Skip to content
Cardan-AI
Analyses

Analyses

Daily AI briefings for regulated and industrial sectors — aerospace, O&G, energy, luxury — read through a business and economics lens. Open access, published by Cardan-AI.

Aero & Defense2 October 2026

Allies, Not Equals: What the Gravity Model of Trade Says About ITAR Easing Among AUKUS Countries

Washington is easing ITAR export controls between the US, UK and Australia for defense technology, AI included. The gravity model of trade (Anderson & van Wincoop) explains why this easing, confined to the AUKUS circle, can mechanically worsen the relative position of allies left outside it.

Read the analysis
Industry1 October 2026

Why US AI-Robotics Firms Are Scaling in the Gulf, Not From the US

Gecko Robotics points to a wider pattern of US industrial robotics firms choosing to scale operations in the Gulf. Dunning's eclectic OLI paradigm (1977) explains why direct investment there — rather than plain export or licensing — is the rational choice for a technology that is both proprietary and sensitive.

Read the analysis
Energy30 September 2026

Who Pays for AI's Electricity? What Ramsey and Boiteux Say About the Bill Coming Due

EQT projects that AI power demand could exceed U.S. residential demand by 2030. Ramsey-Boiteux pricing theory explains why the real battle isn't over megawatts but over how their cost is split between households and data centers.

Read the analysis
Aerospace & Defense29 September 2026

Distribution Settled, Profits Relocated: What Christensen Says About the Rush on GenAI.mil

GenAI.mil topped 2 million users in a week and generated over 100,000 autonomous agents by July 2026. Christensen's law of conservation of attractive profits explains why the real economic battle is no longer about model access.

Read the analysis
AI Regulation28 September 2026

Rules versus discretion: what Kydland and Prescott say about the EU AI Board's silence

On September 17, 2026, the European AI Board met and decided nothing. Kydland and Prescott's framework on policy credibility explains why that silence carries a direct, measurable cost for companies that have to invest in compliance today.

Read the analysis
AI Infrastructure27 September 2026

Space as a factor of production: what Von Thünen's rent theory says about Google's orbital AI bet

Google's Project Suncatcher tests whether AI compute is cheaper to run in orbit, where solar power is up to eight times more abundant. A two-century-old model of land rent explains why this is a genuine site-selection problem, not a stunt — and which workloads could plausibly move there first.

Read the analysis
Aerospace & Defense26 September 2026

The winner's curse is bidding for pre-revenue defense AI

VisionWave's $99.6 million valuation of a pre-revenue decision-latency engine is priced against a $48.76 billion 2029 market forecast, not a track record. Fifty years of auction theory says that is exactly the setup in which buyers systematically overpay.

Read the analysis
AI Governance25 September 2026

Who owns the decision an AI agent makes? Grossman-Hart-Moore and China's control-rights experiment

China's new AI agent guidelines force every developer to pre-assign who controls a decision an autonomous agent will eventually face. That is not a compliance checkbox — it is the incomplete-contracts problem that Grossman, Hart and Moore formalized for the theory of the firm, applied to software for the first time at regulatory scale.

Read the analysis
Aerospace & Defense24 September 2026

Why an uncrewed freighter is worth more than its standalone reliability: Markowitz and the defense logistics portfolio

Textron and Merlin just unveiled an uncrewed freighter for contested military logistics. Portfolio theory, as formalized by Harry Markowitz, explains why its value shouldn't be judged by its standalone reliability, but by what it adds to the existing mix of supply routes.

Read the analysis
Aerospace & Defense23 September 2026

Why the Pentagon is forcing open architecture on defense AI: Farrell and Saloner on standards and coordination failure

GA-ASI and Tactical Air Support just showed that two competing companies' aircraft can share a live targeting picture without custom integration — because both comply with a government-mandated open architecture. The economics of technology standards explains why the Pentagon, not the market, had to impose it.

Read the analysis
Aerospace & Defense22 September 2026

Joby's autonomous flight and the economics of span of control: what Garicano says about remote supervision

Joby's transcontinental autonomous flight, supervised from 2,323 miles away, shifts the binding economic constraint on aerial autonomy away from piloting and toward the design of supervision hierarchies — a problem the economics of organization, via Garicano (2000), lets us formalize.

Read the analysis
Economics of AI21 September 2026

Diminishing returns in the race for bigger models: what research-productivity economics says

TechCrunch reports that frontier AI development costs are rising faster than the returns they generate — a dynamic that Bloom, Jones, Van Reenen & Webb's (2020) framework on declining research productivity reads with precision.

