Investing in the Invisible Stack
These are not moonshot bets.
Investing in the Invisible Stack
Four companies — EMR, TYL, PSN, DELL — are quietly building the AI-powered infrastructure that cities, utilities, and democratic institutions will run on for the next generation. Here is why I own them, and why I think it matters beyond the ticker.
My total investment in individual stocks — these four included — is under $1,500. At roughly $375 per position, no single outcome here will move the needle on anything that actually matters financially. The real money, as it should be at my age, lives in diversified mutual funds that I think about considerably less and trust considerably more. These stocks are recreational in the most honest sense of the word: they keep me engaged, they give me a reason to pay close attention to ideas I find genuinely interesting, and they let me put a token amount of skin in the game on a thesis I believe in. This essay is not a recommendation, and the positions are not a strategy. They are, as much as anything, an excuse to think carefully about what kind of companies deserve to exist — and to write it down.
Why these four?
Most AI investment conversation gravitates toward the obvious: the model makers, the chip giants, the consumer platforms chasing the next billion users. I find myself more interested in a different question — not who builds the most spectacular AI, but whose AI will quietly become load-bearing infrastructure for the societies we actually live in.
The four companies in this portfolio — Emerson Electric, Tyler Technologies, Parsons Corporation, and Dell Technologies — do not look like an AI play at first glance. One makes industrial automation equipment. One sells software to county governments. One builds airports and traffic systems. One makes servers. But examined together, they form something I've come to think of as the Invisible Stack: the layered technological substrate that will govern how cities deliver water, adjudicate parking tickets, time traffic signals, route emergency calls, manage watersheds, and optimize power grids — all through AI, all below the threshold of public scrutiny, and all with extraordinary long-term durability.
The most consequential AI deployments of the next decade will not be chatbots. They will be the quiet operational intelligence embedded in the systems that keep cities alive.
These are not moonshot bets. They are investments in the companies that are building the unglamorous, essential machinery of AI-enabled civilization — and doing it in ways that align, perhaps surprisingly well, with values I care about: environmental responsibility, democratic governance, and social equity.
The environmental argument
Environmental investing often gets reduced to a binary: fossil fuels versus renewables. That framing misses something important. The most impactful environmental technology is not always the generation technology — it is the optimization layer that determines how efficiently existing systems run.
Emerson Electric · EMR
Emerson's Ovation platform now controls approximately 30% of global power generation capacity, and its Ovation Virtual Advisor — the first GenAI system purpose-built for power and water industries — runs locally on pre-trained models, with no cloud dependency. This is a critical design choice: it means the AI operates within the secured operational technology environment of a water treatment plant or power station, making real-time decisions about pump efficiency, anomaly detection, and maintenance scheduling without sending sensitive infrastructure data to a public cloud.
Emerson has reduced its own Scope 1 and 2 greenhouse gas emissions intensity by 52% since 2021, sources 49% of its electricity from renewables, and has Science Based Targets approved to net zero by 2045 — aligned with the 1.5°C Paris Agreement trajectory. More importantly, its products are precisely the tools that allow other organizations to reach their own decarbonization targets. Emerson's commercial success and its environmental mission are structurally the same thing.
Emerson's commercial success and its environmental mission are not in tension. They are the same thing. Every efficiency gain it sells is also a carbon reduction it enables.
Dell Technologies · DELL
Dell's contribution to the environmental picture is less obvious but equally real. Its AI Factory architecture, designed for on-premise inference rather than cloud-based computation, is materially more energy-efficient for organizations running AI at scale. Dell's own research suggests on-premise AI inferencing can be up to 62% more cost-effective than equivalent public cloud workloads — and the energy picture follows the economics. Shifting AI inference from massive centralized data centers to smaller, locally deployed edge hardware reduces transmission losses, allows renewable micro-generation to power local compute, and avoids the enormous water consumption of hyperscale cooling systems.
Dell's prediction for 2026 and beyond — a shift from large language models to small, task-specific models optimized for edge deployment — points toward AI infrastructure that is inherently more sustainable: less compute per decision, less energy per inference, deployable on local renewables.
The democratic governance argument
There is a version of AI development that is deeply centralizing: a small number of enormous technology companies accumulate data, build proprietary models, and sell access to AI capability that governments and citizens increasingly depend on but cannot understand, audit, or control. That trajectory concerns me, for reasons that extend well beyond investment risk.
