<?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 Gustavo Herzer on Medium]]></title>
        <description><![CDATA[Stories by Gustavo Herzer on Medium]]></description>
        <link>https://medium.com/@gustavo.herzer?source=rss-88b87cff7b99------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*0Fnu4OlOz1OEKAfzUvoe3g.png</url>
            <title>Stories by Gustavo Herzer on Medium</title>
            <link>https://medium.com/@gustavo.herzer?source=rss-88b87cff7b99------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 21:48:04 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gustavo.herzer/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[NRED Is Building the Business Side of Its Mining Operating System]]></title>
            <link>https://medium.com/@gustavo.herzer/nred-is-building-the-business-side-of-its-mining-operating-system-b8befe01d4ff?source=rss-88b87cff7b99------2</link>
            <guid isPermaLink="false">https://medium.com/p/b8befe01d4ff</guid>
            <category><![CDATA[business]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[robotics]]></category>
            <category><![CDATA[money]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Gustavo Herzer]]></dc:creator>
            <pubDate>Wed, 22 Jul 2026 13:42:13 GMT</pubDate>
            <atom:updated>2026-07-22T13:42:13.018Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*si6FQf4IZTDtqQa4nuCzQQ.png" /></figure><p>A technology stack is not a business. A company can assemble geological data, machine-learning models, cameras, drones and robotics expertise — and still have no clear product, customer or route to market.<br>That is the gap NRED now appears to be addressing. MetalCore supplies the geological-intelligence layer. EyeX could add computer vision and field monitoring. Robotics creates a longer-term path toward physical execution. Valentin Saitarli gives the technology program an internal executive owner, while Lee Evan Caplin adds experience in intellectual property, commercialization and strategic partnerships.<br>The individual pieces remain early. EyeX is still subject to due diligence, MetalCore has not announced recurring software revenue, and NRED has not deployed robotic drilling or autonomous sampling. Still, the structure is becoming more coherent. NRED is no longer presenting AI as one isolated exploration tool. It is beginning to connect data, field visibility, technical leadership and commercialization around a possible intelligent-mining operating system.</p><h3>The technology stack is taking shape</h3><p><a href="https://www.newsfilecorp.com/release/304193">MetalCore’s analytical dataset has grown beyond <strong>4.1 million records</strong>, including approximately <strong>1.5 million geochemistry records</strong>, <strong>1.4 million oil and gas well records</strong>, more than <strong>800,000 deposit records</strong> and nearly <strong>200,000 geothermal records</strong>. NRED also reported <strong>920 registrations</strong> for an early-access group capped at <strong>1,000 users</strong>.</a><br><br>The registrations are not paying customers, but they move MetalCore beyond an internal tool used only by NRED. An outside user base gives the company a chance to learn which functions people actually use, which property-level questions matter most and whether users return after their first analysis.<br>The raw data is only part of the potential value. Much of it comes from public geological records that other explorers can also access. MetalCore’s advantage will depend on what NRED builds above those records: how information is cleaned, connected and ranked; how gaps are identified; and how the platform turns scattered reports into a decision that saves time or capital.<br><a href="https://www.newsfilecorp.com/release/298327">NRED has filed U.S. Patent Application <strong>№19/680,101</strong>, covering systems related to integrated geological data, probabilistic mineral scoring, parcel-level evaluation and transaction management.</a> The filing is an application rather than a granted patent, but it shows that NRED is attempting to protect the workflow surrounding MetalCore rather than treating it as an ordinary database.<br><br><a href="https://www.newsfilecorp.com/release/304884">EyeX could add a different layer. NRED entered a <strong>30-day mutual due-diligence standstill</strong> concerning a possible acquisition of an interest in EyeX’s computer-vision technology. </a>The system is designed to monitor cameras, drones and other visual feeds, with proposed mining applications including field-progress tracking, equipment risks, environmental events, staff location and site security.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZNeNmKh93zT9IqbPKNRb8w.png" /></figure><p>No transaction had been completed when the agreement was announced. The strategic connection is nevertheless clear: MetalCore can help decide where work should happen, while computer vision could help document what happens after crews and equipment arrive. New field information could then return to the geological model.<br>NRED has also appointed Saitarli as chief technology officer and Dr. Olamide Oladeji as an advisor in robotics and AI. Robotics remains the least developed part of the stack, but the intended loop is becoming easier to understand: geological data informs target selection, field systems collect new observations, and the next model is updated with better evidence.</p><h3>Wilmac can provide the proof</h3><p>NRED’s technology story remains connected to a real exploration asset. The company holds an option to earn a <strong>70% interest</strong> in the <strong>16,078-hectare Wilmac Project</strong>, located in British Columbia’s Quesnel porphyry belt, approximately <strong>10 kilometres west</strong> of the producing Copper Mountain Mine.