<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Gilberto Capelli on Medium]]></title>
        <description><![CDATA[Stories by Gilberto Capelli on Medium]]></description>
        <link>https://medium.com/@gilberto.capelli?source=rss-e00c48f18700------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*C4RV34tMZWlG2rfDgfj2Mg.png</url>
            <title>Stories by Gilberto Capelli on Medium</title>
            <link>https://medium.com/@gilberto.capelli?source=rss-e00c48f18700------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 22:10:21 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gilberto.capelli/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Copper Grades Were Modest. The Geological Clues Were Not]]></title>
            <link>https://medium.com/the-investors-handbook/the-copper-grades-were-modest-the-geological-clues-were-not-7136dfc4a128?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/7136dfc4a128</guid>
            <category><![CDATA[mining]]></category>
            <category><![CDATA[stock-market]]></category>
            <category><![CDATA[money]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[materials]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Fri, 24 Jul 2026 14:10:36 GMT</pubDate>
            <atom:updated>2026-07-24T18:04:59.908Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>Early exploration is rarely about finding the final grade in the first hole. It is about learning whether the system is alive.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6ZbxowDoJiAFtzso7ZhfQw.png" /><figcaption><a href="https://www.newsfilecorp.com/release/295394/NovaRed-Mining-Expands-Wilmac-CopperGold-Project-with-Option-of-TrojanCondor-Corridor">https://www.newsfilecorp.com/release/295394/NovaRed-Mining-Expands-Wilmac-CopperGold-Project-with-Option-of-TrojanCondor-Corridor</a></figcaption></figure><p>Imagine receiving two documents from the same drill program.<br>The first is an assay table. It shows <strong>1,084 ppm copper over 3.13 metres</strong>, <strong>381 ppm over 26.83 metres</strong> and <strong>262 ppm over 24.16 metres</strong>. In percentage terms, those results are approximately <strong>0.108%, 0.038% and 0.026% copper</strong>.<br>The second document is the geological log. It describes diorite, gabbro and pyroxenite, sericite and epidote alteration, quartz-carbonate veinlets, pyrite, traces of chalcopyrite and locally developed stockwork textures.<br>Read only the assays and the historical Trojan–Condor drilling looks modest. Read the core descriptions beside them and the program becomes more informative. The holes entered an intrusive setting that had been fractured, altered and affected by copper-bearing hydrothermal fluids.<br>That does not turn the historical results into a discovery. It shows why the grades should not be interpreted separately from the rocks that carried them.</p><h3>The assay table measures copper. The core records what happened</h3><p>The strongest re-sampled section came from approximately <strong>97.87 to 101.00 metres downhole</strong>. It returned a weighted average of <strong>1,084 ppm copper over 3.13 metres</strong> within altered diorite containing a weak quartz-carbonate-pyrite stockwork.<br>The number is easy to understand. The stockwork may be the more useful clue.<br>A stockwork is a network of intersecting veinlets formed as fluids repeatedly move through fractured rock. Historical descriptions from Trojan–Condor included stockwork zones reaching approximately <strong>0.5 metres</strong>, together with carbonate, epidote and quartz veinlets generally less than <strong>1–2 centimetres</strong> thick.<br>Those dimensions are not remarkable by themselves. What matters is the process they record. The rock did not experience one isolated fracture and one isolated pulse of fluid. It was repeatedly opened and filled.<br>Porphyry systems commonly develop through this interaction between intrusive heat, fluid pressure, fracturing and repeated mineral deposition. <a href="https://www.usgs.gov/publications/porphyry-copper-deposit-model">A stockwork texture does not prove that Trojan–Condor contains an economic porphyry deposit, but it provides a recognized geological framework for interpreting the core.</a><br><br>The two broader copper intervals add context. Results of <strong>381 ppm over 26.83 metres</strong> and <strong>262 ppm over 24.16 metres</strong> show that anomalous copper extended through tens of metres of rock rather than appearing only within one centimetre-scale vein.<br>The grades remained low. The footprint was wider.<br>For NRED, that raises a more useful exploration question than whether the old intervals were economic: what created copper anomalism across those widths, and does the system become stronger in another direction?</p><h3>The mineral sequence can become a map</h3><p>The core was not composed of one uniform intrusive rock. Historical logging identified diorite, leucocratic and melanocratic gabbro, and sections grading toward pyroxenite.<br>That variety suggests a more complicated intrusive history than a single body cooling in place.<br>Hydrothermal systems can develop through several intrusive pulses. One phase may provide heat, another may create new fractures, and later fluids may overprint rocks that were already altered. Not every phase will necessarily carry the same copper potential.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Xv3QLwtFffBSbxRh4SX3sA.png" /></figure><p>This makes the relationships between the rock types important. NRED can examine whether stronger stockworks occur within one intrusive phase or near a contact between two phases. Changes in alteration intensity can show how fluids travelled through the system, while the distribution of sulphides may help identify its internal zoning.<br>Sericite and epidote are particularly useful in that context. Epidote commonly occurs within more external propylitic alteration, while sericite may reflect a different hydrothermal environment or later overprinting. <a href="https://www.usgs.gov/publications/some-phase-relations-hydrothermally-altered-rocks-porphyry-copper-deposits">Their presence alone does not identify the centre of a mineralized system, but changes in their abundance can help establish whether a drill hole was moving through different parts of the hydrothermal footprint.</a><br><br>Pyrite was reported more widely than chalcopyrite, which appeared in traces. In some porphyry environments, pyrite-rich alteration can extend around or above stronger copper-bearing zones. That relationship cannot simply be assumed at Trojan–Condor, but it gives NRED a testable interpretation for detailed re-logging and three-dimensional modelling.<br>The practical value lies in the sequence preserved in the core: intrusive phase, alteration type, veinlet density, sulphide abundance and copper concentration.<br>A section carrying only moderate copper may still show that the hole was approaching a contact or crossing the outer part of a broader hydrothermal zone. Another interval may reveal that alteration weakened in the direction drilled, suggesting that the more prospective area lies elsewhere.<br>The assays report how much copper was present along the hole. The geological sequence can help explain why it was there and where to look next.</p><h3>The next target may not sit beside the best assay</h3><p>The obvious follow-up would be to focus on the <strong>1,084 ppm</strong> interval because it carries the highest reported copper value.<br>That may not be the best use of the historical information.