Computational power is becoming a critical resource as tech giants actively strive to create and deploy artificial intelligence supernodes. This frenzy is driven by competitions to increase model parameters, which creates unprecedented demand for infrastructure.
Companies, including Huawei, ZTE, H3C, and Sugon, are competing to build and deploy AI supernodes scalable from 1024 to 100 thousand GPUs. The increasing demands, driven by models like Kimi K3 and GLM-5, push computational capacity to its limits. Supernodes have replaced models as the central focus of the WAIC 2026 exhibition, as computational capability has become the main limiting factor in AI development. The shift from clusters with 8 cards to clusters of 100 thousand cards makes supernode capacity a direct competitive advantage; consequently, H3C demonstrated a 58% growth, and Inspur showed 41% growth over three weeks.
The most evident demand pressure is observed at the forefront of model development. GLM-5 models from Zhipu AI and Kimi K3 from Moonshot AI, boasting 2.8 trillion parameters, have been forced to raise prices and restrict subscriptions due to the enormous inference demand. Deploying the K3 model requires a supernode configuration with more than 64 GPUs, setting a minimum computational threshold for advanced models. In response, Zhipu acquired Zhongke Jiahe and built a domestic AI data center with a capacity of 1 GW, leading to a 37% rise in the company's stock in one day. The math of the situation is telling: industry participants report computation costs of approximately $0.80 per dollar of revenue from the model, yet every dollar of revenue corresponds to a multiple increase in valuation multipliers, making the scale of computation the primary lever for growth.
Existing achievements include the use of Huawei Ascend 950 with 1024 NPUs and deployed over 750 A384 sets, as well as ZTE OEX with 128 GPUs per rack and a cluster of 10 thousand. Sugon introduced Dawn 8000—the first domestic system with 100 thousand cards. H3C is building an AI factory capable of delivering 240–300 kW to a test point. Technological directions diverge: Huawei utilizes a proprietary full-stack model—from chip to interconnect and cooling—allowing for deep optimization across layers but requiring colossal R&D investment. ZTE and its partners prefer an open approach using multiple chips to avoid dependence on a single vendor and ensure flexibility for clients. Other players, such as Moore Threads, Pingtouge, and Baidu Kunlun, focus on specific cloud or vertical scenarios, paying close attention to inference efficiency and economic performance. The overarching challenge lies in maintaining system integrity across the architecture of chips, interconnect, software stack, and operational tools, as advantages gained in one area cannot compensate for systemic gaps.
Model parameters have not yet reached the 2.8 trillion mark; forecasts indicate reaching 10 trillion within two years. The consensus was clear: supernodes define the future. Companies that cannot secure computational power at the supernode scale will be structurally excluded from the AI frontier competition.
Researchers have discovered a new species of long-necked dinosaur in western India, which was distinguished by a unique horn-like crest on its tail. This dinosaur has been named Ceratocaudia antiqua. The identification was conducted by a group of scientists from the Indian Institute of Technology Bombay (IIT Bombay) and the Birbal Sahni Institute of Palaeontology following excavations in the Caladongar formation in the Kachchh basin.
This small giant lived approximately 170–180 million years ago, during the Jurassic period. It reached only about seven meters in length, significantly smaller than its better-known relatives, such as Brachiosaurus, which reached twenty meters. The team excavated part of the skeleton near the village of Amrapur in the Gujarat district of the state of Gujarat.
Using 3D laser scanning and computed tomography (CT) technologies, researchers were able to reconstruct the animal's anatomy. The analysis indicated that the creature likely used its tail as a defense mechanism against predators. Despite its small size, bone analysis suggests that this particular specimen was not a young animal but an elderly representative of its species, as signs of remodeling and even early arthritis were found in the bones of a limb.