<br><a href="https://www.newsfilecorp.com/release/301870">Wilmac gives NRED exposure to a copper-gold-platinum and broader polymetallic exploration opportunity. MetalCore has already been used to organize <strong>10 mineral occurrences</strong>, approximately <strong>19 historical assessment reports</strong>, <strong>38 regional geochemical samples</strong>, historical production information and regional aeromagnetic data related to the property.<br>The company’s planned field program includes <strong>970 soil samples</strong> and <strong>four IP/AMT geophysical grids</strong> across North Lamont, West Lamont, Wilmac and Plume.</a><br><br>Those numbers matter less than what NRED can eventually demonstrate happened because MetalCore was used. Did the platform reduce the time required to review decades of historical work? Did it identify overlooked relationships? Did it change the order in which targets received capital? Did modern field results support the areas ranked by the system?<br>Wilmac can become MetalCore’s first detailed case study. The property provides real geological problems, while the platform provides a repeatable method for organizing the evidence behind them.<br>That relationship strengthens both sides of NRED. Better field results can improve the understanding of Wilmac, while documented improvements in target selection can make MetalCore more credible to outside users.<br>The key is that Wilmac should become the proof, not the entire market. A platform tested only on NRED’s own property may remain an internal exploration advantage. A platform that produces useful results across outside projects begins to look like a separate commercial product.</p><h3>Caplin adds the business layer</h3><p><a href="https://www.newsfilecorp.com/release/305915">Lee Evan Caplin’s appointment matters because NRED directly connected it to MetalCore’s commercialization.</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/654/1*yQfmC47UXvKGWclsi5X2Xw.png" /></figure><p>Caplin brings more than <strong>40 years of experience</strong> involving intellectual property, technology ventures and strategic partnerships. His role is unlikely to involve improving geological scoring models. The more important questions concern how MetalCore reaches customers.<br>Who should use the first paid version: junior explorers, geological consultants, property vendors or larger mining companies? Should NRED begin with direct software access, a structured property-assessment report or a paid pilot? Which functions should remain proprietary? Which parts should integrate with tools geologists already use?<br><br>These questions may sound less exciting than another AI model, but they determine whether the technology becomes a business. Mining companies do not purchase databases because they contain millions of records. They pay for outcomes: faster property reviews, fewer missed reports, clearer target comparisons and better allocation of exploration capital.<br>Distribution may be just as important as product design. Industrial software rarely scales through online registrations alone. Customers need implementation support, confidence in data security, compatibility with existing workflows and evidence that the technology works on real projects.<br>Caplin may help NRED identify partnerships that provide access to those customers. A strategic partner could contribute distribution, private datasets, industry credibility or integration with established geoscience platforms. NRED would not need to build every channel independently.<br>The wider market has already shown that subsurface software can become a valuable standalone business. In 2021, Bentley Systems agreed to acquire <a href="https://investors.bentley.com/news-releases/news-release-details/bentley-systems-enters-105-billion-agreement-acquire-seequent/">Seequent for approximately <strong>US$1.05 billion</strong>, with Seequent expected to generate more than <strong>US$80 million in annual recurring revenue</strong> that year.</a><br><br>MetalCore is far earlier and should not be compared with Seequent on valuation or commercial maturity. The example establishes one narrower point: software used to interpret geological information can scale independently of ownership of the mineral properties being analyzed.</p><h3>The next numbers will measure the business</h3><p>NRED has already produced several large headline figures: <strong>4.1 million records</strong>, <strong>920 registrations</strong>, a <strong>16,078-hectare</strong> exploration property and a field program built around <strong>970 soil samples</strong>.<br>The next numbers may be less dramatic but more important.<br>How many early users complete onboarding? How many analyze a real property? How many return for a second project? Does NRED secure a paid pilot? Can an outside company show that MetalCore changed a genuine exploration decision?<br>The same applies to EyeX. A broad list of possible mining applications will matter less than one measurable result, such as reducing the hours required to monitor remote sites or identifying an operational problem sooner.<br>NRED has not yet announced MetalCore pricing, paid enterprise clients, recurring software revenue, completed licensing agreements or a finished EyeX acquisition. Those milestones should not be assumed. They are the evidence by which the business strategy should now be judged.<br>Still, the company has assembled more than a typical junior explorer adding AI language to an investor presentation. It has millions of analytical records, external early-access interest, a patent strategy, internal technical leadership, robotics expertise, a possible computer-vision layer and a real exploration environment in which the system can be tested.<br>The technology side of NRED’s mining operating system is beginning to take shape.<br>Caplin may help build the other side: a focused product, protected intellectual property, strategic distribution and customers willing to pay.<br>A mining operating system becomes commercially real when it moves beyond helping NRED make its own decisions.<br>It becomes a business when someone outside NRED wants to use it too.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b8befe01d4ff" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Geological Data to Real-Time Decisions: NRED Next Technology Layer]]></title>
