<?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 Niraj Jha  on Medium]]></title>
        <description><![CDATA[Stories by Niraj Jha  on Medium]]></description>
        <link>https://medium.com/@nkjhaa?source=rss-4555a45a3c1e------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*XiMxm-afk4JmEd9PrVzb6Q.jpeg</url>
            <title>Stories by Niraj Jha  on Medium</title>
            <link>https://medium.com/@nkjhaa?source=rss-4555a45a3c1e------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 01:51:04 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@nkjhaa/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[Throwing the Baby Out with the Bathwater: Will AI Really Kill SaaS?]]></title>
            <link>https://medium.com/@nkjhaa/throwing-the-baby-out-with-the-bathwater-will-ai-really-kill-saas-09db05fc2f50?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/09db05fc2f50</guid>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[software]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Mon, 16 Feb 2026 21:30:35 GMT</pubDate>
            <atom:updated>2026-02-16T21:30:35.400Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/688/1*RZ1a7a81tIy8emYEwjGzeg.png" /></figure><p>What do you do if you have invested billions in a home security company? You tell everyone they will get robbed. What do you do if you sell earthquake insurance? You remind people that the big one is always just around the corner. What do you do when you have poured nearly a trillion dollars into building AI infrastructure? You warn the world that every job may soon be obsolete.</p><p>Every technological revolution claims to be unprecedented. From internal combustion engines to mechanized farming, from frozen food to the internet, each wave promised to permanently displace human labor. Yet history shows a consistent pattern: constraints do not vanish; they shift. Work does not disappear; it reorganizes. Entire categories compress, new ones emerge, and the locus of value migrates to where the new bottlenecks form.</p><p>Is this time different? Maybe. No one knows for sure. But humans have repeatedly shown an uncanny ability to retreat into new economic crevices — cubicles, corner offices, and entirely new professions — each time a prior wave threatened to make their skills obsolete. Predictions that “all jobs will be lost” have appeared in nearly every major technological transition, and yet employment has persistently reconstituted itself around the new frontier of scarcity.</p><p>The recent volatility in software and SaaS valuations is revealing. Capital markets are implicitly signaling where they believe the return on the massive AI infrastructure buildout will come from: margin compression in asset-light software companies whose primary advantage has historically been human-written code and workflow abstraction. If AI dramatically lowers the cost of producing software logic, the economic rents of many software layers may shrink. But extrapolating that trend to conclude that most software companies will vanish is less analysis and more paranoia.</p><p>To understand the real impact, it is useful to deconstruct B2B SaaS companies into three functional layers. First are systems of record — platforms that store and govern critical business data. Second are the middleware and orchestration layers that move, transform, and structure that data across workflows. Third are the interface layers — the user-facing applications that present insights and enable interaction. In some products all three layers are bundled, but economically they behave very differently.</p><p>Consider the system-of-record layer first. These are platforms such as SAP, Oracle, Workday, and Salesforce. They do not merely provide software features; they store the canonical version of truth for finance, HR, customer relationships, and supply chain operations. Replacing them is not just a technical exercise but an organizational upheaval involving compliance risk, retraining costs, data migration, and years of process redesign. Even if AI can generate equivalent code, it cannot instantly replicate decades of embedded workflows, regulatory certifications, and enterprise trust. In many cases, AI will act as a tailwind for these platforms, allowing them to deliver richer automation, embedded intelligence, and decision support on top of already entrenched data foundations.</p><p>The middleware and orchestration layer sits above the systems of record and moves data between them. Companies such as ServiceNow, Snowflake, Databricks, and Palantir operate in this zone. Their value lies in harmonizing fragmented enterprise data and enabling workflows across silos. AI agents that can autonomously transform and route data will pressure parts of this layer, but they will also dramatically expand the total amount of data being generated, simulated, and consumed. As AI proliferates across business processes, the need for governed data movement, lineage, and cross-system coordination is likely to intensify, making this layer selectively disrupted but far from obsolete.</p><p>The most exposed layer is the interface and application tier — the tools that primarily provide user-facing workflows, reporting dashboards, or productivity abstractions. This includes segments of companies like Atlassian, Asana, Dropbox, and a long tail of niche vertical SaaS vendors whose differentiation often resides in front-end workflows rather than deeply embedded data ownership. If AI can generate interfaces, automate workflows, and dynamically compose applications on demand, the switching costs for this top layer decline meaningfully. The danger here is not immediate extinction, but commoditization pressure as features become easier to replicate and bundle into broader AI-native platforms.</p><p>Yet the layered view also reveals why the narrative that “AI will wipe out SaaS” is fundamentally incomplete. For many highly embedded enterprise software providers, AI is not a threat but a force multiplier. Most organizations lack the clarity, data readiness, and cross-functional discipline required to independently operationalize AI within their own processes. They possess the raw ingredients — data, workflows, and domain complexity — but not the integration fabric and execution rigor to convert that into sustained value. Vendors that already sit inside these workflows are uniquely positioned to embed AI capabilities directly into daily operations, offering dramatically more functionality for marginal incremental cost.</p><p>A useful analogy is Formula One racing. If every team were suddenly given access to the same world-class car, the race would not be decided by the machine alone. Execution, strategy, pit-stop coordination, driver skill, and team cohesion would matter even more. Large language models are rapidly becoming commoditized engines — powerful, accessible, and increasingly standardized. Possessing the engine does not guarantee victory; orchestrating it effectively does. In an enterprise context, the vendors that already understand customer processes, data nuances, and operational constraints will be best positioned to turn commoditized AI capability into differentiated outcomes.</p><p>The future, therefore, is unlikely to be a clean sweep where software companies vanish en masse. Instead, we are more likely to see a reordering of value across the stack. Lightweight application layers may face intense margin compression as AI lowers development costs and accelerates feature replication. Middleware and data orchestration platforms will be forced to evolve but may ultimately become more critical as AI multiplies the volume and velocity of enterprise data. Deeply embedded systems of record, far from being displaced, may emerge stronger by becoming the primary control planes through which AI-driven decisions are governed and executed.</p><p>The real disruption is not that AI will eliminate software companies, but that it will shift the economic gravity within software. Code will become abundant. Differentiation will move from writing software to embedding it deeply within mission-critical workflows, governing trusted data, and executing reliably at scale. Some firms will indeed be upended, particularly those whose value is thinly layered on the surface of enterprise processes. Others, especially those with high embeddedness and execution depth, may find that AI is not the force that replaces them, but the lever that makes them indispensable.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=09db05fc2f50" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Manifest 2026- Las Vegas]]></title>
            <link>https://medium.com/@nkjhaa/manifest-2026-las-vegas-55ade75d5a77?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/55ade75d5a77</guid>
            <category><![CDATA[manufacturing]]></category>
            <category><![CDATA[logistics]]></category>
            <category><![CDATA[manifest-2026]]></category>
            <category><![CDATA[supply-chain-management]]></category>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Sun, 15 Feb 2026 00:17:38 GMT</pubDate>
