It’s my independent research project, inspired by my and my family’s needs, that uses advanced AI models to map which public companies are most likely to create outsized value in the AI era, then bundles that work into clear, data-driven updates for serious retail investors.

Whether we can systematically identify companies that will deliver stronger equity returns than the Mag 7 over a 5–10 year horizon as AI-era constraint shifts — in cognition, coordination, energy, and capital substitution — compound. And critically, whether we can build a trackable record that proves or disproves that thesis over time, before the market reprices.

The phrase comes from Emad Mostaque: a world where AI collapses the cost of cognition and coordination, shifting scarcity to energy, compute, capital, and distribution. For investing, that reframes the question — instead of extrapolating current earnings, I map which companies control the binding constraints (compute, power, manufacturing, data, permissioning) and the distribution rails where capability becomes paid workflow. As those constraint shifts compound over 5–10 years, value migrates to whoever owns the gates.

I’m not selling hot tips, trading alerts, or hypey narratives; I apply a consistent methodology and framework (“The Last Economy”) across 100+ symbols in a single, transparent research stack rather than running a chat room or signal service. I’m not paid to promote any asset and declare my holdings.

It’s just me: a retired tech exec and engineer who grew up in the 70s/80s, studied Computer Science, spent decades building products and infrastructure in and around Silicon Valley, and happened to make life-changing early investments in names like AMZN and NVDA; I’m now financially independent, investing my own capital, and using this project to think rigorously about the AI transition.

Both. The explicit goal is to systematically identify companies I expect to outperform the Mag 7 as AI-era constraint shifts compound over 5–10 years — and to build a verifiable record that proves or disproves that thesis over time. My own experience with Amazon and NVIDIA taught me that deep product understanding can reveal futures the market hasn’t priced. Next Arc is the tool I use to do that more rigorously across a broader universe. What it offers members is not a guarantee, but the map, the methodology, and the weekly evidence trail.

You shouldn’t trust anyone blindly, including me; what I offer is a clear methodology, consistent scoring, visible disclosures of my own holdings, and a non-sensational, tech-optimistic lens grounded in both quantitative data and first-principles thinking about AI and “The Last Economy” — your job is to treat it as one thoughtful input into your own decision-making, not a substitute for it.

I track ~100–120 public companies across AI chips, cloud, robotics, biotech, energy, sensors, and frontier tech; it’s a curated universe biased toward technologies that matter in an AI-intensive world.

I was already starting to diversify my AI oriented holdings when I heard about "The Last Economy". The thesis reflected a lot of my own thoughts of the inevitable future. As a result, most of the assets in the universe happen to be those I was already holding. Each time I hear about something new, I research it and determine if it makes sense to include. They’re included if they play a meaningful role in the AI stack—compute, autonomy, energy, data, or distribution—or if they have asymmetric exposure to Last Economy tailwinds; I exclude assets with poor liquidity, low relevance, or too little transparency for consistent analysis.

The full universe is re-run weekly using the latest models, filings, news, and macro signals; large moves in fundamentals or narratives typically surface the same week.

It’s an AI-driven estimate of potential enterprise-value expansion over the next 5 years under realistic but transformative scenarios; the buckets help compare how much value creation each asset might capture in a Last Economy world, not precise price targets.

Each risk dimension (execution, financial, competitive, regulatory, and dependency risk) is scored separately using structured prompts and quantitative inputs; the platform then blends them into a single composite that captures fragility rather than volatility.

Financials come from public filings and market data; qualitative signals come from curated news, earnings calls, and technical documents; scenario exploration and scoring use GPT-5 grade models plus custom prompts shaped around The Last Economy thesis.

The heavy lifting (scenario generation, scoring, and narrative synthesis) is done by frontier models, but I design the prompts, sanity-check outputs, tune the methodology, and set the universe and weighting logic; it’s the human-guided research that I would do for myself (but better now others are using it).

Yes, if a model output doesn’t feel right then I’ll experiment with engineering or prompting fixes then rerun with clearer constraints; transparency matters, so I correct errors rather than letting the pipeline drift.

I think of them as "directional" and "relative". I also expect them to change frequently as the world around them is changing frequently. I treat them as signals to investigate, consider, understand and incorporate into my broader hypotheses. To me they are informed opinions grounded in the zeitgeist of the moment - based on much more data and a much better understanding than I would be able to achieve individually.

The value shifts driven by AI constraint collapse are invisible in current financials. By the time they show up in earnings, the market has already repriced. The platform identifies structural positioning — who controls the gates, who has the flywheels — before the earnings proof arrives. That’s the gap between where the market is pricing today and where the Last Economy thesis says value will migrate over 5–10 years.

Every company in the universe runs through the same 9-stage analytical pipeline weekly, scored by multiple frontier models for cross-validation. The platform produces structured, comparable metrics — including M.I.N.D. scores (Material, Intelligence, Network, Diversity) and AI Industrial alignment — rather than ad-hoc opinions. That consistency means you can track how scores evolve and compare positioning across the entire universe on the same terms.

In my mind Reader is for those new to retail investing who perhaps want to just "dabble". Allocator makes sense once you are a little more comfortable with data driven investing. Builder is for people like me, experienced tech investors who actively take a portfolio approach and are always following the market. The Free tier is a place for folks to see some of the data and understand if upgrading is right for them.

Free members get access to weekly updates with a subset of simplified commentary. At each paid tier, the depth of the data and analysis goes up. At the Builder level you get access to the full analysis PDFs and CSV files for inclusion in your own models.

Builder unlocks every data table and CSV, the most detailed data, and the full interactive screener; it’s the complete research stack I personally use to inform my own thinking.