            <link>https://medium.com/illumination/from-geological-data-to-real-time-decisions-nred-next-technology-layer-8319b6ed58d5?source=rss-88b87cff7b99------2</link>
            <guid isPermaLink="false">https://medium.com/p/8319b6ed58d5</guid>
            <category><![CDATA[money]]></category>
            <category><![CDATA[robotics]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[stocks]]></category>
            <dc:creator><![CDATA[Gustavo Herzer]]></dc:creator>
            <pubDate>Mon, 13 Jul 2026 18:30:41 GMT</pubDate>
            <atom:updated>2026-07-22T13:36:48.258Z</atom:updated>
            <content:encoded><![CDATA[<p>Finding a promising target is only half the job. The next stage is just as important: organizing fieldwork so the team can see what is happening across the property, how equipment is moving, where road conditions are changing, and which decisions need to be made faster.</p><p>That is where the combination of MetalCore and EyeX becomes especially interesting.</p><p>MetalCore can help NRED determine where the next work should take place. EyeX potentially adds the next layer — a digital view of what is happening after the field team arrives.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/444/1*8w39GoVur198scuw58T3wA.png" /><figcaption>AI Image</figcaption></figure><p><a href="https://novaredmining.com/news/novared-mining-to-close-metalcore-early-access-waitlist-in-48-hours-as-platform-dataset-surpasses-4.1-million-records">By early July, the MetalCore database had surpassed 4.1 million records. Approximately 1.5 million of those were geochemical records, while another 1.4 million covered oil and gas wells. A total of 920 users registered for the first early-access group out of a planned 1,000.</a></p><p>At Wilmac, a system like this has a clear practical purpose. The project covers 16,078 hectares, so the challenge is not simply to collect more information. It is to compare historical reports, geochemistry, geophysics, mineral occurrences, and new field results quickly enough to make better decisions.</p><p>What I like here is the logic behind the process. MetalCore can help narrow a large property down to stronger zones. The field team can then work with greater precision instead of spending the same amount of time on every part of the project.</p><p>Once a target has been identified, however, a new stream of information begins. People, equipment, and drones arrive on site. Roads, temporary work areas, and machinery are used every day. Conditions keep changing, and traditional reports are no longer enough to capture everything.</p><p>EyeX could give NRED the ability to see those changes almost in real time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/452/1*qmbTpgl6ZCnWQSxLnLTl9Q.png" /><figcaption>Ai Image</figcaption></figure><p><a href="https://novaredmining.com/news/novared-enters-standstill-agreement-for-mining-surveillance-technology-and-appoints-valentin-saitarli-as-chief-technology-officer">In July, NRED took an initial step toward a possible technology partnership with EyeX, giving the company time to study the platform in detail and consider potential forms of future cooperation. EyeX works with cameras, drones, and video streams, turning visual information into practical signals for site management.</a></p><p>In mining, that could mean much more than conventional video surveillance. The system could potentially help track work progress, equipment condition, road activity, team movements, and the overall situation at a remote site.</p><p>For a project manager, that could be the difference between receiving a report at the end of the day and receiving a signal while there is still time to act.</p><p>Imagine a simple scenario. MetalCore identifies a promising area. The team moves into the field. Drones and cameras record the progress of the work. Computer vision detects changes and helps reveal where delays are developing, how equipment is being used, or which routes are operating more efficiently.</p><p>The human role does not disappear. The team simply gains access to more information for better decisions.</p><p>That is how technology can move from geological analysis into the management of real field operations.</p><p>This is especially important at remote projects, where maintaining a large permanent team can be expensive and difficult. Cameras, drones, sensors, and computer vision systems can create a digital view of the site even where connectivity is limited and distances are significant.</p><p>Over time, this model could develop even further. Once a system can see equipment, roads, people, and the progress of fieldwork, the same data could eventually support route planning, reduced downtime, drone coordination, and future automation.</p><p>To me, this is a logical direction for the next generation of mining companies.