            <atom:updated>2026-02-15T00:17:38.193Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gRPQC-WPjb7eH7OSIJQ1eQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2a2ETtY_Q09e11o86HQFYA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UPrFyJpB14e06YLIajpDrw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HSrXJnHPIK1QDjo1XmgLUw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eOVLJFZil7DpiaktfH1wXA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mliyR3etdrZdpxHxwzVasw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rp4zJwpJuHL4rsmrQYriYg.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*B48ryMUx6AnDQ-wQCAYjsg.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XWaBJ6TZf-pRwD9GeWGceA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zQ2MblFpHv4rJ9Jf4HOa7g.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oeqO4RHGSqbAJCrp5cMsCw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*C_yf9plPPnWddaV9eNkBXg.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*51pkOneaoR7tHYd_KJWSAQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JUrPHifS2rMENrUJn-MZ5w.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gRPQC-WPjb7eH7OSIJQ1eQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2a2ETtY_Q09e11o86HQFYA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UPrFyJpB14e06YLIajpDrw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HSrXJnHPIK1QDjo1XmgLUw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eOVLJFZil7DpiaktfH1wXA.jpeg" /><figcaption>Panel Discussions during Manifest 2026, Las Vegas</figcaption></figure><p>Last week, I had the opportunity to participate in two panels at Manifest Las Vegas, one focused on the future of supply chain leadership and the other on cyber resilience for modern manufacturing networks. At first glance these topics seem distinct, one about leadership and the other about security, but the deeper I reflected the more I realized they are converging into a single, unavoidable truth. The supply chain leader of the future will manage a blended workforce of people, digital products, and AI agents and must ensure that this hybrid system continues to operate even under cyber stress. This is not an incremental shift but a structural redesign of the role itself.</p><p>For decades supply chain leadership was defined by managing people, overseeing processes, and delivering against KPIs. Success was measured in service levels, cost efficiency, and operational stability, and visibility through dashboards and control towers was seen as the major technological leap forward. But visibility is no longer the competitive advantage. We are moving from linear supply chains to adaptive networks powered by data streams, decision platforms, and increasingly autonomous systems. As a result leadership is no longer just about running operations but about orchestrating intelligence across humans and machines. That is why I believe the next generation of supply chain leaders will be half people managers and half product managers governing a digital workforce that includes AI agents.</p><p>Automation does not eliminate the human role but elevates it. Planners, schedulers, and plant operators are no longer just executing tasks but supervising systems, interpreting outcomes, and handling exceptions that automation cannot resolve. Leadership therefore becomes less about command and control and more about orchestration, aligning humans with algorithmic decisions and building trust in system driven outcomes. The real challenge is cultural rather than technical. Leaders must help teams understand why an optimization engine made a decision, coach people to collaborate with AI instead of competing with it, and align IT, operations, finance, and planning around a shared operating logic. In this new environment people remain the most complex part of the system because technology evolves quickly while trust does not.</p><p>The second half of the leadership transformation is less obvious but far more disruptive. Supply chain leaders are no longer just users of systems but owners of internal digital products such as planning platforms, allocation engines, transportation optimizers, and inventory decision systems. These are not background IT tools but operational capabilities that determine how the network behaves in real time. This requires a product mindset where leaders ask what decision should be automated next, what tradeoff logic should be encoded, what feature unlocks the next efficiency gain, and what data is missing to improve decision quality. The focus shifts from firefighting problems to capability building and from reacting to every disruption to designing systems that anticipate and adapt automatically. This is the essence of moving from reactive operations to outcome engineering.</p><p>The third dimension and the one most organizations underestimate is the rise of AI agents as part of the operational workforce. These systems will increasingly replan shipments dynamically, reallocate inventory across nodes, adjust production schedules in real time, and detect anomalies that trigger corrective actions. At that point leadership is no longer supervising only people but governing how autonomous decision systems behave. This becomes a system governance challenge where leaders define where automation has decision rights, where humans override, and what level of risk tolerance is acceptable. The future team will include humans, software products, and AI agents and all of them will require direction, priorities, and performance metrics. The competitive advantage therefore shifts from who has the most labor or capacity to who has the best coordination between humans and systems.</p><p>This shift also explains why decision velocity becomes the new competitive edge. Most organizations today can see what is happening in their supply chains but far fewer can respond quickly in a way that balances cost, service, capacity, and risk simultaneously. Speed without logic creates chaos while logic without speed creates paralysis. Leaders must ensure that systems are constraint aware and encode real operational tradeoffs rather than simplistic optimization targets. One of the most important realizations that surfaced during the leadership panel was that decisions cannot be automated if the business has not agreed on how decisions should be made. Questions about whether to protect margin or service during a shortage, which customers should take priority, and how much volatility can be tolerated are leadership questions rather than technology questions. Only once they are answered can they be translated into system logic that automation can execute consistently at scale.</p><p>This perspective naturally connects to the discussion on cyber resilience, which I approached through the lens of manufacturing reality. In high speed, operational technology heavy environments cybersecurity is not just about protecting data but about maintaining manufacturing continuity. In IT downtime is inconvenient but in operational technology downtime can be existential. Cyber resilience therefore means the ability to continue operating safely under attack, maintaining throughput, protecting product integrity, and ensuring worker safety even if certain systems are degraded or isolated. It requires designing supply chains that can gracefully degrade rather than catastrophically fail. Leaders must ask whether plants can operate in local mode if central systems go down, whether scheduling can fall back to rule based logic if optimization engines are offline, and whether quality checks can continue without full system connectivity.</p><p>These questions shift cybersecurity from a purely technical discipline to an operational one. Security must be embedded into system design rather than bolted on afterward. When done correctly resilience becomes a competitive advantage instead of a compliance exercise. The intersection between leadership transformation and cyber resilience becomes clear because both are ultimately about governing complex adaptive systems. A leader who manages people but does not understand systems will stall while a leader who understands technology but cannot bring people along will fail. The future belongs to hybrids who can orchestrate both dimensions simultaneously.</p><p>Cyber incidents will increasingly test this hybrid leadership model. When systems are disrupted the leader must decide which automated decisions remain trustworthy, which require human intervention, and how to maintain controlled continuity. This is not simply incident response but real time governance of a blended workforce operating under stress. In that sense cyber resilience is not separate from leadership but a proving ground for it. It reveals whether the organization has clearly defined decision rights, encoded tradeoff logic into its systems, and built trust between people and automation.</p><p>The role of the supply chain leader is therefore evolving from operator to architect and from supervisor to orchestrator. They will manage people, digital products, and AI agents as one integrated ecosystem, ensuring that decisions are made quickly, consistently, and safely even during disruption. The organizations that win will not simply have better tools or more data but leaders who understand tradeoff logic, can define clear decision rights, and can govern an adaptive network with both speed and discipline. Ultimately supply chain leadership is moving from reporting on what happened to engineering how outcomes are produced, and the future leader will be a conductor of a hybrid workforce operating at the speed of decision.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=55ade75d5a77" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Toll Collectors of Cognition]]></title>
            <link>https://medium.com/@nkjhaa/toll-collectors-of-cognition-526336470ec8?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/526336470ec8</guid>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Mon, 09 Feb 2026 08:13:41 GMT</pubDate>
            <atom:updated>2026-02-09T08:13:41.025Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6flNsfdMzciwyJSr_R1NZA.png" /><figcaption>Paying toll on passing a bridge. From a painted window in the Tournai Cathedral (15th century).</figcaption></figure><p>In medieval Europe, merchants crossing a stone bridge into a walled city would often stop at a small gatehouse. The guard did not grow the grain, weave the cloth, or forge the tools being carried across the bridge. But he collected a fee from every wagon that passed. The bridge owner understood something timeless: control the infrastructure, and you tax the commerce flowing through it.</p><p>Artificial intelligence is building the modern version of that bridge.</p><p>The companies laying the foundations are <strong>Amazon</strong>, <strong>Microsoft</strong>, <strong>Alphabet</strong> and <strong>Meta</strong>. Once celebrated as asset-light software champions, they are rapidly transforming into something closer to digital infrastructure barons. In 2026, these four are projected to spend roughly <strong>$635bn–$700bn</strong> in capital expenditures, the bulk aimed at AI infrastructure: data centers, GPU clusters, networking equipment, cooling systems and power capacity. That figure approaches <strong>2 per cent of US GDP</strong>, a scale more reminiscent of national infrastructure drives than corporate budgets.</p><h3>From code to concrete</h3><p>Before the AI boom, big tech’s financial identity was built on scalable software and rising margins. Capital spending funded growth, but it rarely defined the story. Free cash flow comfortably exceeded reinvestment needs, and returns on capital were buoyed by digital leverage.</p><p>AI inverts that model. Capacity must now be built before demand is fully visible. Hardware depreciates quickly, energy and cooling costs are real, and refresh cycles are measured in years. In several cases, annual capital expenditures are on track to rival or exceed net income; a characteristic more typical of utilities or telecom networks than of software platforms.