The companies in this portfolio represent a different trajectory — one that is, in important ways, structurally more democratic.
Tyler Technologies · TYL
Tyler Technologies is the largest company in North America solely dedicated to public sector software. Its clients are not corporations — they are county clerks, municipal courts, water authorities, school districts, and sheriff's departments. Its AI does not optimize ad targeting or engagement metrics. It routes building permit applications, helps residents navigate government services in plain language at 9 pm on a Sunday, assists public defenders with case research, and flags budget anomalies that might indicate fraud or misallocation.
This is AI in the service of democratic administration. Tyler's Resident AI Assistant is live in six US states, handling tens of thousands of citizen queries monthly — deflecting routine calls from overstretched public servants and routing complex cases to the humans who can actually help. Its acquisition of For the Record, a courtroom transcription specialist, is building what Tyler calls "judicial intelligence" — AI that makes the justice system faster, more auditable, and more accessible to people without expensive lawyers.
Tyler's AI is also notably sovereign in character. It runs on government data, is contracted to government clients, and operates within the legal frameworks those governments set. It is not a black-box consumer product — it is accountable infrastructure.
Parsons Corporation · PSN
Parsons is harder to characterize. It is, in part, a defense contractor, and defense spending deserves scrutiny. But Parsons' critical infrastructure work — the majority of its revenue and the focus of its civilian business — is genuinely oriented toward a particular vision of how cities function: as integrated, resilient, technology-enabled systems that serve populations rather than extract from them.
Parsons has built digital twin systems for city infrastructure — virtual replicas of transportation networks, utilities, and urban corridors that allow city operators to model and test changes before implementing them. It designed the intelligent transportation network integrating 125 agencies across Southern California. It is building mobility systems for Riyadh, urban public realm for Diriyah, and border systems for multiple sovereign nations. Crucially, it does this as a program manager and designer — bringing cities the capability to own and operate their own infrastructure, rather than creating dependency on proprietary platforms they cannot control.
The social equity argument
I want to be careful here. These are not social enterprises. They are for-profit corporations operating in competitive markets, and their primary obligation is to shareholders. I do not hold them to a standard I would apply to a nonprofit. But I do think investment decisions reflect values, and I think there are honest reasons to believe this portfolio has a positive social valence — not despite its commercial logic, but because of it.
The communities that most benefit from efficient permitting software, from AI-assisted courts, from reliable water systems with predictive maintenance, from better-managed traffic in congested urban corridors — these are not wealthy communities. Wealthy communities hire expensive consultants, navigate bureaucratic friction, and absorb infrastructure failures through private alternatives. Working-class communities and small municipalities bear the full cost of broken systems. A Tyler permitting platform that cuts application review from a week to three minutes makes housing development faster and cheaper — and housing affordability is one of the defining social crises of this decade.
An Emerson water system with AI-enabled anomaly detection that catches pipe failures before they become crises matters more in a city that cannot afford emergency repairs than in one that can. A Parsons-designed transit corridor in a Middle Eastern city creates mobility — and through mobility, economic opportunity — for populations that currently have none. A Dell edge AI system that allows a rural county government to operate AI locally without cloud subscriptions it cannot afford democratizes access to capability that would otherwise require a large IT budget.
The communities that most benefit from efficient public AI are not the wealthy ones. They are the ones that cannot absorb the cost of broken infrastructure by paying for private alternatives.
None of this is guaranteed. Technology deployed without accountability can harm the communities it claims to serve. AI in courts can perpetuate bias. Traffic optimization can prioritize some neighborhoods over others. These are real risks, and they deserve ongoing scrutiny. But the direction of travel — AI embedded in public institutions, accountable to elected governments, built for operational continuity rather than engagement maximization — is the right direction.
The investment case
I have spent most of this essay on values, which is appropriate because values are why I chose this portfolio over others with comparable financial profiles. But the financial case is real and worth stating plainly.
Emerson controls roughly 30% of global power generation automation, operates a $1.5 billion industrial software Annual Contract Value business, and has just completed its acquisition of Aspen Technology — giving it a full-stack AI play targeting a $150 billion addressable market growing at 5–7% annually. Its near-term guidance calls for 5.5% revenue growth in fiscal 2026 with expanding margins. Europe and China remain soft, but the Middle East, Southeast Asia, and Latin America are in aggressive industrialization cycles that favor Emerson's product portfolio precisely.