<br>NRED can instead compare all three reported intervals by intrusive phase, alteration, sulphides and veinlet texture. The <strong>381 ppm over 26.83 metres</strong> result may represent a broad altered envelope surrounding narrower fluid pathways. The <strong>262 ppm over 24.16 metres</strong> interval may sit within another intrusive phase or a different part of the same hydrothermal event.<br>The highest grade, widest alteration and most favourable structural position may not occur in the same place. That is normal in an early-stage intrusive target.<br>Later soil work also reported coincident copper-gold-silver anomalies around intrusive margins, fault intersections and adjacent Nicola Group rocks. The interiors of the mapped Whipsaw Stocks were generally less anomalous, while parts of the soil pattern remained open toward the southwest.<br>That distribution gives NRED another reason to examine the edges of the intrusive bodies rather than simply drilling below their visible centres. Contacts and faults may have provided pathways for fluids, linking the surface anomalies to alteration and sulphides observed in the historical core.<br>A modern three-dimensional model can now combine those observations. Historical drill traces can be positioned relative to faults, magnetic architecture and interpreted intrusive contacts. Sections of stronger alteration can be compared with the direction of soil anomalies and deeper geophysical responses.<br><a href="https://www.geosciencebc.com/projects/2018-047/">Independent work elsewhere in British Columbia’s Quesnel terrane has shown how magnetic and three-dimensional inversion models can improve the interpretation of intrusive bodies beneath cover.</a><br><br>For NRED, the purpose is not to make the old copper grades look larger than they were. It is to extract more geological meaning from the material already collected.<br>Trojan–Condor’s historical core records several intrusive phases, stockwork and veinlet development, sericite-epidote alteration, widespread pyrite and traces of chalcopyrite. Copper anomalism extended across intervals ranging from <strong>3.13 to 26.83 metres</strong>.<br>None of those clues settles the project on its own. Together, they show that the drilling entered an active intrusive and hydrothermal environment.<br>The early assays measured the copper intercepted by four narrow paths through that environment.<br>The geology may now help NRED determine which direction those paths should have taken.<br>The copper grades were modest. The system they recorded was not geologically quiet.<br>That is why the old core may be more valuable as a map toward the next target than as a final judgment on Trojan–Condor.</p><p>📩 <strong>Join </strong><a href="https://investorshandbook-newsletter-c72f94.beehiiv.com/"><strong>Investor’s Handbook Digest</strong></a><strong> — get the best investing, markets, and wealth-building insights each week.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7136dfc4a128" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-investors-handbook/the-copper-grades-were-modest-the-geological-clues-were-not-7136dfc4a128">The Copper Grades Were Modest. The Geological Clues Were Not</a> was originally published in <a href="https://medium.com/the-investors-handbook">Investor’s Handbook</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[$1.65 Billion For Electricians — This Is Where AI Becomes Physical]]></title>
            <link>https://medium.com/@gilberto.capelli/1-65-billion-for-electricians-this-is-where-ai-becomes-physical-dd3cbe0f4785?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/dd3cbe0f4785</guid>
            <category><![CDATA[canada]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[infrastructure]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Fri, 10 Jul 2026 16:43:10 GMT</pubDate>
            <atom:updated>2026-07-10T16:43:10.867Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>MasTec is not buying “AI hype.” It is buying people, cables, substations, electrical rooms, and access inside data centers.</em></strong></p><p>AI infrastructure often appears in the news as chips, cloud, GPUs, hyperscalers, and new models. But sometimes the real signal does not come from a lab or a presentation. It comes from a deal for an electrical contractor.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/898/1*PV0x5jpgXLacCNd-n-qexQ.png" /></figure><p><a href="https://www.investors.com/news/master-stock-ai-infrastructure-play-target-hike-superior-acquisition/">MasTec is buying The Superior Group for roughly $1.65 billion: about $1.175 billion in cash and $475 million in MasTec stock, plus a possible earnout based on performance over the next 36 months. IBD writes that after the deal, Cantor Fitzgerald raised its price target on MTZ to $581 from $545, while MasTec is increasingly being read as an AI infrastructure play.</a></p><p><a href="https://www.businesswire.com/news/home/20260707386169/en/MasTec-to-Acquire-The-Superior-Group-Enhancing-MasTecs-Infrastructure-Capabilities-Across-Data-Center-and-Mission-Critical-End-Markets">What interests me here is not the target hike by itself. It is what MasTec is actually buying. Superior is a major U.S. electrical contractor with roots going back to 1925, around 3,000 employees, and a focus on critical infrastructure.</a> The company works across data centers, healthcare, entertainment, and industrial projects, offering design, preconstruction, construction, engineering, integrated systems, prefabrication, modular manufacturing, maintenance, repair, and retrofit services.</p><p>That does not look much like a normal “AI stock story.” It is a story about who knows how to physically bring electricity where it is needed. Before the deal, MasTec was already strong in outside-the-fence infrastructure: generation, transmission, substations, communications, and civil work. Superior adds the inside-the-fence piece: electrical systems, integrated building systems, commissioning support, and long-term facility services.</p><p>A data center is not just a box with racks. It is an electrical project, a thermal project, and a construction project sitting in one place. There is a transformer yard, switchgear, cable trays, backup power, UPS systems, cooling pumps, controls, fire systems, fiber, and people who have to make sure none of it fails at the worst possible moment.</p><p>And here you can see a bottleneck people talk about less. If AI demand grows faster than skilled electrical crews, the shortage is not only in chips. It is also in the hands that can build and maintain power. Money can be poured in quickly. An experienced electrician or project crew for a mission-critical site cannot be printed that fast.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/890/1*Y5RIpRODjbsDGF_c4aXzwg.png" /></figure><p>Superior’s numbers explain why MasTec is paying this much. The company expects Superior to generate roughly $1.6–$1.7 billion in revenue and $225–$250 million in adjusted EBITDA for full-year 2026. For 2027, expectations are even higher: $2.2–$2.5 billion in revenue and $250–$275 million in adjusted EBITDA. This is not a small purchase. It is a bet on a long, expensive, and very electric AI buildout.</p><p>The deal also quietly brings us back to copper. An electrical contractor does not work with abstract “capacity.” It works with cables, busbars, transformers, panels, switchgear, grounding, and controls. More data centers mean more electrical infrastructure. More electrical infrastructure means more metal that has to be found, mined, processed, and delivered.