</p><p>This shift introduces infrastructure economics into the heart of Big Tech. Depreciation rises, asset intensity climbs and returns hinge more on utilisation than on pure software scaling.</p><h3>The margin paradox</h3><p>At first glance, that sounds like a recipe for margin compression. And in the near term, it often is. The infrastructure layer drags on gross margins through higher depreciation, energy costs and the risk of overcapacity.</p><p>But AI is not only a cost center. It is also a productivity engine. Across large enterprises — including the hyperscalers themselves ; AI promises to automate coding, customer support, sales targeting, forecasting and internal processes. Even modest efficiency gains across vast operating expense bases can lift operating margins meaningfully.</p><p>The result is a structural tension: lower gross margins from heavy infrastructure, offset by potential operating-margin gains from AI-driven efficiency. In the early phase of the buildout, spending leads and monetization lags. Over time, as utilization improves and AI tools scale, the efficiency flywheel can begin to dominate.</p><h3>Digital utilities — but not quite</h3><p>The comparison increasingly heard on Wall Street is to utilities such as <strong>Edison International</strong>, whose subsidiaries provide essential services with high capital intensity and ongoing maintenance needs. The analogy is useful but incomplete.</p><p>Like utilities, hyperscalers are investing in foundational networks with long payback periods and continuous reinvestment. But unlike regulated power companies, they operate in global markets with pricing flexibility, rapid innovation cycles and powerful software and platform layers layered on top of their infrastructure.</p><p>In effect, these firms are becoming hybrid businesses. One layer resembles a digital utility: capital-heavy, depreciation-rich, essential but structurally lower-return. Another remains a software and platform engine, capable of higher-margin growth. Investors are being asked to value both under one roof.</p><h3>Lessons from the past</h3><p>History offers a warning for companies racing to “upload” their institutional knowledge into AI systems while reducing headcount. When railroads and later telecom networks became essential conduits of commerce, firms that depended on them without maintaining alternatives or bargaining leverage often found themselves squeezed on pricing and service terms. The same risk applies to AI. To avoid being caught by rising “cognition tolls,” companies should diversify AI providers, retain internal expertise that understands workflows beyond prompts, negotiate long-term pricing structures, and treat AI capability as a strategic asset rather than a pure cost-cutting tool. Efficiency gains are real, but surrendering too much control over how intelligence is delivered can shift long-term economics away from the enterprise and toward the platform owner.</p><h3>Who benefits — and who pays</h3><p>The consequences extend far beyond Big Tech. AI is a productivity technology. Companies that never build a data center can still use AI to automate back offices, improve forecasting and streamline operations, potentially lifting margins across industries.</p><p>Yet those gains are not entirely retained. Firms that do not own infrastructure will “rent intelligence” via cloud and AI services. As electricity once became a line item on factory ledgers, AI compute is becoming a new operating expense. The hyperscalers, as providers of that compute, are positioned to capture a share of the efficiency gains flowing through the economy.</p><p>In this sense, they resemble the bridge owner from centuries past. They do not participate in every transaction, but they stand at a strategic chokepoint. As cognition itself becomes metered, they are poised to collect a toll.</p><h3>A new era of capital and control</h3><p>For investors, the narrative shift is profound. Big Tech is no longer purely a story of asset-light disruption. It is increasingly about capital allocation, utilization rates and the economics of operating vast physical networks — even as software innovation continues at pace.</p><p>The AI race is often framed as a contest over models and algorithms. It is also a contest over who owns the bridges, the power lines and the server halls of the digital age.</p><p>And history suggests that those who control the crossings tend to do rather well.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=526336470ec8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Elephants and Hummingbirds: Heavy Industry, Hyperactive Markets]]></title>
            <link>https://medium.com/@nkjhaa/elephants-and-hummingbirds-heavy-industry-hyperactive-markets-4bba28de6c24?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/4bba28de6c24</guid>
            <category><![CDATA[manufacturing]]></category>
            <category><![CDATA[supply-chain]]></category>
            <category><![CDATA[futures-trading]]></category>
            <category><![CDATA[silver]]></category>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Mon, 02 Feb 2026 15:45:59 GMT</pubDate>
            <atom:updated>2026-02-02T15:45:59.419Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Why silver’s futures crash says more about financial flows than factory supply chains.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hbB3yBRknc0N1PU4cVwjMQ.jpeg" /><figcaption>The Market Runs Like a Hare. The Supply Chain Moves Like a Snail</figcaption></figure><p>A sharp drop in silver futures has been read in financial markets as a familiar story of cooling demand and fading inflation hedges. Yet on factory floors, from photovoltaic cell lines to electronics assembly, procurement teams describe something very different: tight lead times, limited flexibility in material form, and a scramble to secure reliable supply. The divergence exposes a structural reality that operations leaders understand well but markets often miss: the quoted price of a commodity can move on financial plumbing while the physical supply chain barely changes.</p><p>Silver’s headline price is largely discovered on exchanges such as <strong>COMEX</strong>, where contracts represent claims on metal rather than immediate industrial feedstock. When liquidity tightens, margins rise or investors reduce risk, funds cut exposure. Selling paper ounces can drive prices down rapidly, even if no new mine has started producing and no major end market has suddenly disappeared. A futures sell off is a balance sheet event before it is a geology event.</p><p>Manufacturing, however, runs on a different clock. The industrial silver that goes into solar cells, semiconductors, connectors and medical devices travels through a multi stage chain: mine concentrate, smelting, refining, specialty processing, then fabrication into powders, pastes, wires or coatings. Each step has its own capacity limits, quality controls and working capital needs. A lower futures price does not instantly expand refining throughput, increase the availability of high purity silver powder, or shorten qualification cycles at a solar paste producer.</p><p>The mismatch begins at the source. Much of the world’s silver output is a by product of mining for <strong>Copper</strong>, <strong>Lead</strong> and <strong>Zinc</strong>. Production decisions are therefore driven primarily by base metal economics, not by the silver price. Even a sustained rally in silver would not quickly trigger new dedicated supply. A decline certainly does not flood the market with additional metal. From an operations perspective, this is inelastic supply. Volumes respond slowly and indirectly to price signals.</p><p>Downstream, form matters as much as tonnage. Exchange registered bars or ETF holdings are not the same as fabrication ready inputs. Solar manufacturers need silver paste with specific particle size distributions and conductive properties. Electronics producers require powders or plated components that meet stringent purity and performance standards. Converting bulk bars into these forms involves specialized processors operating near capacity, often with long customer qualification processes. Inventory sitting in a vault is not interchangeable with qualified material on a production line.</p><p>Demand, meanwhile, is anchored in structural trends. Solar capacity additions continue globally, and while manufacturers have reduced the silver used per cell, overall installations keep total consumption high. Electrification of vehicles and expansion of electronics in industrial equipment add further draw. These are not speculative flows that vanish with a change in investor sentiment. They are embedded in capital expenditure plans and product roadmaps. For planners, the issue is not whether silver will be used, but whether it will arrive in the right form, at the right time, with acceptable credit terms.</p><p>This is where the financial and physical disconnect creates operational risk. Many industrial contracts reference exchange prices as benchmarks. When futures fall sharply, buyers may assume relief is imminent. But if midstream processors face high energy costs, tight labor markets or limited access to working capital, they may not pass through the full decline, or may ration supply instead. The result can be the worst of both worlds for manufacturers: volatile reference prices combined with persistent lead time pressure.</p><p>Logistics and finance amplify the gap. Specialty silver products are high value and often move through complex cross border routes. Trade credit, insurance and freight capacity all influence effective supply. A period of financial stress that drives futures lower can simultaneously constrain the very credit lines and risk appetites that enable physical trade. In such cases, the financial signal of abundance coincides with tighter physical flows.</p><p>The divergence in silver is not an isolated anomaly. It is a pattern that increasingly defines how modern economies function. Financial markets move at the speed of conviction, while industrial systems move at the speed of constraint.</p><p>A similar dynamic is emerging in artificial intelligence. Equity markets treat AI as a financial asset class, repricing companies within days based on earnings calls, data center spending plans or shifts in investor positioning. Stocks surge or fall on liquidity, expectations and options flows. This is a hummingbird market, darting on the slightest change in sentiment.