</p><p>First, AI helps the company understand the geology more clearly. Then computer vision makes it possible to observe field operations in real time. Later, autonomous machines, robotics, and remote-control systems could become part of the same structure.</p><p>NRED is already beginning to bring these separate technologies into a broader model.</p><p>MetalCore can help identify opportunity before fieldwork begins. EyeX could give the company a digital view of what is happening once work is underway. Together, they create a bridge between geological data and real-time decisions.</p><p>That combination could gradually help transform NRED from a traditional junior explorer into a mining company built for the future.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8319b6ed58d5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/illumination/from-geological-data-to-real-time-decisions-nred-next-technology-layer-8319b6ed58d5">From Geological Data to Real-Time Decisions: NRED Next Technology Layer</a> was originally published in <a href="https://medium.com/illumination">ILLUMINATION</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Korea’s Offshore Wind Runs Into the Shore]]></title>
            <link>https://medium.com/@gustavo.herzer/koreas-offshore-wind-runs-into-the-shore-7f5b38c7cfa3?source=rss-88b87cff7b99------2</link>
            <guid isPermaLink="false">https://medium.com/p/7f5b38c7cfa3</guid>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[infrastructure]]></category>
            <dc:creator><![CDATA[Gustavo Herzer]]></dc:creator>
            <pubDate>Wed, 01 Jul 2026 18:36:09 GMT</pubDate>
            <atom:updated>2026-07-01T18:36:09.181Z</atom:updated>
            <content:encoded><![CDATA[<p>Offshore wind is always sold through a beautiful image. A flat horizon, turbines at sea, clean electricity, a big number in MW. But a real project rarely breaks on the image itself. It breaks on the shore — where the cable has to come out of the water, pass through permits, connect to a substation and avoid fighting with local communities.</p><p><a href="https://www.techtimes.com/articles/319422/20260701/south-korea-taps-state-power-group-800-mw-offshore-wind-farm-jeonbuk.htm">That is why Jeonbuk Southwest 1 is more interesting than just another wind farm in the news. A consortium led by Korea Hydro &amp; Nuclear Power was selected as preferred bidder for an 800 MW offshore wind project near Buan County in North Jeolla Province. It is the first phase of a larger 2.4 GW complex. The implementation agreement with Jeonbuk Province is expected to be signed by the end of August 2026.</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/971/1*6IBUGhU8AmyAGhLgY9mg2w.png" /></figure><p>800 MW sounds like the main number. But the consortium structure says more. KHNP is working with Hanwha Ocean, Doosan Enerbility, Korea East-West Power, KEPCO KPS, KEPCO Engineering &amp; Construction, IBK Financial Group, KB Financial Group, Samil C&amp;S and Jungang Ocean Heavy Industries. These are not just “partners for a press release.” Korea is pulling energy, shipbuilding, turbines, engineering, finance, service and local industry into one project.</p><p><a href="https://www.offshorewind.biz/2026/06/29/south-koreas-jeonbuk-state-names-preferred-bidder-for-800-mw-offshore-wind-project/">Another detail matters more than it may seem: public institutions are expected to hold more than 50% equity. So offshore wind in this structure does not look like a purely private renewable asset.</a> The state wants to stay inside the project. And that makes sense, because this is no longer only about decarbonization. It is about ports, jobs, supply chains, grid planning, vessels, heavy industry and control over how electricity from the sea enters the system.</p><p>Korea is not just trying to install more turbines. It is changing the rules for offshore wind. The new Special Act on Offshore Wind Farm Promotion, which came into force in March 2026, is meant to move the country away from the old developer-led model, where there were many applications, a lot of friction and not much speed. The new logic is more state coordination, clearer zones and less permitting chaos.</p><p>But even a good law does not replace metal in the water. 800 MW means foundations, installation vessels, an offshore substation, inter-array cables, an export cable, an onshore substation, transformers and a connection to the KEPCO grid. Using modern 12–15 MW turbines as a rough reference, a project like this could need about 53–67 turbines. Each one has to be installed, connected, maintained and protected from an environment that does not forgive mistakes.</p><p>The least photogenic part may be the most important. The wind in Jeonbuk will be off the southwest coast. Major electricity demand sits closer to the Seoul metropolitan area. Between them are transmission, high-voltage routes, onshore substations and years of planning. A turbine can generate electricity at sea, but the economy does not need generation “somewhere out there.” It needs power where factories, data centers, ports and cities actually are.</p><p>There is also a human layer that often sounds boring until it stops a project. In Korea, offshore wind has long run into fishing communities, military clearances, permits and the question of who actually benefits from construction. So “regional co-prosperity” is not just a soft bureaucratic phrase. At this scale, it can be the difference between a project that moves and a project that spends years explaining itself locally.