Data for every membership tier is updated around the 7th, 14th, 21st and 28th of each month. Sometimes I will delay the analysis if there is a big raft of earnings being released or life events get in the way.

I have turned on the "free trial" flag within Patreon so you should be able to try for a week. At the same time you can join at the Free membership level. Finally, the web site includes recent PDFs of all the data from each of the membership tiers so you can see specifically what you would have access to.

You join via Patreon, and your tier automatically unlocks gated pages on nextarcresearch.com through Patreon OAuth; the site checks your membership level and reveals the appropriate tier’s content.

Yes. You control everything through Patreon; upgrades, downgrades, and cancellations take effect through the Patreon site according to their standard practices.

All tiers give access to all covered symbols. The list evolves as new AI-era assets become meaningful or irrelevant.

Use Next Arc to get one perspective where each company or network sits in the AI-era landscape — who has tailwinds, who faces structural risk, and which names merit deeper investigation.

Yes — the research helps you understand what’s inside those ETFs and which specific names are most relevant, if you choose to explore individual positions.

The PDFs are convenient for those looking for an offline accessible document. The CSV files are convenient for those who might be using a spreadsheet to model their investments. You can decide if either of those matches your needs.

You could glance at the week-over-week score changes, risk movements, and narrative shifts; using that to understand how narratives and risk assessments are shifting.

They are a little tongue-in-cheek experiment. You can use them to explain to your non-investing friends/relatives what the hypothesis for a company is.

This is research only — no trading signals, no financial advice; I show one prospective map, you decide the route.

Yes; Builder membership and above includes downloadable CSVs so you can sort, filter, and explore the universe in your own models.

No — it’s research, not advice. I don’t know your risk tolerance, time horizon, or personal circumstances; use this as one input among many.

Yes — I disclose my major holdings and relevant biases so you can understand where my incentives or experience might influence the framing. I am not paid to promote any asset.

Implied multiples reflect potential in a plausible AI-era scenario, not a forecast; real-world outcomes depend on execution, competition, capital cycles, and plain luck.

The platform is designed to build a trackable record over time. Each week’s scores, risk assessments, and growth projections create a trail of falsifiable claims that can be compared against actual outcomes as they unfold. The goal isn’t a backtestable signal service — it’s a transparent methodology where you can watch whether the structural positioning we identify actually translates into value creation.

I rerun the models with updated context and adjust the scores or narratives each week; the weekly refresh ensures the universe evolves with new data rather than freezing old assumptions.

Macro risk sets the backdrop, but company-specific resilience matters more in an AI-driven world; the framework blends both, but emphasizes structural advantage over short-term cycles.

Models can overweight recent narratives, misread technical details, hallucinate connections, or underappreciate real-world constraints like supply chains, physics or regulation.

It’s scenario-driven, not predictive; it depends on public data and AI interpretation; and it’s better at mapping strategic positioning than timing markets — useful for long-term thinking, not short-term trading.

AI models are the engine of The Last Economy — they compress cognitive labor, shift value to compute and energy, and reshape competitive dynamics; the models I use also help illuminate which companies are best positioned to ride that shift.

They are direct extensions of the same trend: intelligence becoming cheap and embodied; I treat them as high-conviction frontier categories where breakthroughs in autonomy, wetware, and physical AI could create new dominant players.

Yes — I skip projects with weak fundamentals, low liquidity, unclear governance, or hype-driven roadmaps; if I can’t reasonably model its trajectory in a Last Economy context, it doesn’t go in the universe.

Price Position measures where the current stock price sits within its 52-week range, normalized to a 0-1 scale. A value near 0 means the price is close to its 52-week low, while a value near 1 means it's near its 52-week high. This helps identify whether a stock is trading at relatively depressed levels (potential value opportunity) or near recent peaks (potential momentum or overextension). It's a purely technical metric that doesn't account for fundamentals. Use it to spot entry/exit timing patterns in combination with the fundamental analysis.

Price Compression measures how tightly the stock has been trading relative to its historical volatility. It's calculated as the ratio of recent price range to the 52-week range. A low compression value (near 0) indicates the stock has been range-bound or consolidating—often a precursor to a breakout move in either direction. A high compression value (near 1) suggests the stock is actively using its full 52-week range, indicating higher volatility or trending behavior. Traders often watch for low compression followed by volume expansion as a signal that accumulation or distribution may be ending.

Upside Compression captures the gap between the stock's current Price Position and its AI-driven 5 year growth forecast. It's calculated as the ratio of the implied 5 year multiple to the current Price Position. A high Upside Compression value suggests the market hasn't yet priced in the growth potential identified by the analysis—potentially an undervalued opportunity if the thesis plays out. A low Upside Compression value indicates the market is already pricing in significant growth, leaving less upside margin for error. This metric helps distinguish between "expensive for a reason" (justified by structural advantage) and "expensive relative to realistic outcomes" (priced for perfection).

I review the universe continuously and make adjustments as technologies mature or become irrelevant; meaningful additions or removals typically happen every few months, not weekly.

Yes — as the AI landscape evolves, I refine prompts, weights, and inputs; major changes are explained in the weekly update notes so you always know what shifted and why.

I never see your Patreon credentials; authentication happens directly between Patreon and the website via OAuth, and the site stores only the minimum non-personal metadata needed to confirm your membership tier.

I post on Patreon with each refresh cycle for each membership tier.

I am constantly making UI improvements or extending the data available. These small changes are reflected in the changelog. As Patreon membership increases and more funds become available I expect to scale to more assets and to more complex analyses by a broader range of models to benefit from their collective wisdom.

You can message me directly on Patreon — I review all requests and add new symbols if they genuinely fit the AI-era framework and have enough transparency to analyze properly.