</p><p>The deployment of AI inside organizations, however, looks far more like the elephant. It advances through process redesign, data integration, workforce adaptation, governance reviews and long testing cycles. Operational systems cannot be re architected at the pace of equity speculation. Enterprises must reconcile legacy infrastructure, regulatory exposure, cyber security risk and change management before AI becomes embedded in everyday decision making.</p><p>Just as silver’s futures price can fall without easing the bottlenecks in refining or specialty fabrication, AI valuations can swing wildly without altering the practical realities of implementation. The technology’s economic impact depends not on model breakthroughs alone, but on whether companies can translate capability into repeatable, scaled processes. This is a journey measured in quarters and years, not trading sessions.</p><p>History suggests this slower path is not a sign of failure but of absorption. The internet, cloud computing and enterprise software followed similar arcs. Markets oscillated long before workflows quietly adapted and the technologies became unremarkable parts of business infrastructure. Financial cycles ran ahead. Operational adoption caught up.</p><p>In both silver and AI, the core tension is the same. Markets behave like hummingbirds, reacting instantly to signals and sentiment. Industrial systems behave like elephants, moving deliberately through terrain defined by physical, organizational and economic limits. Confusing one speed for the other can lead to costly misreading, whether of metal supply or technological progress.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4bba28de6c24" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The AI Investment Paradox: Why Corporate Billions Aren’t Delivering Transformation]]></title>
            <link>https://medium.com/@nkjhaa/the-ai-investment-paradox-why-corporate-billions-arent-delivering-transformation-ddb25b5f2c76?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/ddb25b5f2c76</guid>
            <category><![CDATA[roi]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[capex]]></category>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Thu, 29 Jan 2026 00:44:51 GMT</pubDate>
            <atom:updated>2026-01-29T00:44:51.061Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*mDqlxxmjHIA3djShuSyT4A.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*HU-EX8vTePGLt_I3XEs6Kg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*e7KuvT6gN7M2xaCnRUi9Vg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*g9fK_SJW8JsjJcb-YVgxAg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*nu-YEEbh8mGQTHzG45zLsA.png" /><figcaption><em>New data reveals a striking disconnect between enthusiasm for artificial intelligence and realistic expectations for its impact</em></figcaption></figure><p>Corporate America is in the midst of an AI spending spree, yet a troubling pattern is emerging from the data: the companies writing the largest cheques harbor the most modest expectations for returns.</p><p>According to Duke University’s recent CFO Survey, conducted between November and December 2025, while 77 per cent of firms report investing in AI — rising to nearly 90 per cent among large enterprises — the overwhelming majority expect marginal improvements at best. Most anticipate no change whatsoever in key performance metrics including labour productivity and customer satisfaction.</p><p>This disconnect raises a provocative question: are we witnessing the diffusion of a sustaining innovation rather than the revolutionary transformation promised by technology evangelists?</p><h3>A Tale of Two Strategies</h3><p>The investment landscape reveals a stark bifurcation. Nearly half of firms report zero AI spending — a figure that jumps to 50 per cent among smaller enterprises. Yet among those who are investing, commitment levels vary dramatically, from modest five-figure experiments to investments exceeding $5m.</p><p>This creates what economists might call a ‘missing middle.’ Companies are either betting heavily on AI or sitting out entirely, with few hedging their positions. The pattern suggests that AI adoption requires threshold investments in infrastructure, talent, and organizational change that make tentative approaches unviable.</p><p>Large firms, with their deeper pockets and existing technical capabilities, dominate AI deployment. But whether this represents a durable competitive advantage or an expensive mistake remains an open question.</p><h3>Modest Ambitions</h3><p>The survey’s findings on expected returns are perhaps most revealing. When asked about AI’s anticipated impact on labor productivity, roughly two-thirds of firms investing in the technology expect either no change or are simply unsure. Among those projecting improvements, the vast majority anticipate gains of just 1 to 5 per cent.</p><p>Similar patterns emerge across other metrics. Customer satisfaction, decision-making speed, and time spent on high-value tasks all show modal responses clustering around ‘no change’ — even among firms actively deploying AI systems.</p><p>Perhaps most strikingly, expectations for total employment and operating costs suggest AI is being deployed primarily to augment rather than replace human workers. The majority of firms expect employment levels to remain stable, whilst operating costs excluding labor show similar inertia.</p><p>These are not the expectations of executives anticipating revolutionary change. They are the expectations of managers implementing incremental improvements to existing processes.</p><h3>Productivity Paradox Redux?</h3><p>The pattern bears an uncanny resemblance to the ‘productivity paradox’ that bedeviled early information technology investments. Throughout the 1970s and 1980s, companies poured billions into computer systems yet struggled to demonstrate corresponding productivity gains. Nobel laureate Robert Solow famously quipped in 1987: ‘You can see the computer age everywhere but in productivity statistics.’</p><p>The paradox eventually resolved — but only after organizations learned to redesign workflows, retrain workers, and reimagine business processes around new capabilities. The technology itself wasn’t enough; complementary innovations in management and organization proved essential.</p><p>Today’s AI investments may face similar challenges. Current applications largely automate discrete tasks within existing workflows rather than enabling fundamentally new ways of working. The 1 to 5 percent productivity gains executives anticipate may simply reflect the limits of deploying AI within organizational structures designed for an earlier era.</p><h3>What This Means for Strategy</h3><p>The modest expectations embedded in the survey data suggest a more mature understanding of AI’s near-term business value than much public discourse would suggest. Executives may have moved past the hype cycle’s ‘peak of inflated expectations’ and settled into a more realistic assessment.</p><p>This creates several strategic implications. First, for firms still sitting on the sidelines, the competitive imperative may be less urgent than feared. If early adopters are achieving only marginal improvements, fast followers may be able to learn from their mistakes whilst avoiding expensive dead ends.</p><p>Second, the firms investing heavily in AI may need to meet their organizational expectations. If the goal is marginal efficiency gains rather than transformation, AI projects should be evaluated using standard cost-benefit frameworks rather than treated as strategic imperatives requiring patient capital.</p><p>Third, investors and boards should demand clearer justification for major AI investments. The uncertainty evident in the survey — with roughly a quarter of respondents marking ‘not sure’ across multiple metrics — suggests many firms are deploying the technology without clear hypotheses about expected benefits.</p><p>Finally, the relative absence of firms expecting major productivity leaps (improvements exceeding 5 per cent) lead to a possible opportunity. Companies that can successfully redesign workflows and business models around AI capabilities — rather than simply automating existing tasks — may capture disproportionate value.</p><h3>Unanswered Questions</h3><p>The survey raises as many questions as it answers. Why is uncertainty about AI’s impact so pervasive, even among firms actively deploying it? Shouldn’t organizations have clearer hypotheses before committing millions to new technology?</p><p>The bimodal distribution of investment firms, either spending heavily or not at all, suggests different strategic logics at work. Are non-adopters making a rational calculation that AI doesn’t fit their business models? Or are they falling behind in a race where the advantages to early movers’ compound overtime?</p><p>The stability in expected employment levels is particularly intriguing. If AI augments rather than replace workers, this will represent a departure from previous waves of automation that typically reduced headcount in affected functions. But it would also raise questions about where productivity gains materialize, are firms expected to do more with the same workforce, or simply to do the same work more efficiently.</p><p>Most fundamentally, the gap between investment enthusiasm and modest expectations suggests many firms view AI as a strategic necessity regardless of near-term returns. This ‘build it and they will come’ approach may prove prescient if AI capabilities compound over time. But it could equally result in billions spent on infrastructure that never delivers promised returns.</p><h3>Beyond the Hype</h3><p>The data paints a picture notably at odds with prevailing narratives about AI transformation. Rather than revolutionary change, most executives anticipate incremental improvements. Rather than workforce displacement, they expect stable employment. Rather than certainty about AI’s value, they express widespread uncertainty.</p><p>This disconnect between rhetoric and expectation may prove the most important finding. If the current wave of AI investment represents sustaining innovation rather than disruption — improving existing processes rather than enabling fundamentally new ones — then competitive dynamics will evolve very differently than many anticipate.</p><p>The productivity paradox of the 1980s eventually resolved as organizations learned to harness information technology’s potential. AI may follow a similar trajectory, with today’s modest expectations giving way to more substantial gains as complementary innovations in management and organization emerge.</p><p>But that process took decades for IT and required fundamental rethinking of how work gets done. For now, the survey suggests corporate executives are taking a more measured view of AI’s potential than much public discussion would indicate. Whether this reflects wisdom or a failure of imagination remains to be seen.</p><p>What is clear is that billions are being invested in technology whose benefits remain stubbornly difficult to quantify. In the coming years, the gap between AI investment and AI impact will be one of the most important dynamics in corporate strategy to watch.</p><p><em>Analysis based on Duke University CFO Survey data, Q4 2025 (N=547)</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ddb25b5f2c76" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Great Inversion: When Trades Jobs Supersede Knowledge Workers]]></title>