</p><p>That is why “wind is free” is a very incomplete phrase. The wind really does not send a bill. But the foundation, vessel, cable, substation, transformer, permit, port logistics and grid connection do. Offshore wind in the real world is heavy, expensive and deeply material. Part of that weight is simply hidden under water or inside paperwork.</p><p>Copper does not need to be searched for separately in a project like this. It is already in the export cables, inter-array cables, generators, transformers, switchgear, substations, motors and grid equipment. Before it becomes part of an offshore connection, it has to be found, tested, permitted, mined and processed. A cable for offshore wind is not born the moment a country announces a new gigawatt target.</p><p>That is why early copper exploration matters long before the visible shortage appears. <a href="https://novaredmining.com/">NovaRed Mining is working on the Wilmac Copper-Gold Project in British Columbia — roughly 16,078 hectares in the Quesnel porphyry belt.</a> For 2026, the company plans expanded soil sampling and IP/AMT geophysical surveys across North Lamont, West Lamont, Wilmac and Plume. This is still a very early stage, but every future cable starts not at the factory, but with geology.</p><p>Jeonbuk Southwest 1 shows that offshore wind is not just 800 MW at sea. It is a long road from the seabed to urban demand. And the main test may not be where the blade spins, but where the cable comes ashore and asks the grid to accept new electricity.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7f5b38c7cfa3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Microsoft Is Not Buying Gas. Microsoft Is Buying Time]]></title>
            <link>https://medium.com/@gustavo.herzer/microsoft-is-not-buying-gas-microsoft-is-buying-time-7e18916051d9?source=rss-88b87cff7b99------2</link>
            <guid isPermaLink="false">https://medium.com/p/7e18916051d9</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[infrastructure]]></category>
            <category><![CDATA[materials]]></category>
            <category><![CDATA[investing]]></category>
            <dc:creator><![CDATA[Gustavo Herzer]]></dc:creator>
            <pubDate>Tue, 23 Jun 2026 18:30:50 GMT</pubDate>
            <atom:updated>2026-06-23T18:30:50.196Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="https://www.reuters.com/legal/litigation/chevron-signs-power-supply-deal-with-microsoft-texas-data-center-2026-06-22/">Chevron has signed a 20-year agreement with Microsoft to supply power for a data center campus in Pecos, West Texas. The project is called Kilby, and this is not a small backup-power story.</a> It is a co-located natural gas-fired power facility that is expected to eventually reach about 2.67GW of capacity. First power is expected in 2028, and Chevron plans to make a final investment decision by the end of 2026.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/992/1*aUk-egPVOIRq-L28MLvcNA.png" /></figure><p>On paper, this looks like another large energy deal. I would read it a little differently.</p><p>This is the moment when AI infrastructure is basically saying: waiting in the grid queue takes too long. If a data center wants to operate, it does not need just a “green target” in a presentation. It needs electricity in a specific place, in a specific year, at a specific scale.</p><p><strong>AI is no longer asking where land is cheaper. It is asking where megawatts can arrive faster.</strong></p><p><a href="https://www.chevron.com/newsroom/2026/q2/chevron-signs-20-year-power-agreement-with-microsoft-for-west-texas-data-center">That is why this story is so revealing. Microsoft is not simply signing a power supply deal. It is moving into a model where the data center and the generation source sit next to each other.</a> Chevron, Engine №1, GE Vernova, and Caterpillar through Solar Turbines — this is no longer the classic model where a tech company buys electricity somewhere “in the system.” It looks much closer to an industrial campus where compute and power grow as one object.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/924/1*gYkAiH1yqn0Cf9XAEmJyZQ.png" /></figure><p>And the 2.67GW number sounds almost blunt. This is the kind of scale that cannot be hidden behind words like “cloud” or “AI services.” It is physical demand. Turbines, gas, substations, transformers, switchgear, transmission equipment, water strategy, permits, construction crews and years of buildout. Reuters writes that the campus could add around 2GW of data center capacity for Microsoft and support more than 6,000 construction jobs, plus hundreds of permanent operational roles.</p><p>I think this shows the new phase of the AI boom very clearly. First the market talked about chips. Then it talked about data centers. Now it is talking about power plants. And that makes sense: if a GPU does not have stable electricity, it is not infrastructure. It is just expensive equipment waiting for current.</p><p><strong>In the AI cycle, the power plant becomes part of the server. It just does not sit inside the rack. It sits next to the campus.</strong></p><p>Of course, the natural gas angle will be controversial. Microsoft has its own climate targets, and gas-fired generation for AI does not look like the perfect “clean cloud” picture. But that is exactly why this news is honest. It shows the real tension between the speed of AI buildout and the pace of clean energy and grid expansion. When demand arrives at gigawatt scale, the market starts choosing not the prettiest model, but the one that can be built on time.</p><p>And here the materials layer appears again. Because 2.67GW is not only about fuel and turbines. It is electrical infrastructure, where copper, steel, transformers, cables, busbars, switchgear and substation equipment become part of the AI supply chain.