</p><p>NovaRed Mining can be looked at here simply as one early example of that long chain, not as a data center contractor and not as an AI stock. If MasTec is buying the ability to build electrical rooms inside data centers, then earlier in the chain someone has to search for the copper that may later become a cable tray, transformer winding, or busbar. <a href="https://novaredmining.com/">One fresh positive angle around NovaRed that fits here is the 2026 field program: expanded soil sampling and four IP/AMT surveys across North Lamont, West Lamont, Wilmac, and Plume, with an initial drill program targeted for fall 2026, subject to an approved drill permit.</a></p><p>This is not finished copper for a data center. It is the first rough link: ground, old reports, geophysics, soil data, target, permit, drill. North Lamont has previously shown copper-in-soil up to 1,125 ppm. Those details seem far away from AI until you look at AI as a network of buildings, substations, cables, and people in hard hats.</p><p>MasTec is buying Superior not because “AI” is a nice word. It is buying access to a bottleneck: electrical expertise, field crews, prefabrication, maintenance, and work inside facilities where mistakes are expensive.</p><p>This is how AI becomes physical. Not in a model lab, but in an electrical room. Not only in a chip rack, but in a cable route, a substation, a transformer yard, and in metal that has to travel the full path from the ground to the current.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dd3cbe0f4785" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Quand l’AI en radiology ne ressemble plus à une expérience]]></title>
            <link>https://medium.com/@gilberto.capelli/quand-lai-en-radiology-ne-ressemble-plus-%C3%A0-une-exp%C3%A9rience-744bcadc0bf2?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/744bcadc0bf2</guid>
            <category><![CDATA[robotics]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[macro-technology]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Tue, 30 Jun 2026 18:14:17 GMT</pubDate>
            <atom:updated>2026-06-30T18:14:17.744Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>La FDA a ajouté 68 nouveaux cleared algorithms pour la radiology au cours des 3 premiers mois de 2026 seulement. Cela ne ressemble plus à du hype. Cela ressemble plutôt à la manière dont l’AI devient discrètement une partie de l’infrastructure hospitalière ordinaire.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/756/1*VguzQfk2JKRZ3yEOsX4sJQ.png" /></figure><p>68 nouveaux FDA-cleared algorithms pour la radiology en un trimestre, c’est beaucoup. Pas parce que chacun d’eux va transformer la médecine à lui seul. Mais parce qu’ensemble, ils montrent le rythme.</p><p><a href="https://radiologybusiness.com/topics/artificial-intelligence/radiology-gets-68-new-fda-cleared-algorithms">Radiology Business écrit qu’après la mise à jour de la FDA du 16 juin 2026, le nombre total d’AI-enabled medical devices a atteint 1,524 au 30 mars 2026. Parmi eux, 1,163 concernent précisément la radiology. Autrement dit, la radiology représente aujourd’hui plus de 76% de tous les cleared AI medical devices.</a></p><p>Ce n’est pas un hasard. La radiology convient très bien à l’AI. Il y existe déjà depuis longtemps des images numériques, CT, MRI, X-ray, ultrasound, PACS, DICOM, reporting systems et des tâches répétitives. L’algorithme n’a pas besoin d’attendre que la médecine devienne numérique. En radiology, elle l’est déjà depuis longtemps.</p><p>Mais il ne faut pas exagérer. L’AI n’arrive pas ici comme un grand robot qui va remplacer le radiologist. Elle arrive plutôt sous forme de nombreux petits outils intégrés dans la journée de travail : reconstruction, triage, segmentation, detection, quantification, image enhancement, reporting support.</p><p><a href="https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-enabled-medical-devices">C’est pour cela que la liste des nouveaux cleared products paraît si variée. On y trouve Philips Spectral CT Verida Family, les Canon MRI systems avec AiCE Reconstruction Processing Unit, Lunit Insight DBT pour la breast tomosynthesis, SwiftMR, Aidoc BriefCase-Triage, Siemens AI-Rad Companion Brain MR, qXR-Detect, Annalise Enterprise, RevealAI-Lung, Monaco RTP System, AutoContour et d’autres.</a></p><p>Ce n’est pas une seule histoire d’AI. Ce sont des dizaines de petites entrées dans différentes parties de l’imaging workflow.</p><p>Un système aide à reconstruire l’image plus rapidement. Un autre cherche un cas suspect pour le triage. Un troisième automatise le contouring. Un quatrième aide avec les lung findings. Un cinquième travaille avec le breast imaging. Pour le patient, cela peut rester presque invisible. Pour l’hôpital, c’est un changement dans la manière dont la journée de travail avance.</p><p>Le rythme est encore plus intéressant. Selon Radiology Business, la FDA cleared actuellement environ 30 AI devices par mois. En 2024, le rythme moyen était d’environ 21 par mois. Donc la base n’est pas seulement grande. Elle accélère.</p><p>Et c’est là que commence la partie moins jolie. Acheter ou recevoir un cleared algorithm n’est pas la même chose que changer réellement le travail d’un department. Un hôpital a besoin de scanners, d’une PACS integration, de cybersecurity, de storage, de cloud ou local compute, de monitors, de reporting tools, de QA, de training et d’une vraie intégration dans le workflow. Si l’AI ajoute encore un écran, encore un clic et encore une responsabilité floue, elle devient vite du bruit plutôt qu’une aide.</p><p>Le vrai test pour la medical AI n’est donc pas seulement la FDA clearance. Le vrai test, c’est de savoir si quelqu’un l’utilisera à la 10e heure d’un shift, quand la file d’images ne se termine pas.</p><p>C’est pourquoi la radiology est intéressante. Elle n’accumule pas seulement beaucoup de cleared algorithms en premier. Elle montre comment l’AI en médecine cesse d’être un sujet séparé et commence à entrer dans les équipements, les workstations, le reporting, l’enterprise imaging et le workflow quotidien.</p><p>Mais sous tout cela, il y a une couche physique facile à oublier. La medical imaging AI sonne comme code, models et datasets. En réalité, elle repose sur des scanners, X-ray tubes, detectors, MRI coils, ultrasound probes, servers, cables, cooling, power electronics, monitors et hospital electrical infrastructure. Plus l’imaging devient AI-enabled, plus l’hôpital dépend non seulement du software, mais aussi d’une électronique stable.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/990/1*BrXEt5nxzmM49JO5CPrYPQ.png" /></figure><p>C’est pourquoi le cuivre ne paraît pas éloigné dans ces histoires de medtech. Il se trouve dans le wiring, les motors, les imaging systems, les coils, la power electronics, les sensors, les cooling systems et l’hospital infrastructure. Le nouveau cuivre n’apparaît pas au moment où la FDA ajoute un nouvel AI algorithm à sa liste. Il commence plus tôt — avec la géologie, les permis, les travaux de terrain et les early copper projects, comme NovaRed Mining, qui travaille sur le Wilmac Copper-Gold Project en British Columbia, environ 16,078 hectares dans la Quesnel porphyry belt.</p><p><a href="https://novaredmining.com/">https://novaredmining.com</a></p><p>68 nouveaux algorithms en un trimestre, ce n’est pas seulement un chiffre AI de plus. C’est un signe que la radiology est déjà devenue le principal terrain d’essai pour la medical AI.</p><p>Pas parce que les médecins vont disparaître demain.</p><p>Mais parce qu’à côté du scanner, du monitor, du cable et du PACS, il y aura de plus en plus souvent une couche supplémentaire — un algorithme censé aider à comprendre plus vite la prochaine image.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=744bcadc0bf2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Is No Longer Building a Cloud. It Is Building a Power Plant]]></title>
            <link>https://medium.com/@gilberto.capelli/ai-is-no-longer-building-a-cloud-it-is-building-a-power-plant-2885a63a5e8b?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/2885a63a5e8b</guid>