            <link>https://medium.com/@nkjhaa/the-great-inversion-when-trades-jobs-supersede-knowledge-workers-16fd2827160c?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/16fd2827160c</guid>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Sun, 11 Jan 2026 21:02:51 GMT</pubDate>
            <atom:updated>2026-01-11T21:02:51.264Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5unZuljsCGaFTqFoounITg.png" /></figure><h3>The Great Inversion: When Trades Jobs Supersede Knowledge Workers</h3><p>Ford just revealed something that should make every parent of a college-bound student pause: a master technician at a dealership in Ohio earned $160,000 last year. Meanwhile, Ford CEO Jim Farley is struggling to fill 5,000 technician roles that pay over $120,000.</p><p>Let me be clear: Ted Hummel’s $160K salary represents the exceptional peak — the top 1–2% of his profession. But the broader story here reveals a fundamental shift in how we should think about education, career paths, and the future of work.</p><p>For twenty years, I’ve worked in operations around ships and high-speed bottling lines alongside mechanics, machinists, and technicians who could diagnose and fix a high scrap rate on a blowmolder by watching the transfer arms of a machine rotating at 45 RPM, trace an internal combustion engine’s misfire to a single fouled injector, and troubleshoot problems that would leave most engineers scratching their heads. The best mechanics I’ve known operate like great surgeons — everyone starts as a doctor, but very few have the hands, the instincts, and the relentless precision to become a surgeon.</p><p><strong>Let’s look at the honest numbers:</strong></p><p>Entry-level automotive technicians start around $47,000 annually, while entry-level college graduates average $68,680 (though this varies wildly by major — some start closer to $50,000). At 10 years of experience, auto technicians average around $80,000, comparable to the median for bachelor’s degree holders across their careers.</p><p>The real story isn’t that trades universally outpay degrees. It’s more nuanced and more interesting.</p><p><strong>What’s driving this structural shift?</strong></p><p>Three forces have collided to expose a broken educational model:</p><p><strong>1. The Monolithic Education Myth</strong> We treat education as binary and permanent: either you’re college-bound or trade-school-bound, and that decision at 18 defines your ceiling forever. This is obsolete thinking. We pushed an entire generation toward four-year degrees as the only path to success, creating massive student debt (averaging $39,375 in federal loans) while treating skilled trades as dead ends rather than foundations.</p><p><strong>2. The Digital-Physical Disconnect</strong> Today’s workforce grew up mastering screens but not systems. They have unprecedented digital literacy but minimal mechanical intuition. As we march toward Physical AI — robots in warehouses, autonomous vehicles, smart factories — we’re discovering that the future doesn’t just need programmers. It desperately needs people who understand both the digital controls AND the physical machinery those controls operate.</p><p><strong>3. The AI Reckoning Hits Knowledge Work First</strong> While everyone worried about AI replacing blue-collar workers, the opposite happened. Entry-level hiring in AI-exposed white-collar jobs has dropped 13%. Big Tech companies slashed new graduate hiring by 25% in 2024. Two-thirds of global enterprises plan to cut entry-level hiring due to AI adoption. Anthropic’s CEO predicts AI could eliminate half of all entry-level white-collar jobs within five years.</p><p>Meanwhile, you cannot automate away someone who can keep a high-speed bottling line running, diagnose why a robotic cell is throwing errors, or troubleshoot a ship’s propulsion system in real-time.</p><p><strong>This isn’t just about cars.</strong></p><p>Every sector I’ve touched — manufacturing, aerospace, heavy equipment, industrial facilities — faces the same shortage. Master electricians, millwrights, HVAC technicians, industrial maintenance specialists: these roles can’t be offshored, can’t be automated, and can’t be filled fast enough.</p><p>Some argue that automation is eliminating manufacturing jobs. They’re half right. Automation eliminates repetitive operator roles while creating acute demand for technicians who can maintain, troubleshoot, and optimize increasingly complex systems. A modern high-speed bottling line doesn’t need 20 operators — it needs 3 exceptional technicians who understand pneumatics, PLCs, servo systems, and can diagnose issues on equipment moving at production speeds. The same holds true for automated welding cells, robotic assembly lines, and smart factories.</p><p>Meanwhile, we’re watching organizations restructure away entire layers of middle management and entry-level knowledge work while their production lines sit silent because they can’t find qualified technicians to keep the automation running.</p><p><strong>The real opportunity: A hybrid path forward</strong></p><p>Here’s what we got wrong: education shouldn’t be monolithic or “one and done.”</p><p>Instead of forcing an 18-year-old to choose between a $50K trade certification or $150K in college debt, imagine a system where:</p><p>→ <strong>You start by gaining real skills.</strong> Spend 1–2 years learning to work with your hands, understanding physical systems, earning while you learn with minimal debt.</p><p>→ <strong>Then you ladder up based on intellect, motivation, and ambition.</strong> The technician who shows aptitude can pursue engineering coursework. The machinist interested in optimization can study industrial engineering. The mechanic fascinated by diagnostics can move into systems design.</p><p>→ <strong>You enter the knowledge economy with context.</strong> Instead of fresh business graduates making supply chain decisions with no feel for a factory floor, you have operations managers who’ve actually run the machines. Instead of engineers designing products they’ve never built, you have designers who’ve held the tools.</p><p>This creates professionals who have a visceral connection to the ecosystem beyond their educational credentials. The maintenance technician who becomes a mechanical engineer doesn’t just know the theory — they remember what it feels like when a bearing is about to fail. The machinist who becomes a manufacturing engineer designs fixtures that actual humans can use.</p><p><strong>As Physical AI arrives, this matters more, not less.</strong></p><p>The future isn’t purely digital. It’s the integration of digital intelligence with physical systems. The most valuable professionals won’t be pure coders or pure mechanics — they’ll be people who can bridge both worlds. Who can troubleshoot when the AI-driven predictive maintenance system flags a false positive. Who understand both the machine learning model AND the machine it’s monitoring.</p><p><strong>What needs to change immediately:</strong></p><ul><li>Stop treating skilled trades as terminal careers and start treating them as foundations</li><li>Create clear pathways for technicians to ladder into engineering, operations management, and systems design</li><li>Make technical apprenticeships pay living wages and lead somewhere</li><li>Restructure higher education to be modular and stackable, not monolithic</li><li>Value hands-on experience as much as we value theoretical knowledge</li><li>Build bridge programs that allow working technicians to earn degrees part-time without crushing debt</li></ul><p>Ted Hummel’s $160K story isn’t a fantasy everyone will reach. But it represents a proof point: exceptional skill commands exceptional value. More importantly, it shows what happens when you commit to mastery in a field experiencing acute scarcity.</p><p>The real message isn’t “skip college and become a mechanic.” It’s “rethink the path entirely.”</p><p>Start with skills. Build competence. Earn while you learn. Then ladder up based on capability and ambition, not just credentials. Create a workforce that understands both atoms and bits, both systems and code, both the factory floor and the strategic plan.</p><p>The market is already signaling this shift. Ford can’t fill 5,000 positions at $120K+. Entry-level white-collar hiring is collapsing. Wages for young workers in high-growth trades are outpacing inflation.</p><p>The question is whether our educational institutions and cultural narratives will catch up to the reality the market is already showing us — or whether we’ll keep pushing students toward a model designed for an economy that no longer exists.</p><p>What are you seeing in your industry? Are we ready to build education systems that create hybrid professionals instead of forcing false choices?</p><p>#SkillsGap #WorkforceDevelopment #Manufacturing #SkilledTrades #FutureOfWork #Operations #Education #PhysicalAI #Automation #CareerDevelopment</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=16fd2827160c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Frozen Food Parallel: What AI Automation’s Labor Shifts Really Mean]]></title>
            <link>https://medium.com/@nkjhaa/the-frozen-food-parallel-what-ai-automations-labor-shifts-really-mean-6afa2915218c?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/6afa2915218c</guid>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Fri, 09 Jan 2026 15:13:41 GMT</pubDate>