Read the analysis
AI Sovereignty20 September 2026

The yttrium chokepoint: what global value chain governance theory says about China's rare-earth leverage

China's tightened export licensing on yttrium — a rare earth critical to jet engines, semiconductors and defense electronics — is a textbook case of Gereffi, Humphrey & Sturgeon's (2005) 'captive' value-chain governance: control concentrated at one hard-to-substitute node.

Read the analysis
AI Sovereignty19 September 2026

Critical-minerals exploration as an under-provided public good: what the EIB report reveals about AI sovereignty's blind spot

An EIB report published September 18 quantifies the EU's mineral-exploration gap: investment 6-7 times below world leaders, 75% of drilling concentrated in 4 countries. Samuelson's (1954) public-good framework explains why this underinvestment is structural, not accidental.

Read the analysis
Agentic AI18 September 2026

AI Agents and the Matching Function: What Adecco's Rollout Means for Labor-Market Frictions

Adecco is rolling out agentic AI to 27,000 employees and reporting 20,000 additional placements. Diamond-Mortensen-Pissarides matching theory explains what's really at stake: an efficiency shock to the very function that staffing intermediaries sell — and an existential question for those who sell it.

Read the analysis
Aerospace & Defense17 September 2026

Codifying Tacit Knowledge: What a SpaceX Spinout's Test Software Means for Aerospace's Barriers to Entry

A startup founded by two ex-SpaceX engineers is selling aerospace developers a productized version of the internal data discipline that took SpaceX years of high-cadence testing to build. Von Hippel's 'sticky information' and Cowan, David and Foray's economics of knowledge codification explain why this matters more than the software category suggests.

Read the analysis
Energy16 September 2026

AI Data Centers and Natural Gas: Real Options Theory Meets a Forecast That Doubled in Nine Months

BloombergNEF just doubled its forecast for U.S. data-center natural gas demand in nine months. McDonald and Siegel's (1986) real options theory explains why this instability should slow gas infrastructure investment — and why competitive racing among hyperscalers produces the opposite effect.

Read the analysis
Aerospace & Defense15 September 2026

Rare Earths and the Weaponization of Supply Chains: What Hirschman Knew About Dependency

Deloitte's 2026 midyear outlook names critical minerals as a strategic chokepoint for AI-enabled defense platforms. Albert Hirschman diagnosed this dynamic in 1945 — and Farrell and Newman's 'weaponized interdependence' explains why it cuts deeper in today's networked supply chains.

Read the analysis
Energy14 September 2026

AI Weather Forecasting: What the Value-of-Information Theorem Says About Google's Bet on the Grid

Google DeepMind is now explicitly targeting grid operators with WeatherNext 3. Blackwell's (1953) value-of-information theorem explains why this bet makes growing economic sense: the more the grid decarbonizes, the more a marginally better forecast is worth.

Read the analysis
AI Market & Investment13 September 2026

Europe's Innovation Act: after the capital, the market — what's really holding back AI scale-ups

In five weeks, the European Commission went from a €5 billion fund for tech scale-ups to a proposed horizontal law, the "Innovation Act." Paul Krugman's economies-of-scale trade theory explains why that sequence reveals an implicit diagnosis: Europe's problem isn't just available capital — it's the size of the market that capital gets deployed into.

Read the analysis
AI Regulation12 September 2026

California's "robo-boss" ban: a Brussels Effect without Brussels

California's SB 947, which bars employers from firing or disciplining a worker based solely on an automated system, reproduces a requirement the EU AI Act already imposes on worker-management AI. Anu Bradford's "Brussels Effect" explains why a standard regulates beyond its home territory — and why, here, the convergence happens without any formal export mechanism.

Read the analysis
Aerospace & Defense11 September 2026

Sovereign defense AI: what strategic trade theory explains — and what it leaves out

The NEC-Mitsubishi Heavy Industries partnership for a Japanese "sovereign" defense AI illustrates a broader dilemma: why does an ally choose to build, rather than buy, a capability its own security partners already sell? Brander and Spence's strategic trade theory answers the "why"; Katz and Shapiro's network-externality economics reveals the hidden cost.

Read the analysis
AI Regulation10 September 2026

The AI Act and aviation: why the longest deadline hides the highest cost

The EU AI Act gives AI embedded in already-regulated products — aviation, machinery, medical devices — an extra year (2 August 2027 versus 2 August 2026). Far from a courtesy, that gap reflects a structurally higher governance cost, which Oliver Williamson's transaction cost economics helps explain.

Read the analysis
Economics of AI9 September 2026

Defense AI: capital is abundant, trust is not — what Kenneth Arrow explains

$19.8B in defense tech VC in Q1 2026 (+146% year-over-year), yet Deloitte now names "trusted deployment", not technology, as the binding constraint. Kenneth Arrow's economics of trust (1972) explains why abundant capital alone cannot produce it.

