Oh look…the AI Asian Guy’s voice is being used to manipulate more markets. https://x.com/saylordocs/status/2019443901569069560?s=20
1 Like
that “forehead slap
moment” when you realize…
WORLD WAR II ~ TOKYO ROSE
WORLD WAR III - AI ASIAN GUY
3 Likes
Speaking of AI, this morning I asked Grok how spaced based AI installations vs terrestrial installations would affect physical silver demand. The following was its response:
Space-based AI, often conceptualized as orbital data centers or satellite constellations equipped with AI hardware (e.g., Google’s Project Suncatcher, Starcloud’s gigawatt-scale clusters, or Axiom Space’s orbital nodes), is an emerging paradigm aimed at scaling compute power beyond Earth’s constraints. These systems leverage constant solar exposure in sun-synchronous orbits, radiative cooling in vacuum, and modular satellite designs to achieve efficiencies in energy, deployment speed, and resource use that aren’t feasible terrestrially. However, the impact on physical silver demand—a metal critical for electrical conductivity in electronics, AI chips, circuit boards, switches, and photovoltaic (PV) cells—hinges on balancing surging industrial needs against potential optimizations.Key Drivers of Silver Demand in AI and Related TechSilver’s role in AI infrastructure is multifaceted and growing rapidly:
-
Electronics and AI Hardware: Silver is irreplaceable in many high-performance components due to its superior conductivity and resistance to overheating. It’s used in printed circuit boards (PCBs), membrane switches, electromagnetic interference (EMI) shielding, multilayer ceramic capacitors, and high-speed connectors. AI-specific applications include neural processing units (NPUs), GPUs/TPUs (e.g., NVIDIA H100 or Google TPUs), and advanced chip packaging. Data centers alone consumed an estimated 350 million ounces of silver in 2025 (over 50% of global mining supply), with AI chips requiring 15-20% more silver than traditional electronics. Projections show industrial silver demand hitting record highs of 465-500 million ounces in 2026, driven partly by AI expansion.
-
Solar PV Integration: Solar panels are a massive silver sink, accounting for 29% of industrial demand in 2024 (up from 11% in 2014), with consumption reaching 200 million ounces. Each PV cell uses silver paste for front-side electrodes to conduct electricity efficiently. A 500-megawatt solar array (enough for one hyperscale data center) requires about 300 metric tons (9.6 million ounces) of silver. Global solar installations are forecast to rise, pushing PV silver demand to 195-261 million ounces in 2026 and potentially 321-450 million ounces by 2030 under aggressive deployment scenarios.
-
Broader Ecosystem: EVs, robotics, drones, power grids, and sensors—all integral to AI ecosystems—further amplify demand. For instance, EVs use 25-50 grams of silver per vehicle, and AI-driven electrification could boost automotive silver needs by 3.4% annually through 2031.
Total global silver demand is projected to stabilize at ~1.2 billion ounces in 2026, with industrial uses (including AI and solar) offsetting any dips in investment or jewelry. Supply, however, remains stagnant at ~1 billion ounces annually (mostly as a mining byproduct), leading to persistent deficits through 2030 and price pressures (silver surged 130% in 2025).How Space-Based AI Could Amplify Silver DemandMoving AI to space doesn’t inherently reduce material needs; it scales them up:
-
Massive Solar Requirements: Orbital data centers rely on vast solar arrays for “unlimited” power without batteries or grid ties. Starcloud’s proposed 4km x 4km orbital structure, for example, could replace 50 terrestrial data centers but would need enormous PV wings—potentially consuming millions of ounces of silver per facility. Google’s Suncatcher envisions TPU-laden satellite clusters linked by optical comms, all powered by solar. With higher solar irradiance in space (up to 1.4x Earth’s surface), panels generate more power per unit area, but the sheer scale for gigawatt-level AI training (e.g., training large language models) could still drive overwhelming demand. If space-based AI proliferates by the early 2030s, as predicted by Elon Musk and others, solar silver needs could explode, exacerbating forecasts of 10,000-14,000 metric tons (321-450 million ounces) annual PV demand by 2030.