            <atom:updated>2026-01-09T15:13:41.112Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Nsmj6JcVT56x-9lwI5JkCQ.png" /><figcaption>Freezer to Table</figcaption></figure><p>The weakest year of job growth since 2010 has economists debating whether the U.S. labor market is stalling or simply transforming. History suggests the latter — and offers an unlikely analog in frozen peas.</p><p>The Bureau of Labor Statistics reported in December that the economy added just 50,000 jobs, well below forecasts. For all of 2024, payroll growth totaled 584,000, the slowest expansion outside pandemic years. Manufacturing shed positions, retail employment contracted, and hiring concentrated in food services and healthcare. The unemployment rate dipped to 4.4%, aided partly by declining labor-force participation.</p><p>The numbers have fueled concerns about economic momentum. But they may also signal something more complex: a fundamental reordering of how work gets done, reminiscent of an earlier technological shift that reshaped American life without triggering the alarm bells it perhaps deserved.</p><h3>When Freezing Changed Everything</h3><p>The rise of frozen food between the 1930s and 1950s offers an instructive precedent. Before widespread freezing technology, food supply chains were labor-intensive, localized, and seasonal. Spoilage was common. Distribution required constant human intervention. Household food preparation consumed hours daily, work that fell disproportionately to women and rarely appeared in economic statistics.</p><p>Freezing technology didn’t eliminate food-related employment. It redistributed it. Labor shifted from local handling to centralized processing facilities and cold-storage logistics networks. As frozen food became commonplace, household food expenditure fell as a percentage of income over subsequent decades. Time spent on meal preparation declined sharply, enabling greater workforce participation elsewhere in the economy. What had been a premium convenience became standard practice.</p><p>The transition was largely invisible to policymakers at the time. Yet it fundamentally altered consumer behavior, household economics, and labor allocation across the economy.</p><h3>A Similar Pattern Emerges</h3><p>Today’s labor market exhibits characteristics economists describe as a “low-hire, low-fire” equilibrium. Companies are reluctant to expand headcount aggressively but equally hesitant to make deep cuts. Productivity pressures are mounting. Sectors tied to physical goods production have stopped adding workers — not because demand has collapsed, but because labor is no longer the primary constraint on output.</p><p>Artificial intelligence is beginning to play a comparable role to what freezing technology once did. AI systems centralize planning and coordination tasks that previously required multiple human decision-makers. They compress cycle times and reduce operational waste without necessarily reducing output. Like frozen food’s replacement of repetitive household tasks with industrial systems, AI automates cognitive work that was once manual and distributed.</p><p>The parallel extends to where job growth now concentrates. Food services and healthcare — sectors where physical presence and human judgment remain difficult to automate — continue hiring. Manufacturing employment falls even as industrial output holds relatively steady. The pattern suggests labor is being reallocated, not simply eliminated.</p><h3>The Migration of Opportunity</h3><p>Economic history offers a counterintuitive lesson: when labor intensity in one area declines, opportunity typically doesn’t vanish — it moves. Frozen food reduced time spent on food preparation, which enabled increased participation in other parts of the economy while lowering real costs for households. AI automation appears to be following a similar trajectory, reducing labor devoted to planning and coordination while shifting work toward system design, exception handling, and judgment calls that machines handle poorly.</p><p>This doesn’t guarantee smooth adjustment. Workers displaced from shrinking sectors face real costs. Retraining takes time and resources. Geographic mismatches persist. But the aggregate pattern points toward evolution rather than collapse.</p><h3>What Markets Miss</h3><p>Some investors worry that equity valuations assume near-perfect economic conditions while job growth slows. For corporate operators, however, the calculation is different. Waiting for traditional labor-market dynamics to reassert themselves may mean ceding ground to competitors already redesigning operations around greater automation.</p><p>The Federal Reserve’s next moves remain uncertain. Market volatility is likely. But the operational reality facing businesses is already clear: labor has become less elastic as a scaling mechanism, automation provides greater leverage, and returns increasingly depend on intelligence rather than headcount.</p><h3>The Longer View</h3><p>Frozen food didn’t diminish the quality of American diets. It made time less expensive. AI automation isn’t making work disappear. It’s making certain decisions less expensive. In both cases, consumers adapt, habits evolve, new roles emerge, and old assumptions prove inadequate.</p><p>The latest employment data may not signal impending collapse. It may instead mark the early stages of a system being fundamentally restructured — a transformation that only becomes legible in retrospect.</p><p>Much like frozen peas did seventy years ago.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6afa2915218c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Tale of Two Sams]]></title>
            <link>https://medium.com/@nkjhaa/the-tale-of-two-sams-ab7edd41b0a9?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/ab7edd41b0a9</guid>
            <category><![CDATA[jobs]]></category>
            <category><![CDATA[universal-basic-income]]></category>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Sat, 11 Oct 2025 02:47:23 GMT</pubDate>
            <atom:updated>2025-10-11T02:47:23.003Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PBdwAuaCIKW51BtOm9xvfQ.png" /><figcaption>The Epic Divergence</figcaption></figure><p><em>One sits on top of the AI mountain. The other worries about the valley below.</em></p><p>There’s Sam Altman, the architect of a new digital order — standing at the peak of the AI revolution, building models that can reason, write, and soon, run companies. And then there’s Uncle Sam, staring down a familiar American problem: what happens when technology outpaces labor faster than policy can adapt?</p><p>Look at the chart above. For the first time in decades, the stock market and the real economy seem to have stopped talking to each other. Job openings (blue) are falling fast, while the S&amp;P 500 (orange) keeps climbing. The labor market is cooling; capital markets are on fire.</p><p><strong>The Great Divergence</strong></p><p>In previous cycles, this disconnect didn’t exist. Rising unemployment or shrinking job openings meant lower corporate earnings — and markets would follow suit. But something fundamental has changed. Today, both retail and institutional investors are piling into equities even as the number of open jobs drops, betting that the next wave of productivity won’t come from people — it’ll come from data centers.</p><p>Billions are being poured into hyperscale computing facilities, Nvidia clusters, and liquid-cooled racks of inference servers: the new “factories” of cognitive labor. These aren’t building cars or shoes; they’re building intelligence. And when investors see margins expand without hiring, they cheer. When policymakers see the same chart, they worry.</p><p>This divergence isn’t a statistical oddity — it’s a structural signal. Money is no longer chasing labor productivity; it’s chasing model productivity. In 2024 and 2025, corporate CAPEX reports revealed something striking: spending on AI infrastructure was up over 40%, while headcount growth in many large firms was flat or negative. Capital is finding its next frontier in silicon-based cognition, even as a growing pool of human potential sits under-utilized.</p><p><strong>The AI–UBI Feedback Loop</strong></p><p>That brings us back to our two Sams — because their stories are converging faster than most people realize.</p><p>Sam Altman dreams of artificial general intelligence and also happens to be one of the biggest proponents of Universal Basic Income. Uncle Sam may soon have no choice but to take that idea seriously. As AI systems begin to absorb white-collar tasks — drafting reports, managing operations, even writing code — the displacement won’t look like the factory layoffs of the past. It’ll be quieter, more diffuse, and harder to measure. But it will be real.</p><p>Every previous industrial revolution — from engines to electricity to software — eventually created more jobs than it destroyed. But AI may rewrite that rule. This time, the marginal cost of cognitive labor is heading toward zero. And when intelligence becomes abundant, human scarcity may become our new constraint. The question is not whether AI will replace jobs — it’s whether society can evolve fast enough to replace income, and meaning, for those whose work is automated.</p><p><strong>Closing the Gap</strong></p><p>Sam Altman’s vision and Uncle Sam’s responsibility are two ends of the same equation: how to fund abundance in a world where fewer people earn wages. One side sees AI models as assets. The other must see citizens as stakeholders.</p><p>If we get it right, this divergence will narrow — not because markets fall, but because prosperity broadens. If we get it wrong, we’ll have the world’s most powerful intelligence systems and no one left to buy what they produce.</p><p><strong>🧭 The Takeaway</strong></p><p>The chart isn’t just about jobs and stocks. It’s about the transition from a labor-driven economy to a compute-driven one. We’re watching two Sams — innovation and governance — write the next chapter of capitalism. And the question that will define this decade is simple:</p><p><em>Can the AI mountain lift the valley below?</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab7edd41b0a9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Great Operators Never Hide Behind Technology: Lessons from Chipotle and Sweetgreen]]></title>
            <link>https://medium.com/@nkjhaa/why-great-operators-never-hide-behind-technology-lessons-from-chipotle-and-sweetgreen-9678e6341def?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/9678e6341def</guid>
            <category><![CDATA[digital-transformation]]></category>
            <category><![CDATA[health-foods]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[operations-management]]></category>
