Ai of 2028 and 2030 will look vastly different from Ai in 2024 or 2026 <\/h2>\n\n\n\n
It is my contention that what the common man talks about as ‘Ai’, is at the beginning of the beginning!<\/p>\n\n\n\n However, as the models like NemoTron and Nanobanana etc take on both more complex and compute heavy workloads, there is a ‘different’ Ai’s that we need to consider: Virtual world <\/em>and Physical world<\/em> Ai. <\/p>\n\n\n\n The largest companies in the world are now scaling up for the demand driven, not just by democratising LLM’s (Text based Ai) but for getting infront of the wave of Virtual world <\/em>and Physical world<\/em> Ai. <\/p>\n\n\n\n This will likely see a trippling of Data Centre compute across the globe, starting with and explosion in the number and size of data centres on the North American continent.<\/p>\n\n\n\n To understand why datacenter demand will quadruple year-over-year, we have to look at how data utilizes a microchip.<\/p>\n\n\n\n Text inference is fundamentally memory-bound<\/strong>. When you type a prompt into an LLM, the GPU spend a massive amount of time waiting for model weights to load from high-bandwidth memory into its cache. Because the chip\u2019s mathematical units are idling during this wait, data centers can “batch” dozens of different users’ text requests simultaneously onto a single GPU.<\/p>\n\n\n\n Video is entirely different. Video is compute-bound.<\/strong><\/p>\n\n\n\n When an AI model generates video (like OpenAI\u2019s Sora or Meta\u2019s Movie Gen), it isn’t processing words; it is managing millions of spatio-temporal visual patches across multiple dimensions.<\/p>\n\n\n\n If we assume that over the next decade, just half of the global creator economy’s daily video uploads (such as YouTube or TikTok) leverage AI augmentation or generation, the math dictates a baseline requirement of hundreds of thousands of top-tier GPUs pinned 24\/7<\/strong>. That represents hundreds of megawatts of continuous power just to act as the “printing press” for the next generation of media\u2014before a single viewer hits play.<\/p>\n\n\n\n While creative video drives immense data volume, moving AI into scientific discovery, 3D CAD engineering, and real-world modeling<\/strong> drives extreme computational density.<\/p>\n\n\n\n When an AI helps design an airplane wing, a consumer electronics chassis, or an autonomous vehicle world-model, the operational mandate shifts from visual plausibility<\/em> to deterministic physical precision<\/em>.<\/p>\n\n\n\n Text and video models thrive on low-precision math ($FP8$ or $FP16$) to maximize speed. If a video model makes a rounding error, a pixel shifts slightly, and the human eye ignores it.<\/p>\n\n\n\n But if an engineering AI makes a rounding error while calculating stress tensors on a bridge or a biomedical implant, structures fail and lives are lost. Engineering AI requires Double Precision ($FP64$) mathematics<\/strong>. When forced to compute at this level of accuracy, the theoretical throughput of standard AI chips drops drastically, requiring a massive increase in physical silicon footprint to process the same volume of work.<\/p>\n\n\n\n Furthermore, an engineering AI cannot operate in a vacuum. It triggers a continuous computational loop: the AI generates thousands of geometric permutations (represented as dense, complex boundary graphs), which must immediately be fed into traditional multi-physics solvers (Finite Element Analysis and Computational Fluid Dynamics) to verify structural integrity.<\/p>\n\n\n\n This requires an entirely new class of “Industrial Simulation Factories”<\/strong>\u2014datacenters featuring massive internal network switching capacities to allow thousands of computing nodes to sync instantly without data starvation.<\/p>\n\n\n\n The sheer weight of these workloads will eventually push computing past the boundaries of Earth’s grid limitations. While space-based datacenters would be useless for low-latency text chats, they possess a massive structural advantage for the world\u2019s hardest asynchronous workloads.<\/p>\n\n\n\n Because the market is treating AI as a monolithic block of conversational software, the structural enablers of this multi-dimensional transition are deeply mispriced. As workloads double and quadruple year-over-year, the real value accumulates at the physical layers:<\/p>\n\n\n\n Text is read-heavy; spatial digital twins and video generation are intensely write-heavy<\/strong>. Logging parametric CAD histories, multi-layered physics simulation states, and massive video patches requires specialized, parallelized NVMe storage architectures capable of millions of IOPS. Companies that control high-performance enterprise data storage fabrics are sitting on an unappreciated goldmine.<\/p>\n\n\n\n An AI cannot invent the physical world without a ground-truth validator. The legacy software monopolies that own the proprietary multi-physics simulation engines (FEA\/CFD) hold a structural bottleneck. Every generative engineering tool must pay a mathematical licensing tax to these engines to verify its designs.<\/p>\n\n\n\n As clusters are forced to act as single, synchronous machines to process massive geometric graphs, the bottleneck shifts from chip speed to interconnect bandwidth and high-precision processing<\/strong>. Ultra-low latency network switches, electro-optical interconnects, and high-bandwidth memory (HBM) suppliers will experience structural tailwinds that standard software-as-a-service providers cannot duplicate.<\/p>\n\n\n\n We are moving away from the era of centralized “chatbots” and entering the era of generative physical reality<\/strong>. The total mega-capacity of datacenters required to act as the design, simulation, and rendering engine for the physical and visual world is staggering.<\/p>\n\n\n\n The infrastructure boom isn’t nearing its end; it is merely graduating from words to worlds. The picks, shovels, chips, and storage architectures enabling this multi-dimensional regime shift represent the defining investment opportunity of the next two decades.<\/p>\n\n\n\n![\"\"](\"https:\/\/i0.wp.com\/murrayslatter.me\/wp-content\/uploads\/2026\/06\/IMG_4636.jpg?resize=640%2C768&ssl=1\")
![\"\"](\"https:\/\/i0.wp.com\/murrayslatter.me\/wp-content\/uploads\/2026\/06\/Ai-Data-Centre-Spend.jpg?resize=640%2C401&ssl=1\")
<\/figure>\n\n\n\n1. The Physics Shift: Memory-Bound Text vs. Compute-Bound Video<\/h2>\n\n\n\n
\n
2. Entering the Physical World: The Heavy Industrial “Simulation Factory”<\/h2>\n\n\n\n
[ Traditional LLM Datacenter ]\nLow Intra-Network Traffic -> Many Independent Users -> High Memory Bottleneck\n\n[ AI Industrial Simulation Factory ]\nMassive Inter-GPU Traffic -> High-Precision Math (FP64) -> Total Compute Saturation\n<\/code><\/pre>\n\n\n\nThe Precision Wall ($FP64$)<\/h3>\n\n\n\n
The Dual-Engine Loop<\/h3>\n\n\n\n
3. The Ultimate Edge: The Case for Space-Based Datacenters<\/h2>\n\n\n\n
\n
4. The Picks and Shovels: Where the Alpha Generates<\/h2>\n\n\n\n
The Write-Heavy Storage Fabrics<\/h3>\n\n\n\n
The Physics Verification Moats<\/h3>\n\n\n\n
High-Precision and Interconnect Silicon<\/h3>\n\n\n\n
The Next 20 Years<\/h2>\n\n\n\n
Connect to discuss more<\/h2>\n\n\n