Tyler Technologies generated $2.3 billion in revenue in 2025, growing at over 9% year-on-year, with SaaS revenue growing at 20%. It carries 45,000 installations across 15,000 government locations — a customer base with extraordinarily high switching costs and multi-decade contract cycles. Its forward multiple is elevated (roughly 71x earnings), which demands continued growth execution, but its competitive moat — owning the operating software that governments cannot practically replace — is about as durable as a business advantage gets.
Parsons carries $8.7 billion in backlog plus $11 billion in unbooked contract wins where it is the sole prime contractor. Its win rate in 2025 was 61%. Its cyber and intelligence segment has grown more than 20% annually for two consecutive years. Its Middle East infrastructure business is expanding rapidly as Gulf nations invest in post-oil urban economies. It is a smaller, less-followed company than the others — which means less efficient pricing and potentially more upside.
Dell is the largest and most financially powerful of the four: $111.7 billion in revenue guidance for fiscal 2026, up 17%, with AI server shipments expected to reach $25 billion — up over 150% year-on-year. Its Infrastructure Solutions Group reported $14.1 billion in revenue in Q3 FY26 alone, up 24%. Dell is the clearest direct beneficiary of enterprise AI infrastructure spending, and its on-premise AI Factory architecture positions it to capture a category of AI deployment — sovereign, secure, locally controlled — that is distinct from the hyperscaler cloud market and potentially much larger.
Caveats and risks I hold alongside this thesis
I try to hold my convictions with calibrated uncertainty, which means being explicit about what could go wrong.
Tyler's valuation is demanding. At roughly 71x forward earnings, any stumble in growth execution or a shift in government spending priorities could compress the multiple significantly. The Q4 2025 revenue miss — modest in absolute terms but enough to trigger a 14% stock decline — is a reminder that the market prices in perfection and punishes any deviation.
Emerson faces near-term headwinds in Europe and China. Its AspenTech integration is proceeding well but adds complexity. The industrial automation market is competitive — Honeywell and Siemens are investing heavily in AI as well, and the competitive dynamics in this space are not fully settled.
Parsons operates significantly in defense and national security, where contract risk, classification constraints, and geopolitical sensitivity create uncertainties that are genuinely hard to model. Its Middle East infrastructure growth is real but concentrated in a region with its own political dynamics.
Dell's hardware business is fundamentally cyclical, and AI server demand, while strong now, could be subject to inventory digestion cycles similar to those that have periodically whipsawed semiconductor and server markets. Its deliberate exit from Chinese supply chains adds short-term cost and complexity.
And across all four: AI regulation is coming. Not yet, not comprehensively, not at the sub-city operational level where most of this technology currently operates — but eventually. The companies best positioned for that regulatory future are those whose AI is already accountable to democratic institutions, which is part of why I find this portfolio's governance profile reassuring. But it is not a guarantee.
Why I write this here
This blog is called AI Almost Daily because I find myself thinking about AI almost every day — not as a technology enthusiast chasing novelty, but as someone trying to understand what is actually happening to the world and to make considered choices within it.
Investment is one of those choices. Money directed toward companies that build AI in accountable, environmentally responsible, publicly beneficial ways is a different kind of choice than money directed toward companies that do not. I do not think personal investment portfolios change the world — the amounts involved are too small and the mechanisms too indirect. But I do think they reflect and reinforce values, and I think writing about them publicly is a form of accountability.
The Invisible Stack — Emerson, Tyler, Parsons, Dell — is where I have chosen to place a meaningful portion of my attention and capital. I think it is where a significant part of the AI story will actually play out: not in the spotlight, not in congressional hearings about chatbots, but in the water plant at 3am when the AI catches an anomaly before it becomes a crisis, in the county courthouse when a resident gets an answer about their case without waiting two weeks, in the traffic cabinet at an intersection in Bellevue where a rugged server is quietly making the signal smarter.
That is the AI that matters most to most people. I want to own a piece of the companies building it.
© Man Withahat, flextalks.org. Licensed under CC BY-SA 4.0.