</p><p>In the copper-gold space, I would not look only at operating mines. <a href="https://novaredmining.com/">NovaRed Mining</a> is working on the Wilmac Copper-Gold Project in British Columbia — 16,078 hectares in the Quesnel porphyry belt, roughly 10 km west of Copper Mountain Mine. Its 2026 field program includes expanded soil sampling, four IP/AMT surveys on the North Lamont, West Lamont, Wilmac and Plume grids, as well as an initial drill program subject to an approved permit. This is an early stage, but the future grid-heavy economy begins with upstream maps like these.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/557/1*kS0HMTRtDvvU5TYu00u-YQ.png" /></figure><p>Project Kilby matters not because Chevron has suddenly become an “AI company.” It matters because Microsoft is showing a new logic: for large-scale AI, it is no longer enough to rent racks and wait for the grid.</p><p>You have to buy the path to electricity.</p><p>Because in the next phase of AI, the winner will not be the company with the loudest model. It will be the one that connects 2.67GW to it first.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7e18916051d9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The City Tripped Over Its Smallest Infrastructure]]></title>
            <link>https://medium.com/@gustavo.herzer/the-city-tripped-over-its-smallest-infrastructure-4d42f7635c8f?source=rss-88b87cff7b99------2</link>
            <guid isPermaLink="false">https://medium.com/p/4d42f7635c8f</guid>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[infrastructure]]></category>
            <category><![CDATA[copper]]></category>
            <category><![CDATA[materials]]></category>
            <category><![CDATA[money]]></category>
            <dc:creator><![CDATA[Gustavo Herzer]]></dc:creator>
            <pubDate>Tue, 16 Jun 2026 18:35:28 GMT</pubDate>
            <atom:updated>2026-06-16T18:35:28.627Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>Waterbury is repairing sidewalks, and that sounds ordinary. But projects like this show where a city really breaks — not on major highways, but under people’s feet.</em></strong></p><p><a href="https://www.ctinsider.com/waterbury/article/waterbury-ct-sidewalk-improvement-project-22302799.php">Waterbury has announced a $3,483,612 sidewalk improvement project scheduled to begin in July 2026 and run through June 2027</a>. The contract was awarded to Colonna Masonry Concrete and Asphalt Paving LLC. On the surface, this does not look like a major infrastructure story. Not a billion-dollar bridge, not a data center, not a new power plant. Just sidewalks, curbs, crossings, accessibility, and parts of the city that have needed repair for a long time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/834/1*dVdJN4Hs2ozXiVBYYki2eA.png" /></figure><p>But there is something interesting in news like this. It shows infrastructure in places where we are not used to looking for it. We often talk about grids, highways, ports, bridges, and power plants. But for an ordinary person, a city starts much more simply: can you walk safely down the street? Can a child get to school? Can an older person cross the road without fear? Can someone in a wheelchair avoid going around an entire block because of a broken curb?</p><p>The list of streets includes North Walnut Street, Rose Street, Huntingdon Avenue, Peach Orchard Road, Branch Street, Congress Avenue, East Main Street, Hawkins Street, and parts of Baldwin Street. This is not one point on a map. It is a scattered set of small urban pain points. One place needs a new sidewalk run, another needs a curb, another a ramp, another pavement restoration. And that is exactly why the project matters. It does not create one beautiful image. It fixes many small places where the city rubs against its own wear every day.</p><p>To me, the sidewalk is one of the most underrated parts of infrastructure. People notice it only when it is broken. When everything works, it simply exists. A person walks, a stroller rolls, a bus stop is reachable, and the pedestrian route does not suddenly disappear. But when concrete sinks, the curb is too high, the traffic signal is hard to see, or the road forces someone into the street, the city suddenly stops being convenient.</p><p>Alongside the sidewalk work, Waterbury also allocated $89,130 to A&amp;S Electrical Services Inc. to replace traffic signal lights with LED models carrying a 10-year warranty. That is another layer of the same story. A safe street is not only concrete underfoot. It is also light, signals, visibility, wiring, electrical cabinets, maintenance costs, and small pieces of urban electricity that nobody notices while they work.</p><p>This is where an ordinary sidewalk project begins to look wider. The city is repairing the surface, but beneath that surface sits a whole material chain: concrete, asphalt, drainage, pipes, signals, wiring, steel, copper, lighting, and contractor labor. Even a small traffic signal needs metal, cables, electronics, and maintenance. If a city wants to be safe and accessible, it needs more than plans. It needs materials, workers, and a budget for things that do not look heroic.</p><p>Copper does not appear in the headline of this story, but it is inside the body of the city. It sits in signals, LED traffic systems, electrical connections, control cabinets, lighting infrastructure, and the broader network that makes the road manageable. And when hundreds of cities are upgrading crossings, signals, chargers, lighting, and urban grids at the same time, “small” materials demand suddenly adds up to something much larger.