Read the analysis
Agentic AI8 September 2026

Agentic AI in industry: the real bottleneck is organizational co-investment, not technology

IDC (via Deloitte) puts 36% of industrial tasks within reach of agentic AI automation, and forecasts a 3.5x growth in US aerospace-defense AI spending by 2029. The Autor-Levy-Murnane (2003) and Bresnahan-Brynjolfsson-Hitt (2002) frameworks explain why this technical ceiling says nothing about the real pace of diffusion.

Read the analysis
AI Regulation7 September 2026

Mandatory AI content labeling gives luxury brands back a costly-to-fake signal

Since August 2, 2026, the EU AI Act (Article 50) and California's AI Transparency Act have independently imposed the same instrument: labeling AI-generated content. Costly-signal theory (Spence, 1973) and the Brussels Effect (Bradford, 2012) explain why this convergence matters especially for luxury brands.

Read the analysis
Energy6 September 2026

Industrial AI Cybersecurity: Why the UK Chooses Mandates Over Pledges

The UK is widening its legal duty to report cyber incidents across critical infrastructure and weighing dedicated AI controls. Moral hazard in information economics, network security games, and real-options theory explain why binding regulation, here, beats voluntary commitment.

Read the analysis
Aerospace & Defense5 September 2026

Pentagon's AI Blacklist: From Arbitrary Uncertainty to Measurable Risk

A federal judge ruled unconstitutional the Pentagon's 'supply chain risk' designation against an AI vendor, applied without a contradictory hearing. Frank Knight's risk/uncertainty distinction and the hold-up problem explain why this ruling matters for the entire defense AI vendor base.

Read the analysis
Markets & Capital4 September 2026

AI Compute Futures: Why Oil Can Be Stored and Compute Cannot

Big Tech's 2026 AI capex reaches parity with global oil & gas capex ($680B each), just as Wall Street launches the first regulated futures contracts on GPU computing power. Commodity storage theory reveals a fundamental structural gap between the two markets.

Read the analysis
Energy3 September 2026

Who Pays for AI's Grid? FERC Picks Price Over Rules

FERC's June 18, 2026 order (RM26-4-000) ends decades of socialized grid interconnection costs, forcing AI data centers to internalize the marginal cost they create — a textbook case study in choosing price over rules to address an externality.

Read the analysis
Aerospace & Defense2 September 2026

The market for military AI is a market for lemons — and the Pentagon knows it

The FY2026 NDAA requires the Pentagon to build a standardized AI model assessment framework, not complete until 2028. The immediate exclusion of Chinese vendors is not just a stand-alone national security measure: it is the second-best solution to an information asymmetry problem Akerlof formalized back in 1970.

Read the analysis
AI Cybersecurity1 September 2026

AI Cyber Defense: Why 116 Rivals Signed the Same Letter — and What That Doesn't Guarantee

116 technology companies, including head-to-head rivals, co-signed a call for collective action on AI cyber defense in late August. Collective-action economics explains why voluntary coordination is rational here — and why, by construction, it remains insufficient without a binding mechanism.

Read the analysis
Agentic AI31 August 2026

Agentic AI in industry: why 36% of potential doesn't mean 36% of adoption

Deloitte puts 36% of industrial manufacturing tasks within reach of agentic AI augmentation — a technical figure, not an adoption rate. The gap between the two is explained by task economics and complementary assets, not a cyclical lag that closes on its own.

Read the analysis
Aerospace & Defense30 August 2026

AI market sizing: the real problem isn't forecasting, it's definition

Two reputable firms size the same “AI in aerospace” market at $3.3B and $29.3B for 2025 — a 9x gap. The discrepancy doesn't measure forecast uncertainty; it reflects the absence of a shared definition of what “AI in aerospace” even counts, a problem measurement economics has documented for over sixty years.

Read the analysis
Energy & O&G29 August 2026

AI and energy: complementary inputs, not two separate files

The president of the American Petroleum Institute argues the US must “win the energy war before the AI war.” Behind the sectoral pitch for natural gas lies a real economic complementarity between compute capacity and energy capacity — with a risk of mutual under-investment if the two are planned separately.

Read the analysis
Aerospace & Defense28 August 2026

VENOM: when military autonomy becomes a kit rather than an aircraft

On July 16, 2026, DARPA and the US Air Force flew a production F-16 under full AI control using a removable kit rather than a dedicated airframe. This architectural choice shifts the economics of military autonomy from a fixed, bespoke capital investment to a modular component reusable across an existing fleet.