            <category><![CDATA[mining]]></category>
            <category><![CDATA[stock-market]]></category>
            <category><![CDATA[infrastructure]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[investing]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Mon, 29 Jun 2026 18:29:55 GMT</pubDate>
            <atom:updated>2026-06-29T18:29:55.427Z</atom:updated>
            <content:encoded><![CDATA[<p>When hyperscaler capex moves into the hundreds of billions, this stops being only a GPU story. Behind AI come land, substations, cables, cooling, water, transformers and metal.</p><p>AI sells itself very well as something light. You open a browser, type a prompt, get an answer. No smoke, no pipes, no factory outside the window. Just “cloud.”</p><p>But that cloud no longer looks light in the budgets.</p><p>Intellectia writes that tech giants may invest more than $1.5 trillion in AI infrastructure over the next several years. <a href="https://intellectia.ai/blog/ai-infrastructure-investment-june-2026">The article also notes that combined annual capex from Amazon, Microsoft, Alphabet and Meta rose from about $100 billion in 2023 to more than $300 billion in 2025, and could exceed $500 billion in the coming years</a>. Nvidia sits at the center of this picture, with a projected $1 trillion in confirmed AI chip demand through 2027.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/983/1*U_SQgBEMgIF2Jcj9u9RP2g.png" /></figure><p>These numbers are interesting not only as another tech bull case. They show that AI has started pulling money out of software and into physical infrastructure. $1.5 trillion is no longer just about chips. It is data centers, power infrastructure, networking equipment, cooling systems, land, substations, transmission, backup generation, specialized construction materials and permitting.</p><p>In other words, everything that cannot be scaled with a code update.</p><p><a href="https://www.reuters.com/business/big-tech-invest-about-650-billion-ai-2026-bridgewater-says-2026-02-23/">Reuters previously reported, citing Bridgewater, that Alphabet, Amazon, Meta and Microsoft could invest around $650 billion in AI-related infrastructure in 2026, compared with $410 billion in 2025.</a> Bridgewater also warned that the boom is entering a “more dangerous phase”: compute demand is far above supply, while physical infrastructure spending is starting to create risks for capital, electricity prices and the broader market.</p><p>And this is where the story becomes less comfortable for the market. If AI needs more compute, it automatically needs more electricity. And electricity does not live in a presentation. It lives in transformers, switchgear, copper cables, busbars, substations, grid interconnections, cooling loops, power modules, land rights and years of approvals.</p><p>Old software could grow almost without friction. The new AI stack does not work that way. You can buy GPUs, but then you need somewhere to put the rack. You need to bring in power. You need to cool the equipment. You need to find grid capacity. You need permits. You need to wait for transformers, breakers, cable, high-voltage equipment and the people who can put all of it together.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/642/1*CK_orj0btYP2mIRj6rWwOQ.png" /></figure><p>That is why the AI infrastructure boom looks less and less like a classic tech cycle. It looks more like an industrial wave, where next to Nvidia we suddenly see utilities, electrical equipment, gas turbines, construction, cooling, grid upgrades, semiconductors, copper, steel and critical minerals.</p><p>The market calls it AI.</p><p>Physically, it often looks like a very old economy: concrete, land, metal and electricity.</p><p>Copper is not a decorative detail on the side of this story. It sits inside the electrical logic itself: cables, transformer windings, grounding, motors, pumps, busbars, switchgear, inverters, power distribution, cooling systems and electrical hardware. The larger the data center, the more points there are where electricity has to move without overheating, losses or weak spots.</p><p>And this is not just about one campus. If the AI buildout is really moving at the scale of hundreds of billions of dollars per year, copper demand starts appearing across the whole chain — from grid connections to backup systems, from cooling pumps to substations, from transmission upgrades to internal power rooms.</p><p>If AI capex is moving into land, substations, cables and transformers, then copper supply has to be found not only in ready-made miners. It begins much earlier — with exploration, soil anomalies, geophysics and the first drill programs. <a href="https://novaredmining.com/">One such early example is NovaRed Mining with Wilmac in British Columbia: historical AMT/3DIP data, porphyry-style targets and North Lamont copper-in-soil anomalies of up to 1,125 ppm</a>. It is still early stage, but an AI power room also begins not with a server rack, but with metal in the ground.</p><p>The biggest mistake right now is looking at AI infrastructure only through Nvidia shares or hyperscaler capex. That is the top layer. Important, expensive, visible, but not the only one.</p><p>Under it is electricity. Under electricity is grid equipment. Under equipment are materials. And under materials are years of supply chains, permits, drilling, construction and production.</p><p>AI may live in the cloud.</p><p>But that cloud increasingly needs a very real socket.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2885a63a5e8b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Je ne vois pas ici une simple formalité boursière.]]></title>
            <link>https://medium.com/@gilberto.capelli/je-ne-vois-pas-ici-une-simple-formalit%C3%A9-boursi%C3%A8re-1d065a46d289?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/1d065a46d289</guid>
            <category><![CDATA[stock-market]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[money]]></category>
            <category><![CDATA[stocks]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Mon, 22 Jun 2026 16:01:15 GMT</pubDate>
            <atom:updated>2026-06-22T16:01:15.200Z</atom:updated>
            <content:encoded><![CDATA[<h3>Je ne vois pas ici une simple formalité boursière. Je vois comment le cuivre devient un document géopolitique</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/413/1*1pJ1hV1ym6evetcNUfdVPA.png" /></figure><p>Quand je lis des nouvelles sur les LME warranting rules, je ne vois pas seulement un petit détail juridique pour les traders. En surface, cela paraît effectivement étroit : Russian-origin copper, cobalt, EU warehouses, attestations, sanctions compliance, warranting process. Mais pour moi, derrière ces mots, il y a une histoire beaucoup plus large — le marché des métaux devient de plus en plus documenté, politique et contrôlé.</p><p><a href="https://www.reuters.com/business/lme-issues-notice-warranting-russian-origin-copper-cobalt-eu-2026-06-17/">Reuters écrit que London Metal Exchange a issued a notice concernant le warranting de Russian-origin copper and cobalt dans les EU LME-listed warehouses. </a>La condition principale est simple : à partir du 25 July 2026, ce métal pourra être registered on warrant dans les EU warehouses uniquement s’il existe une evidence qu’il a été imported into the EU before 25 July 2026 et qu’il respecte les amended EU sanctions.</p><p>Pour quelqu’un qui n’est pas dans le marché, le mot “warrant” peut sembler abstrait. Mais sur le LME, un warrant n’est pas simplement un papier. C’est un document qui confirme ownership of metal dans un registered warehouse et rend ce métal deliverable against exchange contracts. Si un métal ne peut pas être warranted, il perd une partie de sa liquidité financière dans le LME system.