</p><p>Somewhere near the beginning of that picture is the search for raw materials. <a href="https://novaredmining.com/">NovaRed is working with the Wilmac Copper-Gold Project in British Columbia and developing MetalCore</a> — a platform that uses geological, geochemical, geophysical, and historical data to identify critical-minerals opportunities. This is far from a sidewalk in Connecticut and has no connection to Waterbury’s specific contract. But before every traffic light, cable, and urban electrical cabinet, there is a slower stage — the metal that has to be found long before it is called part of “public safety.”</p><p>Waterbury also updated the load rating for the Frost Road Bridge over the Mad River: 17 tons for single-axle emergency vehicles, 24 tons for tandem-axle vehicles, and a maximum gross vehicle weight of 33 tons. That sums up the whole story well. Urban infrastructure does not always look like a major breakthrough. Often, it is a simple question: can a fire truck pass, can the signal be seen, and can a person cross the street without tripping?</p><p>On paper, this is a sidewalk improvement project. In real life, it is a repair of the layer of the city that most people only notice when it stops working.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4d42f7635c8f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Tonne Hasn’t Arrived Yet]]></title>
            <link>https://medium.com/@gustavo.herzer/the-tonne-hasnt-arrived-yet-956a6f158b06?source=rss-88b87cff7b99------2</link>
            <guid isPermaLink="false">https://medium.com/p/956a6f158b06</guid>
            <category><![CDATA[money]]></category>
            <category><![CDATA[macroeconomics]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[stocks]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Gustavo Herzer]]></dc:creator>
            <pubDate>Fri, 12 Jun 2026 16:55:19 GMT</pubDate>
            <atom:updated>2026-06-12T16:55:19.370Z</atom:updated>
            <content:encoded><![CDATA[<p>Copper may have a beautiful price on the screen. But the real question now is different: can it complete the whole route to the cable?</p><p><a href="https://www.ft.com/content/44d4b188-45ed-4e7e-a5eb-278da62ac601">Financial Times described the situation in industrial metals as a “super-squeeze” — not just another growth cycle, but pressure coming from several sides at once. Copper and aluminium were already strong because of data centers, power grids, EVs and clean energy</a>. Then the Middle East was added to the equation, along with risks around the Strait of Hormuz, more expensive fuel, sulfur issues and logistics. Suddenly, metal stopped being just a line in a trading terminal. It became cargo again — something that actually has to be delivered.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/847/1*FB-rhF2v5DME2e9hmuQ2OQ.png" /></figure><p>What caught my attention in this story was not the phrase “super-squeeze” itself. Markets like phrases like that. The more interesting part is this: according to FT, Goldman Sachs raised its forecast for the copper deficit outside the U.S. from 60,000 tonnes to 640,000 tonnes and lifted its price forecast by 10% to $13,735 per tonne by year-end. That is not a cosmetic adjustment. It is almost an admission that the previous supply map looked too calm.</p><p>Copper has one simple problem: it does not respond to demand as quickly as a stock responds to a headline. AI workloads can expand in months. A new data center can be announced at 100 MW or more. A power grid upgrade can be added to a government plan. But a mine, processing, acid, diesel, ports, insurance, vessels and cables do not move at the same speed. What happens when digital demand is flying, while the physical route is still walking?</p><p>Aluminium also showed that “infrastructure metals” no longer live in a vacuum. According to FT, the Middle East accounts for almost 10% of global refined aluminium production, and some producers in the region have already limited output because of infrastructure damage and problems with alumina deliveries. That does not sound like an AI story. But server racks, transmission lines, EVs, solar farms and industrial sites suddenly remember very quickly what they are made of.</p><p>It is easy here to look only at the LME and the large mines. But the chain begins much earlier — with land packages, rights, field data and years of quiet work. <a href="https://novaredmining.com/">NRED, with Wilmac in British Columbia</a>, sits in exactly that early part of the cycle: not finished metal in a cable yet, but the level without which the future cable often exists only in a presentation. The company has an option for a 70% interest in Wilmac, exploration commitments of up to $8.5M and $1.5M staged in 2026. In a market where copper is increasingly running into availability rather than demand, these early-stage stories start to look less abstract.</p><p>The most uncomfortable question here is not whether copper is needed. Almost everyone needs it. The question is who gets it first when data centers, grid upgrades, EV factories, defense, construction and renewable projects all show up at the same time. And is the buyer ready to pay not only for the tonne of metal, but also for the confidence that this tonne will not get stuck somewhere between a port, acid and politics?</p><p>I would not romanticize this squeeze. For the economy, it is not just “bullish metals.” It means more expensive projects, more difficult budgets and more delays in places where everyone got used to drawing straight arrows from left to right. If the supply chain starts to creak, even the cleanest electrification plan becomes dependent on very boring things: diesel, permits, warehouses, contracts, insurance and people on the ground.</p><p>The digital economy loves speed. It talks about AI, automation, cloud and instant decisions. But copper does not arrive through an API. It has to be extracted, refined, transported, insured, sold, processed and finally placed into a cable.</p><p>And maybe that is exactly what the metals market is showing now. The problem is not only that the tonne has become more expensive. The problem is that it still has to arrive.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=956a6f158b06" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Factory Let the Robot In. Then the Hard Part Began]]></title>