-
Hardware Scale-Up: AI chips and servers in space still require silver for conductivity and reliability under radiation. Launching constellations (e.g., hundreds of interconnected satellites) means more hardware overall, not less. Radiation tolerance adds complexity, potentially increasing material use in shielding or redundant components.
-
Supply Chain Multiplier: Building and deploying these systems involves terrestrial manufacturing spikes in electronics, EVs (for ground support), and robotics— all silver-intensive. Short-term imbalances could force prices higher, as seen in X discussions where AI is blamed for silver shortages.
This trajectory points to “overwhelming demand,” with AI and space tech as key culprits in a structural supply gap.Counterarguments: Efficiencies That Could Obviate Large Silver QuantitiesSpace offers unique advantages that might mitigate or even reduce net silver needs compared to Earth-bound infrastructure:
-
Energy and Design Optimizations: Constant sunlight in sun-synchronous orbits eliminates downtime from weather or night, allowing 24/7 operation without energy storage. Radiative cooling in vacuum dissipates heat more efficiently than Earth’s air-cooled or water-cooled systems, enabling denser, more power-efficient chip packing. This could mean fewer servers/chips per compute task, indirectly reducing silver in hardware. Modular, clustered designs (e.g., Google’s “string of pearls” formations) facilitate rapid scaling without massive earthly footprints, potentially lowering overall material intensity.
-
PV Thrifting and Substitution: Solar industry trends like “thrifting” (reducing silver per cell via finer electrodes or multi-busbar designs) have already cut usage by 5-7% year-over-year, with 2026 PV demand forecast to dip despite rising installations. In space, advanced tech like cadmium telluride thin-film panels (which use little to no silver) or perovskite-based cells could dominate, as they prioritize efficiency and radiation resistance over cost. If adopted widely, this could cap PV silver demand at 29-41% of supply by 2030, avoiding the worst deficits.
-
Holistic Resource Savings: Space bypasses terrestrial bottlenecks like permitting, land use, and grid strain—which often require silver-heavy infrastructure (e.g., switchgear, circuit breakers). Orbital systems could process data closer to satellites (e.g., for Earth observation or edge AI), reducing the need for ground-based data centers and their silver footprint. Long-term, AI in space might enable breakthroughs in material science, like silver alternatives (e.g., copper-graphene composites) that aren’t viable on Earth due to environmental factors.
-
Economic Feedback Loops: Sky-high silver prices (already a risk per J.P. Morgan) could accelerate substitution in both space and Earth apps, forcing innovation. If space-based AI becomes cost-effective (Musk predicts within 5 years), it might displace silver-intensive terrestrial builds entirely, stabilizing demand.
These efficiencies are space-exclusive: Earth’s atmosphere attenuates solar power by ~30%, weather disrupts operations, and cooling demands massive energy (up to 40% of data center power). Space could thus “obviate” large quantities by enabling hyper-efficient, low-material designs that make AI compute more sustainable overall.Balanced OutlookSpace-based AI is more likely to create overwhelming demand for silver in the near term (2026-2030) due to the explosive build-out of solar-powered hardware. Deficits could widen to 14,000-18,000 metric tons annually if supply doesn’t ramp up, pushing prices toward $50-100/ounce and creating a “short squeeze” as AI, solar, and EVs compete. However, long-term efficiencies (post-2030) could obviate the need for vast quantities by leveraging space’s advantages for thrifting, substitution, and optimized architectures—potentially capping or reversing demand growth. This isn’t a binary; it’s a race between scaling ambition and innovation. If substitution wins (e.g., via non-silver PV), space AI could even ease pressure on silver markets. Watch for pilots like Suncatcher’s 2027 demo or Starcloud’s LLM training in orbit for early signals.
2 Likes
And let’s not forget the new laser weapons and in manufacturing. Silver mirrors offer superior reflectivity and can withstand high power densities without degrading. Silver’s high thermal conductivity can be used to manage heat generated by high power lasers.
1 Like
Personally, i believe the likelihood that AI will find a way around or through the supply side silver bottleneck is relatively high since by virtue of its relative scarcity at some point, and probably sooner rather than later, increasing deficits associated with current material supply will necessitate replacement with materials or technology that performs at least equally as well.