</p><p>La question n’est donc pas seulement de savoir où se trouve physiquement le copper. La question est de savoir si ce copper peut entrer dans le système mondial de circulation boursière.</p><p><a href="https://www.lme.com/api/sitecore/MemberNoticesSearchApi/Download?id=af28365d-6146-4054-a2c2-b75fdc16c16a">LME explique que le EU Council a adopted Regulation 2026/506 le 23 April 2026. Ce regulation amended existing sanctions et interdit le purchase, import or transfer de copper HS 7403 et cobalt HS 8105 into the EU, s’ils originate in ou are exported from Russia</a>. Between 24 April and 24 July 2026, des exceptions existent pour les contracts concluded before 24 April 2026. Mais à partir du 25 July 2026, l’importation de Russian Copper and Cobalt into the EU is prohibited.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/583/1*sls-0dME9QO5J-7GM09H9A.png" /></figure><p>Ce qui m’intéresse ici, c’est la mécanique elle-même. Jusqu’au 25 July, le métal peut encore passer s’il est lié à d’anciens contrats et correctement documenté. Après cette date, la logique devient case-by-case pour le métal qui avait déjà été imported earlier but not placed on warrant. Cela signifie que le métal physique vit de plus en plus non seulement selon les lois de supply and demand, mais aussi selon les lois de origin, timing et paperwork.</p><p>Le cuivre devient non seulement une commodity. Il devient un document avec une géographie.</p><p>LME indique aussi clairement que no cobalt or copper of Russian origin has been warranted at an LME-listed warehouse in the EU for over a year. C’est pour cela que la bourse ne s’attend pas à un significant market impact de ce nouveau process. À première vue, cette phrase rassure. Si ce métal n’était déjà plus présent dans les EU warehouses depuis plus d’un an, le changement peut sembler presque symbolique.</p><p>Mais je ne sous-estimerais pas les choses symboliques dans les commodity markets.</p><p>Car parfois, ce qui compte n’est pas le nombre de tonnes qui bougent immédiatement. Ce qui compte, c’est la façon dont le framework change. Si metal origin devient de plus en plus important, si les sanctions regimes entrent dans les warehouse rules, si les attestations deviennent une partie du warranting process, alors le marché reçoit une nouvelle réalité : le métal doit être non seulement de bonne qualité, mais aussi “admissible”.</p><p>C’est un changement très important pour le copper.</p><p>Avant, le copper market semblait souvent presque universel. Il y a des cathodes, des brands, des warehouses, un price, une delivery. Mais maintenant, une autre couche s’ajoute : cette jurisdiction accepte-t-elle ce métal ? Existe-t-il un EU nexus ? Y a-t-il une preuve d’importation avant une date précise ? Le warranting crée-t-il un sanctions risk pour le member, la warehouse company ou LME Clear ?</p><p>Autrement dit, le copper devient non seulement un industrial material, mais aussi un compliance material.</p><p>Cela se voit bien dans le LME notice lui-même. Le metal owner et le Member doivent fournir des attestation letters. La warehouse company ne peut pas accept onto warrant Russian Copper and Cobalt dans un EU warehouse sans LME confirmation. Si la warehouse company a des concerns concernant la documentation, des discrepancies ou des deficiencies, le metal should not be warranted. Ce n’est déjà plus seulement de la logistics. C’est de la due diligence infrastructure.</p><p>Et c’est ici que je vois l’histoire de marché plus large. Dans un monde où le copper est nécessaire aux grids, EVs, AI data centers, renewable power, industrial automation, metro systems, water infrastructure et defense supply chains, l’origine du métal devient plus importante. Tout copper n’est pas identique pour tous les acheteurs. Tout cobalt n’a pas la même liquidité. Chaque tonne ne peut pas circuler facilement dans le système boursier.</p><p>Physiquement, le métal peut exister. Mais financièrement et juridiquement, il peut être moins pratique.</p><p>Pour moi, c’est un rappel que la copper supply chain devient plus complexe non seulement à cause de la geology ou des mine delays. Elle devient plus complexe à cause des sanctions, trade flows, warehouse rules, documentation, brand eligibility, financing restrictions et jurisdictional risk. Le supply risk ne signifie donc plus seulement “y a-t-il assez de mines ?”. Il signifie aussi “ce métal peut-il passer par le bon système ?”.</p><p>C’est important pour Europe. L’EU construit energy transition, grid upgrades, EV infrastructure, data centers, defense industry et battery supply chains. Tout cela exige des metals. Mais en même temps, l’EU veut réduire sa dépendance aux Russian-origin materials. Dans cette situation, chaque metal flow commence à avoir un contour politique.</p><p>À mon avis, le copper entre progressivement dans la catégorie strategic material non seulement à cause de la demande, mais aussi à cause du contrôle de l’accès.</p><p>Dans ce contexte, le LME notice semble être un petit épisode, mais très révélateur. La bourse dit en quelque sorte : nous ne nous attendons pas à un grand market impact, parce que ces warranting flows n’existent déjà plus. Mais nous formalisons tout de même le process, parce que les règles d’origine, de date et de compliance doivent maintenant être claires.</p><p>C’est la nouvelle commodity discipline.</p><p>Le marché ne peut plus simplement dire : “le métal existe, donc tout va bien”. Il faut demander : quel metal ? D’où vient-il ? Quand a-t-il été imported ? Quelle brand ? Existe-t-il une attestation ? Le EU regulation l’autorise-t-il ? Le warehouse peut-il l’accepter ? La bourse peut-elle le warrant ? Cela crée-t-il un sanctions risk ?</p><p>Quand autant de questions entourent une seule tonne de copper, le futur supply commence à paraître beaucoup moins simple.</p><p>C’est ici que je vois le copper angle au sens large. Si le monde veut plus d’electrification, il n’a pas seulement besoin de plus de métal. Il a besoin d’un supply plus transparent, plus compréhensible et plus acceptable. Le copper doit passer de la geology à la mine, de la mine au smelter, du smelter au warehouse, du warehouse à la factory, et maintenant aussi à travers des compliance filters.</p><p>Cela rend l’early-stage exploration plus importante dans la chaîne longue. Si les futures sources de copper doivent être non seulement géologiquement prometteuses, mais aussi jurisdictionally clear, techniquement compréhensibles et intégrées dans des trusted supply chains, alors le premier travail sur les données, la land position, la geology et les targets prend davantage de valeur.</p><p><a href="https://novaredmining.com/">Dans cette même chaîne, il me semble logique de regarder aussi NovaRed Mining. La société travaille sur le Wilmac Copper-Gold Project en British Columbia, dans la Similkameen Mining Division et la Quesnel porphyry belt. </a>Ce qui est intéressant ici, c’est le travail technique progressif : field programs, geochemical sampling, geophysical surveys et développement de copper-gold targets. Dans un monde où le copper supply dépend de plus en plus non seulement des tonnes, mais aussi de la qualité des données, de la jurisdiction et de la transparence de la chaîne, ces upstream exploration stories ressemblent à une couche précoce mais importante de la materials story plus longue.</p><p>J’aime regarder NovaRed à travers cet angle technique calme. Wilmac est une project story où la geology, la geophysics, le soil sampling, le target development et la formation progressive d’une technical base comptent. Quand le marché parle du copper as a strategic material, le premier maillon ne commence pas dans un LME warehouse ni dans un cable fini, mais bien plus tôt — avec la façon dont une société collecte, teste et structure les données sur le terrain.