Ceratocaudia is distinguished by several physical traits that set it apart from its closest relatives. The most noticeable feature is the curved, horn-like crest on the anterior vertebrae of the tail. Furthermore, it possessed a specific indentation on its cervical vertebrae and unique shapes in the bones of its lower limbs—the tibia and fibula—which were not observed in other sauropods. Its caudal vertebrae also show a rare transition from flat joints to 'hinge' type joints closer to the tip, which is typically characteristic of later dinosaur species.
The name Ceratocaudia antiqua reflects these physical and historical characteristics: the genus derives from the Greek words 'cerato' (horn) and 'caud' (tail), indicating the unique spine on the vertebra. The species name 'antiqua' is Latin and means 'old' or 'ancient,' emphasizing that it is one of the earliest known macronarian dinosaurs—a group of long-necked giants with large, angular skulls and wide nostrils. This find represents an important piece of the evolutionary mosaic for paleontologists, suggesting that this successful group of dinosaurs spread across the planet much earlier than previously thought. Prior to this discovery, the record of such dinosaurs in India during the Jurassic period was extremely sparse, and Ceratocaudia is now the earliest diagnosed member of its lineage in India and one of the oldest in the world.
Due to a global shortage of RAM spurred by the demands of artificial intelligence, the cost of various gadgets, from Apple's MacBook line to Xbox and PlayStation consoles, has significantly increased in South Africa, making 2026 a challenging period for buyers.
The main culprit behind this phenomenon is the global scarcity of RAM, unified memory, and storage space, which is caused by the insatiable demand of AI workflows. Modern users require more memory to run resource-intensive applications such as ChatGPT, Google Gemini, Microsoft Copilot, Claude, and Perplexity AI.
Although Apple was the last major tech corporation to raise prices, it managed to avoid an 'RAM apocalypse' until recently thanks to long-term supply contracts signed in advance. However, over the past year, Windows-based laptops have shown multiple price increases, which briefly made Apple a more attractive brand, but now the old order has been restored, and Apple is once again setting high prices for its hardware.
For instance, the 13-inch MacBook Air M5, which cost R19,999 before the recent price hike, is now selling for R26,499, representing a sharp jump of over 32%. Furthermore, Apple's most affordable 2026 model, the MacBook Neo, has risen from an initial price of R11,999 to R13,999, an increase of R2,000 or 16.67% in a short time. This means that the entry-level MacBook Air is almost twice as expensive as the Neo, and budget buyers are likely to opt for the Neo despite performance compromises.
The problem is not limited to computers. The Xbox Series S and Series X consoles, released about six years ago, have also seen price increases. The Series S model launched at $300 (equivalent to approximately R5,100 at the standard exchange rate), and the Series X cost $500 (about R8,600). Although local import costs initially made them slightly more expensive in South Africa, the rise was not as dramatic. The author purchased an Xbox Series S in March 2023 for R5,999, but today the same console costs R9,998, an increase of R4,000 for a six-year-old device.
Instead of the expected price drop, the price continues to climb. To get the top-tier Series X, it will cost around R14,999. When the author bought the Series S, the Series X price of R11,000 seemed too high given the R5,000 difference between the models, which would allow buying almost two Series S units for the price of one. Microsoft is not alone in this situation: Sony's base PlayStation 5 Digital Edition has risen from $400 at launch in 2020 to $500 by 2025. PlayStation maintains its lead in the console race, with about 75 million units sold compared to 31 million for Xbox, meaning millions of gamers are feeling the same pressure.
Thus, purchasing almost any new electronics during 2026 will be expensive. However, there is uncertainty about when prices might stabilize, making the current moment both the best and worst time to make a purchase.
A new investigation offers a possible explanation for the enigma of the Little Red Dots, intriguing phenomena detected by the James Webb Space Telescope (JWST). Researchers propose that these objects did not simply disappear in the primordial universe, but rather evolved to become globular clusters, which are vast groupings of densely packed stars, like those found in the Milky Way.