I think the key will soon be space based infrastructure that uses elements in unlimited supply. Hydrogen, and Helium for example which in the extremely cold temps of outer space (perhaps with enhanced solar or nuclear powered refrigeration) become excellent conductors, and are the most abundant elements in the Universe.
Once these types of technologies come online, I believe silver and gold will very likely be quickly dethroned as must have elements.
I also asked Grok AI about transmutation of existing elements through nuclear fusion (as a way to overcome material shortages of elements like silver and gold). This was its reply:
Yes, nuclear fusion technology can theoretically enable the creation of both silver and gold from other elements, but not through the direct fusion of lighter nuclei into these heavy elements in the way stars fuse hydrogen into helium or helium into carbon.
Direct Fusion Limitations
Standard nuclear fusion (combining light nuclei like hydrogen or helium) releases energy only up to iron (atomic number 26). Beyond iron, fusion becomes endothermic — it consumes more energy than it produces. Elements like silver (atomic number 47) and gold (atomic number 79) are far heavier, so they cannot be built efficiently through direct fusion chains in current or foreseeable fusion reactors. In nature, these elements form primarily via the rapid neutron capture process (r-process) in extreme events like neutron star mergers or certain supernovae, not ordinary stellar fusion.
How Fusion Technology Can Produce Them Indirectly
Modern fusion reactor designs produce massive fluxes of high-energy neutrons (especially from deuterium-tritium fusion). These neutrons can be used for neutron capture transmutation on target materials placed in or around the reactor blanket:
- For gold: A promising concept (proposed by companies like Marathon Fusion around 2025) involves bombarding the abundant mercury-198 isotope with neutrons from a fusion reactor. This creates mercury-197 (via neutron capture and subsequent reactions), which beta-decays into stable gold-197. Simulations suggest a large-scale fusion power plant could produce several tonnes of gold per gigawatt of thermal power per year as a byproduct, though this remains theoretical and unproven at scale as of 2026.
- For silver: Similar neutron irradiation can transmute elements like palladium or cadmium isotopes into silver via neutron capture followed by beta decay. While less discussed than gold, the principle is the same and feasible with fusion neutron sources.
These methods are transmutation via neutron reactions rather than pure fusion of light nuclei into heavy ones. They leverage fusion’s neutron output as a tool, much like how fission reactors have produced trace gold from mercury in the past.
Practical Reality
- Small amounts of gold have already been created artificially using particle accelerators or fission reactors (e.g., from bismuth or mercury), but the cost is astronomically higher than mining natural gold.
- Fusion-based approaches could be more scalable if commercial fusion power becomes viable, potentially turning reactors into dual-purpose energy + materials producers.
- However, no fusion reactor has yet achieved this at meaningful quantities — current experiments focus on net energy gain, not side-product synthesis.
In summary, direct fusion of light elements won’t make silver or gold, but fusion-powered neutron sources could enable their production from heavier starting elements like mercury (for gold) through transmutation. It’s modern nuclear alchemy — scientifically valid, but economically challenging and still emerging.
So naturally I asked how long it would take AI to replace Silver and Gold in AI infrastructure. The answer is a qualified 1 -2 year window:
4 Likes
Interesting articles. Thanks for posting them.
1 Like
What is your take on this article? It seems gold and silver are too expensive to make via fusion, it’s a few years away from massive space adaption. I think silver will be in short supply over the next 2-3 years and by the time miners get their new mines on line, silver demand will be down. But who knows. Diversification is the key.
Grok says making gold by fusion… the resulting gold would be mildly radioactive, requiring 14-18 years of storage before safe use
2 Likes
Definitely wouldn’t put too many eggs in that particular basket. The hype is that values will continue to climb as demand outstrips supply for industrial use. As a practical matter though, you can’t expand your data center business relying on increasingly expensive material with increasingly unreliable availability, so my money is on big tech moving to alternative materials, and away from precious metals.
2 Likes
Correct. Many are using copper and aluminum with electroplating. If silver gets too expensive, alternatives will be found.
1 Like