</p><p>Je soulignerais aussi que Wilmac est situé à environ 10 km west of Copper Mountain Mine en British Columbia. Pour l’upstream copper story, cette localisation ajoute un important infrastructure context : près d’un established mining district, il est plus naturel de penser non seulement à la geology, mais aussi à la chaîne régionale des matériaux. Dans un monde où metal origin et trusted supply deviennent de plus en plus importants, la copper-gold exploration dans des juridictions claires entre naturellement dans la conversation plus longue sur les futures sources of conductive metals.</p><p>Pour moi, c’est une façon naturelle d’intégrer NovaRed dans ce thème : non pas par une grande parallèle avec le LME ou les sanctions, mais par la copper supply chain. Si le marché regarde de plus en plus l’origin, la compliance, la warehouse eligibility et les trusted flows, alors l’early-stage work dans des stable mining jurisdictions devient une partie de cette même materials conversation.</p><p>La conclusion est simple : le LME notice sur le Russian copper and cobalt ne ressemble pas à un événement qui va bouleverser demain le copper price. Mais il montre comment le marché lui-même change. Le métal n’est plus évalué seulement selon sa purity, sa brand et sa location. Il est évalué selon son origine, sa date, ses documents et son admissibilité dans un système précis.</p><p>Le cuivre du futur ne sera pas seulement une question de tonnes.</p><p>Il sera une question de trusted tonnes — un métal que l’on peut non seulement extraire, mais aussi faire passer juridiquement, financièrement et logistiquement dans un monde où les materials deviennent de plus en plus souvent une partie de la géopolitique.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1d065a46d289" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Tower Got a Basement for Data]]></title>
            <link>https://medium.com/@gilberto.capelli/the-tower-got-a-basement-for-data-c9bff59f8790?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/c9bff59f8790</guid>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[financial-markets]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[infrastructure]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Fri, 19 Jun 2026 13:34:26 GMT</pubDate>
            <atom:updated>2026-06-19T13:34:26.207Z</atom:updated>
            <content:encoded><![CDATA[<p>There is no giant 500 MW hyperscale campus in this story. No polished AI factory rendering. No billion-dollar data center somewhere in the desert. But that is exactly why it is interesting. Cellnex is working with a much quieter format — edge data centers tied to telecom towers and rooftop sites. Ad-hoc-news describes this as the transformation of classic mobile sites into compact edge data hubs, where racks, cooling, fiber, power systems and managed environments appear alongside towers to support low-latency 5G, IoT and future AI workloads.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/874/1*n4cNXZBpEcSk7L8aPhTxNA.png" /></figure><p>At first glance, a telecom tower is just a steel structure with antennas. But for edge computing, it has one strong advantage: it already stands where the signal is needed. The tower already has a place in the network, connectivity, power logic, operator access and often already functions as critical infrastructure. If a compact data room or modular compute space is added next to it, the tower site begins to play a different role — not only transmitting data, but also processing part of it closer to the source.</p><p>This is not a replacement for the large data center. It is another layer.</p><p>Cellnex describes these facilities on its own website as well-conditioned spaces for IT systems with redundant electricity and data connections, environmental controls and security systems. In the Netherlands, the company lists 24 telecom &amp; data towers, as well as the Media Gateway in Hilversum. For France, the page lists more than 100 locations, and for Spain, 4 locations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*eNzT3kEjndOIzDODqhhY-w.png" /></figure><p>These numbers matter because edge infrastructure does not work like one large campus. Its logic is distribution. One site may be small, but a network of such sites changes the latency map. If a user, camera, industrial sensor, vehicle, drone or private 5G network is far from the core cloud, every extra mile can mean latency, backhaul load or a weaker system response.</p><p>The cloud becomes faster when part of the cloud moves closer to the ground.</p><p>That is why a Cellnex edge data center should not be seen simply as “a small data center near a tower.” It is better understood as an intermediate node between the radio access network, fiber, enterprise systems and the central cloud. Ad-hoc-news writes that such edge sites can support content caching, local video analytics or initial AI inference, where data is processed closer to where it appears, and the core cloud receives not the full raw stream, but a selected result.</p><p>This is especially important for 5G. Many 5G use cases have been sold for years as the future: connected vehicles, smart-city sensors, industrial automation, autonomous drones, public safety and private networks. But all of these things require more than a radio signal. They need compute nearby. If a video analytics camera or an industrial system has to wait for data to travel to a remote cloud and return, part of the 5G promise disappears.</p><p>Cellnex itself talks about next-generation connectivity as the backbone of essential services, public infrastructure and Europe’s digital transformation. At MWC Barcelona 2026, the company emphasized real European use cases — transport systems, emergency and public safety communications, energy-as-a-service, major venues, corporate facilities, broadcasting and everyday services on which millions of people depend.</p><p>In this context, the tower becomes interesting not as a tower, but as real estate with a function. It is already integrated into the telecom network. It already has a role in coverage. It often stands closer to local demand than a hyperscale campus. And if a modular edge room appears at the base of the tower, it begins to absorb part of the digital work that previously would have been automatically sent farther away.</p><p>But there are limits. An edge data center near a telecom tower is not designed for large GPU clusters or massive AI training. Space, power and the thermal envelope are much tighter there than in a full colocation or hyperscale facility. Ad-hoc-news directly notes that high-density GPU clusters for large-scale AI training do not comfortably fit into every site physically or thermally.</p><p>That is why this topic is not about “all AI data centers moving to towers.” No. Large training workloads will remain where there is massive power, cooling and campus-scale infrastructure. But inference, caching, local analytics, private networks, IoT processing, smart-city systems and industrial control follow a very different logic. There, proximity matters more than size.</p><p>The most interesting part is that Cellnex is not just a tower landlord in the old sense. Reuters wrote in 2024 that Cellnex is Europe’s largest mobile phone tower company, and that towers accounted for more than 80% of its revenue at the time. The company also focused on Towers, Fiber, Connectivity &amp; Housing services, Broadcast, and DAS &amp; Small Cells. So edge data centers fit not as a random side product, but as an attempt to extract more functions from an existing infrastructure base.