            <link>https://medium.com/@gustavo.herzer/the-factory-let-the-robot-in-then-the-hard-part-began-d02346861ed3?source=rss-88b87cff7b99------2</link>
            <guid isPermaLink="false">https://medium.com/p/d02346861ed3</guid>
            <category><![CDATA[financial-markets]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[copper]]></category>
            <category><![CDATA[technology-news]]></category>
            <dc:creator><![CDATA[Gustavo Herzer]]></dc:creator>
            <pubDate>Wed, 03 Jun 2026 15:43:32 GMT</pubDate>
            <atom:updated>2026-06-03T15:43:32.335Z</atom:updated>
            <content:encoded><![CDATA[<p>When I saw the news about PUDU D7, I did not stop at the fact that another robot had entered the industrial world. That part already feels familiar. New machines with arms, wheels, sensors, and polished demonstrations are becoming part of the background noise of automation.</p><p>What caught me was something smaller: the moment a robot reaches toward a real object in a real factory.</p><p>A box is never just a box in that moment. It has weight, position, edges, friction, distance from a person, and the possibility of a mistake. A warehouse is not a clean simulation. Someone walks past. A shelf is slightly off. A route changes. A part needs care, not strength. And suddenly the question is not whether the robot can move. The question is whether it understands enough before it moves.</p><p><a href="https://www.prnewswire.com/news-releases/pudu-embodied-unveils-the-next-generation-pudu-d7-opening-a-new-chapter-for-industrial-semi-humanoid-robotics-302786976.html">Pudu Robotics introduced PUDU D7 as an industrial semi-humanoid robot for manufacturing and warehouse environments. It is built on PuduFM 1.0, a foundation model for embodied intelligence — the kind of AI that has to leave the screen and deal with physical space.</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/471/1*cxCq6rE61QJ86cUewOgIEA.png" /></figure><p>The headline numbers are easy to notice: up to 14 kg of payload, operation at heights up to 2 meters, tactile sensors, LiDAR, 360° environmental awareness, autonomous battery swapping, and support for 24/7 industrial operations. But the longer I looked at those specifications, the less they felt like a list of features.</p><p>They felt like a list of fears the robot has to survive.</p><p>It has to know how hard it is pressing. It has to notice what is behind it. It has to keep working when people are nearby. It has to recharge without waiting for someone to rescue it. It has to turn a messy physical environment into a readable map, again and again, without the luxury of pretending the world is perfectly arranged.</p><p>That is the real shift. Older industrial robots could live behind barriers, repeating controlled movements in controlled spaces. Machines like D7 are moving closer to shelves, workers, tools, routes, and decisions. In that world, strength matters. But attention matters more.</p><p>A robot becomes useful not when it can lift more. It becomes useful when it knows the cost of being wrong.</p><p>This is where embodied AI stops sounding like a clean software phrase. Behind it are motors, actuators, batteries, sensors, boards, copper conductors, charging stations, cables, controllers, casings, servers, and stable electricity. The more intelligence enters the factory, the more physical the factory becomes.</p><p>There is a quiet parallel here with mineral exploration. Before metal becomes part of a motor, sensor, cable, or charging system, someone has to read another kind of physical space: the ground. In British Columbia, <a href="https://www.tradingview.com/news/tmx_newsfile%3Ac1f69afc2094b%3A0-novared-reports-soil-geochemistry-results-from-north-lamont-target-area-wilmac-copper-gold-project/">NovaRed Mining</a> has been working with this early layer at North Lamont within the Wilmac Copper-Gold Project. A 43-sample soil geochemistry program was focused on a mapped pyroxenite exposure and a strong magnetic anomaly, while the area received No Permit Required authorization for an IP/AMT survey to better define a future target.</p><p>It is not the same industry, but the logic rhymes. A robot reads pressure, distance, shape, and motion before it acts. Geology reads anomalies, soil signals, structures, and old patterns before the metal becomes visible. In both cases, the future starts with a weak signal that has to be understood before anyone can trust the next move.</p><p>Maybe the most interesting part of PUDU D7 is not that it can lift 14 kg. It is that industry is asking machines for something harder than strength. It is asking them to enter the real world and behave as if the real world matters.</p><p>And once a robot touches that world, the future stops looking weightless. It becomes data, yes — but also metal, current, sensors, batteries, and the quiet ability to read the room before making a mistake.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d02346861ed3" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>