
The signal: abundance still needs physical capacity
Artificial intelligence promises an economy that produces more with less effort. Yet a warehouse still needs to move goods, a hospital needs reliable ventilation, and a data centre needs an electricity connection. Software can improve decisions; it cannot remove every physical constraint.
Our central investment hypothesis is that companies which reduce a measurable shortage of labour, power or usable capacity may capture durable value. The test is not whether they belong to a fashionable theme. It is whether customers obtain a defensible economic payback and suppliers retain attractive returns after delivering it.
Reported facts — ageing changes the productivity equation
The IMF’s October 2026 book, Shaping the Future of Asia, examines demographic change alongside technology, fragmentation and the energy transition. Its first chapter explains that ageing can reduce aggregate growth through labour supply, without necessarily reducing output per worker or per person: investment and productivity still matter. Conditions also differ across countries; Asia is not a single demographic story. [1][2]
MITUXA View — this strengthens the case for studying automation where staff are difficult to recruit and workflows are repetitive. It does not establish that every robot project pays for itself. Labour participation, training, migration and organisational changes are competing or complementary responses. The relevant unit of analysis is the customer’s actual operation.
Reported facts — the electricity constraint is local
The IEA’s Powering Artificial Intelligence, published on 14 October, surveys 34 policy jurisdictions representing 90% of global data centre capacity. It describes connection queues, permitting delays and pressure on electrical equipment supply chains. Its executive summary estimates that the largest operators are set to spend more than $700 billion in capital expenditure in 2026, mostly on new data centres. This is an outlook, not completed expenditure. [3]
MITUXA View — abundant annual energy production is not the same as deliverable power at the right site and hour. That distinction directs research towards electrical equipment, controls, cooling and retrofit efficiency. A large order book is encouraging only if equipment can be delivered, customers can connect and the supplier gets paid.
Modelled scenario — energy security changes the investment mix
Rystad’s public Global Energy Scenarios 2026 summary presents three possible pathways. In its reference Pragmatism scenario, reliance on imported energy peaks around 2034 as electrification advances. This is conditional modelling, not an observed turning point or a promise that fossil fuels disappear. The summary also notes that cleaner technology supply chains can create a different kind of dependence. [4]
MITUXA View — the common research question across these scenarios is how economies use energy more productively and build resilience. We favour examining the economics of indispensable equipment over treating a single energy forecast as a stock-selection rule. We reviewed the public summary, not the full proprietary scenario database.
Daifuku and Kardex: the same demand story, different execution
Reported — Daifuku’s first-half orders rose 31.6% to ¥440.2 billion, sales rose 8.9% to ¥355.5 billion and operating income rose 10.9% to ¥56.7 billion. Cleanroom systems were a major contributor. In August, management raised full-year sales guidance to ¥735 billion and operating income guidance to ¥113 billion. Guidance remains a management forecast. [5]
MITUXA View — Daifuku is a useful research priority because it converts material-handling complexity into installed systems. But its semiconductor exposure means that calling it an ageing-population beneficiary is incomplete. Project profitability, the quality of backlog and customer concentration deserve as much attention as order growth.
Reported — Kardex provides a useful counterexample. First-half bookings grew 25.8%, while EBIT fell 38.4% and the group EBIT margin dropped to 6.8% from 11.8%. Management cited product mix, weaker utilisation in Automated Products and growth investment; it expects recovery, which still needs to be delivered. [6]
MITUXA View — a shared structural tailwind does not make these companies interchangeable. Integration, product mix and execution can separate demand growth from shareholder value. A recovery forecast is an item to test, not a margin already earned.
Belimo: efficiency can grow faster than cash
Reported — Belimo’s first-half sales reached CHF676.4 million, up 29.6% in local currencies and 20.5% in Swiss francs. Data centre cooling contributed slightly more than half of absolute sales growth. The EBIT margin was 22.5%, but free cash flow excluding term deposits fell to CHF38.7 million from CHF52.8 million, with working capital and capacity investment absorbing cash. [7]
MITUXA View — this is the attraction and the discipline in the same set of accounts. Controls and cooling can solve expensive operating problems. Rapid growth can nevertheless require inventories, production capacity and funding before it releases cash. Monitor cash conversion over a full cycle rather than judging the business by revenue alone.
Valuation: the market can already own the story
A popular investment narrative says scarcity should support suppliers’ pricing power. This is a narrative, not a quantified consensus forecast. Competition, customer bargaining power, capacity expansion and regulation may transfer the benefit elsewhere.
We have not established current fair values or an entry price for the companies in this issue. Operational evidence alone cannot establish that any share is cheap. Start by testing the growth, margin and cash conversion required to justify today’s enterprise value, then allow for a higher discount rate or slower delivery.
Illustrative calculation — if earnings rise 25% but a share’s price/earnings multiple falls from 40 to 30, its price falls 6.25%, before dividends. These are hypothetical inputs, not actual company multiples or a return forecast. Strong business growth can coexist with a disappointing investment.
Risks and conditions that would weaken the thesis
The thesis weakens if customers postpone automation because payback is too long, if labour shortages ease, or if installations fail to deliver promised productivity. For equipment suppliers, competition and added capacity can remove the scarcity premium. Grid delays can defer revenue rather than increase it.
We would also reconsider company-level conviction if backlog repeatedly fails to convert into profitable sales, cash conversion stays weak after capacity spending normalises, or leverage grows faster than operating cash generation. A structural story cannot excuse a deteriorating balance sheet.
What we will monitor
At Daifuku: order conversion, cleanroom concentration, project margins and cash generation. At Kardex: whether the expected second-half margin recovery appears in reported results. At Belimo: the balance between cooling growth, working capital, capacity spending and cash conversion.
Across the theme: customer payback, cancellations, grid connection lead times and the cost of financing new projects. The investment conclusion remains selective research: identify who removes a costly constraint, assess who retains the savings, and demand a valuation that leaves room for execution risk.
Sources and scope
- IMF — Shaping the Future of Asia, October 2026
- IMF — Asia and Pacific at a Turning Point, chapter 1
- IEA — Powering Artificial Intelligence, 14 October 2026
- Rystad Energy — Global Energy Scenarios 2026, public summary
- Daifuku — H1 2026 results and August outlook
- Kardex — Half-Year Results, 30 July 2026
- Belimo — Semiannual Report 2026, Management Report
Source check: 20 October 2026. The IMF and IEA publications are the current structural signal; company disclosures refer to the first half of 2026 and are not presented as new quarterly releases. Figures are reported, management forecasts or scenarios as labelled.
Research disclaimer
Independent research for information and educational purposes. This article is not personalised investment advice or an instruction to trade. Scenarios, forecasts and valuations can be wrong. The companies discussed are a research universe, not an announced model-portfolio change.
Which operating constraint deserves closer examination? Write to questions@mituxa.com.