The theory establishes an analogy with the evolution of terrestrial dinosaurs: just as many dinosaurs were not extinct but gave rise to current birds, the Little Red Dots could have survived, transforming into structures that persist in the contemporary universe.
The Little Red Dots began to attract astronomers' attention in 2022, when James Webb started detecting them in large numbers at a phase corresponding to approximately 600 million years after the Big Bang. The mystery lies in the fact that these bodies appear to disappear before the universe reaches about two billion years old.
Several theories have been raised about their nature, including the possibility of being 'black hole stars,' meaning black holes surrounded by extensive clouds of gas and dust. However, the current research points to another path: these objects could be globular clusters in a formation stage, possessing a supermassive star at their core.
This hypothetical type of star would have a mass between one and ten thousand times that of the Sun and a very short life, but it would be capable of generating an appearance similar to that of the Little Red Dots observed by the telescope. John Chisholm, lead researcher from the University of Texas in Austin (USA), stated that these objects may not just be a strange new population from JWST unrelated to the current universe, but rather that they may endure beyond the primitive universe, evolving into something more known.
Chisholm added that the Little Red Dots could be galaxies, involve black holes, or represent something even more unexpected, and that their work demonstrates that the formation of globular clusters with supermassive stars must be considered in this debate.
Globular clusters are typically located in large galaxies and manage to concentrate millions of ancient stars in extremely compact areas. The Milky Way, for example, contains at least 150 of these clusters. Although well-studied, their origin remains a topic of scientific discussion.
Danielle Berg, also from the University of Texas in Austin, mentioned that astronomers usually observe these clusters after billions of years of evolution. She explained that at this point, their massive stars have already gone extinct, their gas has dispersed, and dynamic processes have modified their masses and structures, making it difficult to reconstruct the initial conditions of their formation.
Scientists believe that the stars within globular clusters formed approximately at the same time, during the early stages of the universe, when the cosmos was mainly composed of hydrogen, helium, and small amounts of heavier elements, called metals by astronomers.
However, many of these stars exhibit a peculiar chemical composition. They show a high abundance of helium and elements such as nitrogen, sodium, and aluminum, while showing lower levels than expected of carbon, oxygen, and magnesium. Mike Boylan-Kolchin, a member of the team, indicated that this chemical signature points to extreme nuclear fusion conditions.
He specified that this particular pattern signals nuclear fusion at temperatures much higher than those found in the cores of common stars, even massive ones. A supermassive star is exactly the type of environment capable of generating this chemical combination.
According to the researchers' proposed model, these supermassive stars would emerge in environments of extremely high density during the formation of the first globular clusters, where stellar collisions and mergers would occur frequently. Despite their colossal size, they would have a lifespan of only about one million years, an infinitesimal period compared to the 4.6 billion years of the Sun's life.
Even with this short existence, these stars would be capable of producing the chemical elements necessary to justify the composition observed in globular clusters. After dying in supernova explosions, these elements would be released into space, being used in the creation of new generations of stars.
Chisholm concluded that this would also explain the disappearance of the Little Red Dots. In their model, the supermassive star that makes the object look like a Little Red Dot lives for a short time; after its death, the object may cease to resemble a Little Red Dot, even if the cluster itself persists for billions of years.
In addition to chemical analysis, the team identified other signs linking the Little Red Dots to globular clusters. The distribution of these objects in the early universe mirrors the distribution currently seen in globular clusters. Evolution models also suggest that the estimated masses of the Little Red Dots could naturally progress to reach the masses of existing globular clusters today.
An additional relevant factor is the timing of their appearance: the Little Red Dots appear about 600 million years after the Big Bang, coinciding with estimates for the beginning of globular cluster formation. Despite the evidence, the researchers warn that the hypothesis still lacks definitive confirmation. Boylan-Kolchin stated that although there is no conclusive proof that the Little Red Dots are globular clusters, this idea would explain several surprising and distinct observations. The study is in a preliminary version on the scientific repository arXiv, implying that its results have not yet undergone peer review.