            <category><![CDATA[qsr]]></category>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Sun, 28 Sep 2025 01:25:41 GMT</pubDate>
            <atom:updated>2025-09-28T02:26:53.158Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*J3j9YdTVHyIlfRShvgU0zQ.png" /></figure><h3>Technology’s Double-Edged Sword</h3><p>Technology has become the most powerful buzzword in business today. From artificial intelligence to robotics, executives are under constant pressure to frame their organizations as “tech-first.” Investors reward stories of disruption, analysts lean in when they hear automation or digital transformation, and boards demand innovation roadmaps heavy with buzzwords. But the reality is much simpler and far less glamorous: technology cannot mask weak operations, nor can it substitute for clarity of identity.</p><p>The greatest operators have always understood this. They know their product inside out, they understand their customer better than anyone else, and they execute with relentless discipline. For them, technology is not a costume or an identity — it is a tool, deployed carefully and precisely to amplify strengths that already exist.</p><p>The recent stories of Chipotle Mexican Grill and Sweetgreen reveal this truth vividly. Both companies entered the fast-casual space promising healthier, customizable meals for an urban, health-conscious consumer. Both leaned into freshness, sustainability, and the allure of better-for-you food. But their approaches to identity diverged. Chipotle doubled down on being a food company, one that happened to use technology to sharpen execution. Sweetgreen, by contrast, sought to rebrand itself as a technology company that happened to sell food. That difference in self-definition led to radically different outcomes.</p><h3>Two companies, Two destinies</h3><p>The contrast between Chipotle and Sweetgreen is visible in every performance metric. Chipotle’s market capitalization stands at roughly $53 billion. In June 2024, it executed a historic 50-for-1 stock split — one of the largest in NYSE history. Its financial results in 2025 underscore its resilience: $3.1 billion in revenue for the second quarter alone, up three percent year over year, with net income of $436 million. Over the trailing twelve months, Chipotle generated more than $11.6 billion in revenue, an increase of 8.6 percent year over year.</p><p>Sweetgreen tells a different story. Once hailed as the darling of health-conscious dining, its market capitalization has shrunk to around $1 billion. Shares have plummeted more than 80 percent from their IPO highs, leaving early investors with only $172 of value for every $1,000 initially invested. In the second quarter of 2025, Sweetgreen reported revenue of $185.6 million — just one-sixteenth the scale of Chipotle — alongside a net loss of $23 million. Even more troubling, same-store sales declined 7.6 percent that quarter, driven by a 10 percent drop in customer traffic.</p><p>💡 <strong>Chipotle generated $3.1B in Q2 2025 revenue vs. Sweetgreen’s $185M.</strong></p><h3>Chipotle: Technology as an Enabler</h3><p>Chipotle has always been clear about what it is: a food company. Its promise of “Food with Integrity” has anchored its strategy for decades. That clarity allows the company to make disciplined, pragmatic choices about technology.</p><p>When Chipotle invests in digital ordering, it does so to make the customer experience faster and easier. When it introduces new equipment — such as dual-sided planchas, three-pan rice cookers, or automated produce slicers — it is in service of consistency, throughput, and food quality. When it expands, as it plans to do by opening more than 300 new restaurants in 2025, it does so with operational discipline, ensuring that most new units include Chipotlanes, the digital drive-thrus that align seamlessly with its growing digital sales channel.</p><p>💰 <strong>Chipotle repurchased $435.9M of stock in Q2 2025, funded entirely by cash flow.</strong></p><p>Technology, for Chipotle, is never the main event. It is always a supporting act, an enabler of the company’s food-first identity. As leadership has often emphasized, Chipotle is not a tech company. It is a restaurant company that uses technology to serve better food, faster and more consistently. That clarity is visible in its financial strength, its profitability, and its enduring customer loyalty.</p><h3>Sweetgreen: Technology as Identity</h3><p>Sweetgreen took the opposite path. Its founders increasingly positioned the company not as a restaurant brand but as a “food platform.” It adopted the language and posture of Silicon Valley, branding itself as a disruptor not of menus but of operations. Its flagship innovation, the Infinite Kitchen, promised to revolutionize the dining experience through robotics and automation. By 2024, Sweetgreen operated a dozen of these kitchens, marketing them as the future of food service.</p><p>But the problem with making technology your identity is that it distracts from the fundamentals. Sweetgreen struggled to deliver consistent execution across its stores. By its own admission, only one-third of restaurants operated at or above standard, while two-thirds fell short. Efforts to diversify offerings — like the introduction of RippleFries — were abandoned because they “complicated delivery on the core.” Even its loyalty program backfired, creating headwinds that hurt same-store sales.</p><p>🥗 <strong>Sweetgreen’s traffic fell 10.1% in Q2 2025 — its steepest decline since going public.</strong></p><p>The result was a brand that looked innovative on paper but failed at the ground level where customers make their decisions: food quality, speed of service, and value. Heavy investments in automation and digital tools did little to change the fact that customers were walking away, leaving Sweetgreen with declining traffic, eroding margins, and shrinking investor confidence.</p><h3>What the Comparison Reveals</h3><p>The comparison between Chipotle and Sweetgreen illustrates a fundamental truth. Technology should enable, not define. Chipotle integrates digital tools and new equipment only when they enhance the dining experience or sharpen operational consistency. Sweetgreen made technology its story, a narrative that ultimately overshadowed its food.</p><p>Operational excellence is identity in action. Chipotle’s burritos taste the same whether you are in New York or New Mexico. That consistency builds trust. Sweetgreen, by contrast, struggled to execute at scale, with most stores failing to meet standards.</p><p>Customers buy food, not algorithms. Chipotle sells burritos and bowls. Sweetgreen sells salads. That is what customers care about. No diner shows up hoping for an “AI-powered ordering experience.” They want meals that are fresh, fast, and fairly priced.</p><p>Financial discipline is a mirror of strategic clarity. Chipotle generated so much free cash in the second quarter of 2025 that it could return nearly half a billion dollars to shareholders while still funding expansion. Sweetgreen, by contrast, was forced to cut staff just to preserve cash.</p><p>And finally, identity shapes perception. Analysts describe Chipotle’s brand as “never stronger.” Sweetgreen’s identity crisis has left investors skeptical about its ability to execute consistently at scale.</p><p>📊 <strong>Chipotle’s market cap: $53B. Sweetgreen’s market cap: $1B.</strong></p><h3>Lessons Beyond Restaurants</h3><p>The Chipotle–Sweetgreen contrast is not just about burritos and salads. It is a parable for every industry wrestling with the role of technology.</p><p>In manufacturing, digital twins and IoT sensors can transform operations, but they cannot replace process discipline and quality control. In logistics, robotics and automation are powerful, but reliability, safety, and on-time performance remain the real differentiators. In healthcare, AI diagnostics are promising, but patient trust and care quality are still the bedrock of value.</p><p>Across sectors, the same principle holds true: companies succeed when technology serves the operation, not when the operation is forced to serve the technology.</p><h3>The Seduction of Buzzwords</h3><p>It is easy to understand why companies fall into Sweetgreen’s trap. Tech narratives are seductive. They attract media attention, excite investors, and signal modernity and ambition. A company that describes itself as “tech-first” or “platform-driven” feels like it is riding the wave of the future.</p><p>But customers are not seduced by narratives. They care about outcomes. They judge companies on product quality, reliability, service consistency, and value for money. No amount of automation or AI gloss can disguise failure on those basics. And when customers leave, investors eventually follow.</p><p>📉 <strong>Sweetgreen IPO investors are down more than 80% — a $1,000 stake is now worth just $172.</strong></p><h3>Conclusion: The Operator’s Edge</h3><p>The story of Chipotle and Sweetgreen is not just a tale of two restaurant chains. It is a fundamental reminder about the nature of competitive advantage. Chipotle thrives because it never lost sight of what it is: a restaurant company committed to food quality and operational excellence. Sweetgreen struggles because it let technology become its identity, diverting focus from the fundamentals of food and service.</p><p>The lesson for leaders is timeless. Know your product better than anyone else. Know your customer better than anyone else. Execute with relentless discipline. And only then, once the core is solid, use technology to enhance, accelerate, and enable.</p><p>Technology is a force multiplier, but only when it is in the hands of operators who know who they are. The greatest companies never hide behind tools or buzzwords. They stay anchored in their craft, their product, and their customer. That is why they win — and why others, who confuse identity with innovation, eventually falter.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9678e6341def" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Sovereign AI: The Strategic Imperative for Legacy Manufacturers in the Age of Intelligent…]]></title>
            <link>https://medium.com/@nkjhaa/sovereign-ai-the-strategic-imperative-for-legacy-manufacturers-in-the-age-of-intelligent-9652aa516f90?source=rss-4555a45a3c1e------2</link>
            <guid isPermaLink="false">https://medium.com/p/9652aa516f90</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-ownership]]></category>
            <category><![CDATA[sovereign-ai]]></category>
            <category><![CDATA[intellectual-property]]></category>