</p><p>This is where the familiar physical layer appears. Edge computing does not need only software. Even a small modular edge room needs power distribution, UPS, electrical switchboards, wiring, cooling, security, fire protection, racks, fiber, monitoring and remote operations. Cellnex directly describes electrical infrastructure, backup systems, connectivity infrastructure and modular pre-wired spaces as part of its data center offering.</p><p>And this is where the copper bridge naturally appears. Telecom towers, edge data rooms, fiber nodes, power cabinets, cooling systems, UPS, switchgear, sensors and control systems are all physical infrastructure. They require wiring, connectors, busbars, motors, boards and electrical equipment. The more compute moves closer to tower sites, the more the digital network becomes material.</p><p>At the early exploration level, this logic also touches NovaRed Mining . The Wilmac Copper-Gold Project in British Columbia is not an edge data center or a telecom tower. It is a distant beginning of the material chain: soil sampling, IP/AMT geophysics, a planned 2026 field program and early copper-gold targets. But before copper becomes part of a tower power cabinet, cooling control, fiber site or edge infrastructure, it passes through exactly these quiet stages of testing in the ground.</p><p>Cellnex shows one very simple thing: the future network will not only be faster. It will be more distributed. Compute will not live only in large data centers. Part of it will move down into small technical rooms, closer to antennas, roads, factories, cameras and sensors.</p><p>Before, the tower caught the signal.</p><p>Now it is starting to hold part of the cloud.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c9bff59f8790" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[First came not code, but 400kV]]></title>
            <link>https://medium.com/@gilberto.capelli/first-came-not-code-but-400kv-dc81c24e2343?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/dc81c24e2343</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[stocks]]></category>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[investing]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Mon, 15 Jun 2026 19:03:12 GMT</pubDate>
            <atom:updated>2026-06-15T19:03:12.781Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Konza is often called a smart city. But before apps, data centers and offices, something much simpler had to appear first — a powerful electricity network.</em></p><p><a href="https://techtrendske.co.ke/2026/06/15/ketraco-isinya-konza-line-konza-technopolis/">KETRACO has commissioned the Isinya-Konza line and the Konza substation</a>. At first glance, this looks like a normal energy-sector update. But the numbers say more than the headline: a 98-kilometer double-circuit 400kV transmission line, a 400/132/66kV substation, two 400/132kV transformers rated at 350MVA each, and two more 132/66kV transformers rated at 100MVA each. This is not a small grid upgrade. For a technopolis, this kind of line is almost a permission slip for growth.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/946/1*sVZ05HiVNONoPOETmiRDOA.png" /></figure><p>What caught my attention in this story was not just the word “Konza”. And not even “smart city”. The more interesting part is that before all the polished language about innovation, there is a very physical question: can this place actually receive enough electricity? Because a technology hub can look ready in a presentation long before the substation is ready in real life.</p><p>Konza Technopolis has long been presented as one of Kenya’s important technology projects. But a city does not become smart just because it is called that. If data centers, campuses, research labs, offices, industrial zones or manufacturing sites are supposed to operate there, someone has to solve much less fashionable problems in advance: where the connection will come from, how much backup capacity exists, whether the grid can handle peaks, and what happens during maintenance or a failure.</p><p>That is why the Isinya-Konza line matters beyond the energy sector. It does not create demand by itself, but it gives demand room to grow. 400kV is not a number for a glossy press release. It is the kind of infrastructure level that shows the project is not being prepared for a small load. When an investor looks at a city of the future, the question is not only “is there land?”. The question is: can people work there without constantly worrying that electricity will become the bottleneck?</p><p>In that sense, transmission is often underestimated. Everyone likes talking about solar plants, geothermal power, data centers, AI and startups. But what happens if a country has electricity, yet the capacity cannot properly reach the place where new demand is appearing? Transmission lines do not look as fashionable as a technology campus, but they often decide where business actually gets built and where it simply waits.</p><p>TechTrendsKE links this project to the lower eastern region, industrial development and Vision 2030. KETRACO talks about transfer capacity, system stability and operational flexibility. That sounds dry, but the meaning is simple: the grid should not just exist. It should be able to move power, carry load and give operators room to maneuver. For a city that wants to grow, this is not a detail. It is the foundation.</p><p>And this is where copper is easy to forget. It is not at the center of this news, but it is inside the body of the whole story: transformers, substations, cables, distribution systems, control equipment and future connections. When a country builds a technology corridor, it is also building demand for materials. Not only concrete and steel, but conductivity.</p><p>A cable does not begin at the substation. Somewhere earlier, there is metal exploration, land rights, financing and years of early-stage work. NRED can be mentioned here as one junior example in the broader copper story. At Wilmac, the company has an option for a 70% interest, while exploration commitments could reach CAD $8.5M, including CAD $1.5M staged in 2026. This is not production and not a finished cable. But these numbers show how long the road really is from geology to real infrastructure.</p><p>That is why the Isinya-Konza news reads as more than just the launch of a power line. It shows the moment when a city’s future stops being a picture and starts demanding heavy things: kilometers, transformers, metal, permits, installation and time.</p><p>A city can call itself smart as much as it wants. But if it cannot receive power, it is still just a plan on a map.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dc81c24e2343" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Future Stopped Before the First Kilometer]]></title>
            <link>https://medium.com/@gilberto.capelli/the-future-stopped-before-the-first-kilometer-db1f0628dc47?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/db1f0628dc47</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[money]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Tue, 09 Jun 2026 17:19:42 GMT</pubDate>
