The idea behind a volumetric 3D display is to present images that appear to have three physical dimensions, regardless of the viewpoint. This is a completely different approach than the current method of converting flat, 2D images, and then using special glasses or multiple screens to trick the naked eye.

Volumetric 3D images would not only appear solid from one point of view--as in standing directly in front of the display--but would look the same when moving around the display a full 360 degrees. While this technology doesn't offer 3D visuals that literally reach out to your face, volumetric 3D displays in the living room would mean a more realistic experience.

In October 2009, Sony revealed such a device, although it was rather small. Viewers watched the object rotate within the display or walked around the device to see the object in its entirety, as if it stood within a virtual cage.

Self-Driving Automobiles

Self-Driving Automobiles

Self-driving vehicles would be part of that evolutionary step from pre-programmed AI to cognitive mechanics. These vehicles would still require the rider to physically interact with the car. But instead of driving, the pilot would merely step in and select a destination with an oral command. The vehicle would then deliver the rider using an auto-pilot system.

This would mean that the AI would need to have complex cognitive skills in order to make quick decisions, such as avoiding lost puppies wandering around in the middle of the road or letting the rider exit the vehicle while in motion. According to General Motors, the first driverless car could be on the road by 2018. The company revealed its plans a few years back, and said it would begin testing by 2015. Former General Motors Chief Executive Rick Wagoner said the biggest hurdle wouldn't be the technology, but rather liability laws and government regulations. “This is not science fiction,” Larry Burns, GM’s vice president for research and development, said in 2008. Since the report, other car manufacturers, including Nissan, Honda, and Audi have begun development of automated cars.

Augmented Reality

Augmented Reality

We've seen some augmented-reality applications already begin to take shape. But what if it was standard technology, embedded in your sunglasses? Wherever you went, whether it would be on foot, in an automobile, or even on a bicycle, you would receive information relevant to your surroundings. Need to find a local toilet? An augmented-reality device could show you where. Have you wandered too far in the woods? An enabled device could point you in the right direction, drawing an actual virtual road on the ground right beneath your feet. We previously reported on plans in the automotive market to install augmented-reality technology in windshields and rear-view mirrors. Connected to an onboard computer, the augmented-reality fields would display most of what you would find on a normal dashboard, along with weather conditions, distance to your destination, the distance to the next car, and more.

Hardcore Virtual Gaming

Hardcore Virtual Gaming

This really should need no explanation--we've seen examples of virtual gaming in various sci-fi movies and TV shows. Although 3D is slowly starting to creep into our PC and console titles, virtual gaming is already taking shape on a more engrossing level. A good example is the Virtusphere, which allows players to climb inside, don special glasses, and virtually walk and run during the game. One day we'd like to see a real holodeck-like experience, allowing us to play games in artificial, replicated environments without the need for special glasses or other connected devices. In the meantime, this small step will certainly do.

Atomic Memory

Atomic Memory

Imagine if you could store a bit of data--binary zero or binary one--on one individual atom. Now imagine a grain of sand composed of trillions of binary bits of information stored on each atom--a definite far cry from using millions of atoms just to store one bit of data. The idea of atomic memory has been around since the late 1950s, while scientists have been able to move closer to atomic storage during the past 10 years. By 2001, scientists were able to store one bit of data on a single atom, but it had to be shielded from its neighboring atoms by placing it within a five-by-four cell of atoms. The end result took 20 atoms to secure one bit of data. Close, but not the 1:1 ratio scientists eventually want to achieve.

Shape-Changing Alloys

Shape-Changing Alloys

Shape-changing alloys could be a good or bad thing. On one hand, the technology could make automation quick and simple. For example, developing an alloy that could shape itself into a particular component, whether that would be an army knife or a key part in an automobile engine. On the other hand, this could eliminate jobs by making hand-crafted work redundant.

However, the shape-changing technology is already in existence, developed by scientists at Harvard and MIT. In its present state, the shape-changing alloy is folding sheets into small shapes like boats and airplanes. But that's merely the beginning. Could it be possible to use the technology to build real boats and airplanes within the next 10 years?

Home-Based Nuclear Generators

Home-Based Nuclear Generators

In 2007, Toshiba revealed plans to sell nuclear reactors for every home across the globe based on a prototype it had developed. Measuring 20 by six feet, the 200 KW generator was fail-safe, fully automatic, self-sustaining, and could even serve the power needs of entire city blocks, Toshiba said.

Instead of traditional control rods, the personal reactor used special liquid-lithium reservoirs so it could last up to 40 years and, once commercialized, would cost consumers around 5 cents per KW/h of electricity consumed. For whatever reason, the device has not reached homes yet. However, earlier this year, Bill Gates (via TerraPower) announced a possible joint venture with Toshiba to refine and develop the original design. Gates' technology was said to run without refueling for up to 60 years on depleted uranium. Future generators are expected to be safer, smaller, and more socially acceptable than what Toshiba originally had in mind. Unfortunately, the technology isn't expected to go commercial until the early 2020s.

Petroleum-Free Automobiles

Petroleum-Free Automobiles

After the recent BP spill in the Gulf of Mexico, it's clearer than ever that our dependence on fossil fuels needs to be at least partially alleviated by alternative energy sources. We've already seen hybrid vehicles enter the market, utilizing conventional fuels and electricity.

However, gas consumption isn't the only factor--manufacturers use petroleum during the process of building automobile parts. Recently, Ford revealed its aim to eliminate petroleum altogether in its processes by using bio-based soy foam. According to Debbie Mielewski, Ford technical leader of plastics research, the company conserves around 1.5 million pounds of petroleum annually when using soy-based foam to manufacture parts. "While these numbers are not big compared to the burning of fuel, we think that everything we can do, without compromising performance and durability, moves us in the right direction for the long term," Mielewski said.

Liquid Breathing

Liquid Breathing

The idea behind liquid breathing is to allow humans to venture into the depths of the ocean--or even the cold black void of space--without the need to breathe air in gas form. Instead, explorers would wear a device that pumps oxygen-rich liquids into the lungs. This could help mitigate some of the pressure-based challenges encountered by divers in deep water and the problems associated with returning to the surface.

In the medical field, the application is already in use--premature babies born with underdeveloped lungs have used liquid breathing to stay alive until their lungs are capable of taking in air. Liquid breathing has also been used to help adults with acute respiratory failure.

Wireless Energy

Wireless Energy

The concept of wireless energy came about during the late 1800s, while an article featured on Space.com sparked an interest with mainstream media as it discussed wireless energy transfers in space. Now, 10 years later, we've recently reported on devices in development that actually do recharge and power other devices without the need for cables and wires.

The process works by plugging a copper coil into a wall outlet while another copper coil is plugged into an electronic device. Both coils are tuned to the same frequency. The wall coil generates an electromagnetic field and resonates with the device coil. This resonant coupling is what generates the energy for the device. But don't fret--this process isn't harmful to humans, as apparently our bodies don't respond to magnetic fields in a negative way. Much like WiFi is now available in restaurants, gas stations, book stores, and many other public places, resonant coupling is expected to be just as abundant. Many companies are expected to hit the market with wireless power technology by the second quarter of 2011.