            <category><![CDATA[data-sovereignty]]></category>
            <dc:creator><![CDATA[Niraj Jha ]]></dc:creator>
            <pubDate>Sat, 26 Jul 2025 19:38:18 GMT</pubDate>
            <atom:updated>2025-07-26T19:38:18.391Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Sovereign AI: The Strategic Imperative for Legacy Manufacturers in the Age of Intelligent Automation</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Icprxp6yKXtEsTSbsXYEJA.png" /><figcaption>Sovereign AI- The Strategic Imperative for Legacy Manufacturers</figcaption></figure><p>For legacy manufacturing companies with decades of market dominance, robust cash flow, and unparalleled expertise in their core domain, the digital future holds tremendous promise — but also significant risk. These companies are often rooted in a culture of operational excellence, continuous improvement, and deeply embedded institutional knowledge. In this context, quoting Jensen Huang and replacing “countries” with “companies” holds a lot of significance, “Sovereign AI for companies codifies your culture, your company’s intelligence, your common sense, your history — you own your own data,” is not just a philosophical stance. It is a strategic necessity.</p><p>When a manufacturing company with decades of success contemplates its AI roadmap, the temptation to turn to the hyperscalers (<em>large-scale cloud service providers that offers vast and flexible computing resources, storage, and networking capabilities, often on a global scale— Amazon, Google, Microsoft, and Oracle </em>) — is understandable. They offer compelling platforms, massive compute resources, and off-the-shelf generative AI capabilities. But doing so can mean exporting a company’s proprietary knowledge and internal know-how to a third party, effectively leasing back their own intelligence. This model is not only economically inefficient in the long term but potentially perilous to the company’s autonomy, IP security, and strategic advantage.</p><h3>The Real Threat of Hyperscaler Dependency</h3><p>For a legacy manufacturer, the core moat is often built on decades of trial-and-error learning, proprietary processes, custom machinery, supplier relationships, and workforce know-how. This accumulated intelligence is embedded in everything from production line workflows to safety protocols, from supply chain heuristics to customer service ethos. Turning over this crown jewel to a hyperscaler to build agentic AI solutions introduces several major risks:</p><h4>1. Loss of Proprietary Advantage</h4><p>By training AI models on company-specific data within hyperscaler platforms, organizations risk revealing their process efficiencies and proprietary workflows to the cloud provider. Even under strict legal protections, the data ingestion and model refinement processes may lead to leakage or replication of the company’s strategic advantage into generalized AI offerings.</p><h4>2. Strategic Vendor Lock-In</h4><p>Once core AI capabilities are built and operated within a hyperscaler ecosystem, switching costs become prohibitively high. The AI models, APIs, and data storage architectures are often tightly coupled with the vendor’s proprietary formats and workflows. This makes future innovation dependent on the hyperscaler’s roadmap, pricing, and terms of service.</p><h4>3. Data Sovereignty and Compliance Risks</h4><p>Manufacturers operating across multiple jurisdictions must adhere to local regulations regarding data storage, access, and usage. By leveraging global hyperscalers, companies can inadvertently expose themselves to compliance violations, especially when data crosses borders or is co-mingled in multi-tenant environments.</p><h4>4. Resale of Aggregated Intelligence</h4><p>Perhaps the most concerning is the possibility that hyperscalers, having learned from a company’s operations, may resell generalized versions of this intelligence to competitors. Even anonymized or abstracted learnings can erode a firm’s differentiation if similar AI agents are offered as a service to others in the industry.</p><h3>Why Sovereign AI is a Better Path</h3><p>For a cash-rich, moat-heavy manufacturer, Sovereign AI is a more strategic and secure approach. This model advocates for building and running AI on infrastructure that is either on-premises or hosted in private, controlled environments. The AI models are trained solely on internal data, for internal use, and never shared with external platforms.</p><h4>1. Preservation of Institutional Knowledge</h4><p>Sovereign AI codifies tribal knowledge, unwritten best practices, and domain-specific intuition into intelligent systems. This ensures that decades of expertise are retained, scaled, and passed down, rather than lost to attrition or hidden in hyperscaler pipelines.</p><h4>2. Full Data Ownership and Control</h4><p>With Sovereign AI, the company maintains complete control over how data is collected, stored, and processed. No data leaves the organization without explicit governance. This ensures compliance with evolving global regulations and internal risk policies.</p><h4>3. Custom-Tuned Agentic Systems</h4><p>Rather than relying on generic AI assistants, companies can develop agentic systems finely tuned to their specific workflows, terminologies, and business logic. This leads to better operational accuracy, lower error rates, and enhanced trust from internal users.</p><h4>4. Cost Control and Predictable Economics</h4><p>Instead of opaque pricing models and usage-based billing from hyperscalers, Sovereign AI allows companies to amortize their investments in hardware, software, and talent over time. Long-term, this often results in lower TCO (Total Cost of Ownership) and higher ROI.</p><h3>Practical Strategies for Implementation</h3><p>To pursue a Sovereign AI strategy effectively, legacy manufacturers should consider the following roadmap:</p><h4>1. Establish a Data Governance Framework</h4><p>Begin with a robust data governance plan that includes clear policies on data classification, ownership, access controls, and usage rights. Identify critical operational datasets, label them appropriately, and centralize storage in a secure data lake.</p><h4>2. Build Internal AI Capabilities</h4><p>Invest in AI engineering talent and reskill existing IT staff to handle machine learning operations, data engineering, and model deployment internally. Leverage open-source LLMs and frameworks (e.g., Llama 3, Mistral, or Falcon) for customized AI agent development.</p><h4>3. Deploy AI on Private Infrastructure</h4><p>Whether on-premises or in a private cloud with full tenancy control, deploy models on infrastructure that guarantees data sovereignty. Emerging technologies like AI accelerators (NVIDIA, Groq, Graphcore) can provide competitive compute power with predictable performance.</p><h4>4. Create Digital Twins and Simulations</h4><p>Utilize AI to build digital twins of production environments, logistics networks, and supply chains. These simulations can identify bottlenecks, predict equipment failures, and optimize throughput without risking live operations.</p><h4>5. Embed AI in Daily Operations</h4><p>Integrate agentic AI into frontline processes: predictive maintenance, quality assurance, supplier negotiations, employee onboarding, and real-time safety monitoring. Each deployment should reflect the company’s unique logic and legacy practices.</p><h4>6. Govern for Ethical and Secure AI Use</h4><p>Establish an internal AI ethics board and cybersecurity protocols. Ensure that AI models are auditable, explainable, and aligned with company values. Avoid black-box deployments that lack transparency or accountability.</p><p>“<strong>How does a legacy manufacturer compete with a hyper-scaler, with billions in Capex spend and existing infrastructure to attract top notch talent, they would be better off investing their Capex in machines that produce the widget instead of getting involved with GPU’s and trying to set up server farms</strong> “</p><p><strong>A Practical Middle Ground: The Hybrid Sovereign AI Model</strong></p><p>This raises a valid concern: how can legacy manufacturers justify competing with hyperscalers (Microsoft, Meta, Amazon, Google )in terms of capital investment and the war for AI talent? Most of these companies are naturally inclined to prioritize investments in core machinery and equipment — the very assets that produce their revenue-generating widgets — rather than diverting resources toward GPU clusters or building server farms. They often lack the in-house talent and infrastructure to manage sophisticated AI operations at scale.</p><p>A practical middle ground lies in a hybrid Sovereign AI approach. Instead of building everything from scratch, manufacturers can partner with regional or industry-specific cloud providers that offer dedicated environments with full data segregation and transparency. These providers often support open-source AI models and allow for tighter governance and customization without vendor lock-in.</p><p>Additionally, manufacturers can adopt containerized or edge-based AI deployments, where models are trained or fine-tuned centrally but deployed close to the machines or production lines. This not only preserves data sovereignty but also improves latency and operational efficiency.</p><p>By blending strategic outsourcing with localized control, companies can still protect their proprietary data and workflows — without incurring the full cost of scaling AI infrastructure or exposing their crown jewels.</p><h3>Closing the Loop Between Legacy and Innovation</h3><p>For a 60+ year-old manufacturing titan, the objective is not merely to digitize for the sake of modernization. It is to extend and amplify a legacy of excellence. Sovereign AI offers a bridge between tradition and transformation — a way to encode a company’s culture and ingenuity into intelligent systems that serve the enterprise, not a third-party vendor.</p><p>By owning their AI stack, controlling their data, and encoding their unique workflows into private agents, these companies can not only defend their moats but widen them. They can improve efficiencies, innovate faster, and future-proof their operations in a world that is becoming increasingly automated and intelligence-driven.</p><p>As the adage goes: “If data is the new oil, then Sovereign AI is the refinery you own.”</p><p>The path forward is not through dependency but through mastery — and that starts with reclaiming AI as a strategic, sovereign asset.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9652aa516f90" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>