            <atom:updated>2026-06-09T17:19:42.475Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4-_HDTTLBszhcf_bdCj1cg.png" /></figure><p>BMW is preparing Neue Klasse as an electric leap. But the Northvolt story showed where the car of the future can break before it ever reaches the road.</p><p>It is easy to imagine an electric car already finished. A clean silhouette, a quiet road, a large screen, fast acceleration, a battery under the floor. In the advertising image, everything looks as if the future simply rolls out of the factory.</p><p>But sometimes the most important part of an electric car does not move anywhere. It waits in the production queue.</p><p><a href="https://www.press.bmwgroup.com/global/article/detail/T0311274EN/bmw-group-continues-to-drive-electromobility%3A-long-term-supply-contract-with-northvolt-for-battery-cells-from-europe-concluded?language=en">Back in 2020, BMW signed a long-term €2 billion contract with Northvolt for battery cells from Europe.</a> The idea sounded almost perfect: a European automaker, a European battery supplier, production in Sweden, 100% energy from wind and hydropower, and a lower carbon footprint in the heaviest part of the electric vehicle.</p><p>Then reality turned harder than the presentation. In June 2024, Reuters reported that BMW had cancelled the €2 billion contract, while German media pointed to Northvolt’s problems meeting delivery timelines. In March 2025, Northvolt filed for bankruptcy in Sweden. For Europe, this was not just the failure of one company. It was a cold lesson: battery sovereignty is not built with a slogan.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/1*M_5r79YjtX32J7xQmJc9xA.png" /></figure><p>This is where BMW’s story becomes much bigger than one agreement. <a href="https://www.press.bmwgroup.com/global/article/detail/T0457069EN/successful-market-launch-for-neue-klasse%3A-bmw-group-posts-significant-first-quarter-growth-in-european-new-orders?language=en">Neue Klasse is meant to be BMW’s new electric architecture — with a different design, software layer, battery logic and longer range. BMW has already shown the iX3 50 xDrive with a stated WLTP range of 678–805 km</a>. On paper, that sounds like a confident step into the future. But that step rests on a very physical question: can the industrial chain deliver the right cells, at the right quality, at the right time?</p><p>What caught me here is not the Northvolt drama by itself. It is how quickly the electric car becomes physical again. We talk about software, platforms, range, design and charging. Beneath all of that sit factories, drying lines, chemistry, cathodes, anodes, copper foil, electrodes, quality control, energy and time.</p><p>In the battery chain, copper is not always at the center of the conversation, but it is always nearby. It is needed in the vehicle’s electrical systems, charging infrastructure, power electronics, cables, transformers and the grid that has to feed the new fleet. An electric car may look light and digital. Its industrial shadow is heavy.</p><p>While automakers count platforms and suppliers, the materials chain moves on a slower calendar. At CSE: NRED, that early layer appears through the commitments around Wilmac and Trojan-Condor: up to $8.5 million in exploration expenditures, including $1.5 million staged for 2026. This is not a story about one specific BMW battery. It is a reminder that electrification begins long before the car interior — with maps, capital, fieldwork and patience.</p><p>BMW can build a beautiful Neue Klasse. Europe can announce battery independence. But between ambition and the road stands a factory that has to keep up.</p><p>The future does not always break on the highway. Sometimes it simply does not get its battery on time.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=db1f0628dc47" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Europe Is Protecting Its Factories. But What If the Real Battle Starts Before the Factory?]]></title>
            <link>https://medium.com/@gilberto.capelli/europe-is-protecting-its-factories-but-what-if-the-real-battle-starts-before-the-factory-c6fab92ca739?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/c6fab92ca739</guid>
            <category><![CDATA[stocks]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[energy]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Fri, 29 May 2026 13:05:20 GMT</pubDate>
            <atom:updated>2026-05-29T13:05:20.680Z</atom:updated>
            <content:encoded><![CDATA[<p><em>When Brussels talks about China, tariffs and industrial policy, another question appears behind the scenes: who will have access to the metals behind power grids, AI infrastructure and new manufacturing?</em></p><p><a href="https://www.reuters.com/world/china/european-commission-debates-policy-shift-protect-industry-china-2026-05-29/">Europe is increasingly talking about China not as a distant trading partner, but as an industrial challenge</a>. Reuters reports that the European Commission is debating a tougher approach to protecting its own industry, including in metals, chemicals and clean energy technology.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/827/1*DYitWMHWWzIN_pI1drQVBA.png" /><figcaption><a href="https://www.reuters.com/world/china/european-commission-debates-policy-shift-protect-industry-china-2026-05-29/">https://www.reuters.com/world/china/european-commission-debates-policy-shift-protect-industry-china-2026-05-29/</a></figcaption></figure><p>On the surface, this looks like another trade dispute. But to me, the real fear runs deeper: Europe is afraid of losing control over what its future economy is built from.</p><p>Factories can be subsidized. Laws can be changed. But if cables, transformers, electric motors, grids, EV infrastructure, data centers and clean energy systems depend on outside supply chains, industrial independence remains a beautiful phrase.</p><p>Copper does not shout from the headlines in this story. It quietly runs through almost everything Europe wants to protect: energy, automation, defense, charging networks, manufacturing and digital infrastructure. That is why copper supply is gradually becoming part of the conversation about industrial security.</p><p>As markets look for more predictable sources of metals, attention shifts toward stable jurisdictions and early copper-gold projects. The Wilmac Project in British Columbia, advanced by NovaRed Mining, sits in the Quesnel porphyry belt roughly 10 km from the Copper Mountain Mine. Stories like this matter not because they solve the problem in one day, but because future supply chains begin long before the first tonne of metal is produced.</p><p>Europe’s debate about China shows one simple thing: industrial strength is no longer measured only by the number of factories. It begins earlier — with geology, access to resources, mining and the ability to control the material foundation of technology.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c6fab92ca739" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Moving Less Dirt, Finding More Metal: Sensor-Based Sorting]]></title>
            <link>https://medium.com/@gilberto.capelli/moving-less-dirt-finding-more-metal-sensor-based-sorting-6d029d70d358?source=rss-e00c48f18700------2</link>
            <guid isPermaLink="false">https://medium.com/p/6d029d70d358</guid>
            <category><![CDATA[mining]]></category>
            <dc:creator><![CDATA[Gilberto Capelli]]></dc:creator>
            <pubDate>Tue, 03 Feb 2026 13:27:57 GMT</pubDate>
            <atom:updated>2026-02-03T13:27:57.285Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*P_dotxatZZ8PiKYm.jpg" /></figure><p>The most energy-intensive part of mining is crushing and grinding rock. What if we only crushed the rock that actually contained metal? Sensor-based sorting (SBS) uses X-rays and near-infrared sensors to scan rocks on a conveyor belt, ejecting waste before it ever reaches the mill. This technology can reduce energy and water usage by 30%. In an era of declining ore grades, efficiency isn’t just a bonus - it’s the only way to stay profitable.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6d029d70d358" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>