Read the analysis
AI Sovereignty27 August 2026

Genesis Mission: public coordination as an answer to underinvestment in scientific AI

Over $5B committed across 15 US federal agencies for the "Genesis Mission" (July 22, 2026) illustrates a precise economic mechanism: public coordination of complementary investments to overcome externalities that the market structurally underfunds on its own. Read through Rosenstein-Rodan's "big push" and Mazzucato's entrepreneurial state, with implications for European industrial firms.

Read the analysis
Aerospace & Defense26 August 2026

Palantir and the economics of dual use: when defense funds the commercial platform

In Q2 2026, Palantir posted $1.94B in revenue (+93% year-over-year), driven by 149% US commercial growth that now outpaces its defense and government base. An economist's reading: a textbook case of defense-to-civilian technology diffusion and increasing returns in software platforms, with direct implications for European aerospace and defense.

Read the analysis
AI Regulation25 August 2026

AI Act: what the ‘Digital Omnibus’ reveals about the economics of regulation

On 27 July 2026 the EU brought the Digital Omnibus on AI into force — the first substantive revision of the AI Act since it entered into force in August 2024 — amending implementation deadlines, simplifying part of the compliance regime, and adding protections against intimate deepfakes, while leaving Article 50 transparency duties (applicable since 2 August 2026) untouched. An economist's reading: a regulatory calendar that revises itself becomes a risk variable in its own right for compliance investment decisions.

Read the analysis
Energy & O&G24 August 2026

AI in oil & gas: why 92% invest but only 50% deploy

The global AI in oil and gas market triples by 2035 ($5.1B → $18.7B, 13.8% CAGR), yet 92% of enterprises invest while only 50% have deployed and hardware captures 43.4% of spend. An economist's reading: the sector is in the 'installation phase' of a general-purpose technology — capital precedes productivity, and the intent-adoption gap is a timing lag, not a failure.

Read the analysis
Aerospace & Defense23 August 2026

Aerospace AI: The Temptation to Extrapolate an Early Growth Rate Over Eight Years

The global aerospace AI market jumps from $2.29B (2024) to $3.28B (2025), and SkyQuest's report extends that pace (43.1% CAGR) to $57.61B by 2033. An economist's read: extrapolating a first-year growth rate off a tiny base is the classic trap of nascent markets — the law of small numbers applied to market forecasting.

Read the analysis
Luxury & Cosmetics22 August 2026

Beauty and AI: when a booming market is already slowing down

The global AI-in-beauty-and-cosmetics market grows from $4.38B (2025) to $10.86B (2030), yet its CAGR falls from 21.1% to 19.6%. An economist's read: diminishing marginal returns and base effects — the same mechanics that explain why growth rates slow as an economy, or a market, gets larger.

Read the analysis
Aerospace & Defense21 August 2026

Defense: a venture-capital version of the Solow paradox

$19.8B in defense tech VC in Q1 2026 (+146% YoY, PitchBook), a third of it flowing to autonomous systems — while the US Department of War's 2026 AI strategy provides no adoption metrics at all. An economist's read: Solow's productivity paradox (1987), reapplied to defense financing, where capital is pricing a productivity gain no instrument yet measures.

Read the analysis
AI Regulation20 August 2026

The EU AI Act's energy rules: how a fixed compliance cost reshapes market structure

Since August 2, 2026, AI safety systems in critical energy infrastructure fall under the AI Act's high-risk obligations (fines up to €15M or 3% of global turnover). An economist's read: a largely fixed compliance cost that weighs proportionally heavier on mid-sized operators — regulation as a barrier to entry (Stigler, 1971).

Read the analysis
AI Regulation19 August 2026

US AI regulatory fragmentation: the hidden cost of federalism

Illinois now requires mandatory audits of "frontier" AI systems (up to $3M per violation), Colorado regulates chatbots, and the federal GAAIA preemption bill remains stalled. An economist's read: a case of uncoordinated regulatory federalism (Tiebout, 1956; Vogel, 1995), where compliance cost scales with the number of jurisdictions rather than actual risk.

Read the analysis
Aerospace & Defense18 August 2026

Aerospace-defense AI: the race between technology and skill

US aerospace and defense AI spending is set to grow 3.5x by 2029 ($5.8B, IDC/Deloitte), while the data-science share of job postings grows only from 3% to 5% over the same period. An economist's read: a race between technology and education (Tinbergen, 1974; Goldin & Katz, 2008), where skill supply is struggling to keep pace with AI capital diffusion.

Read the analysis
Aerospace & Defense17 August 2026

Defense venture capital: buying the option, not the technology

Defense-contractor-led funding rounds hit a record $4.1B in 2026, with Lockheed Martin growing its fund from $400M to $1B. An economist's read: minority corporate venture capital functions as a real option on still-uncertain technology (Kogut, 1991), cheaper than an outright acquisition or full in-house development.

Read the analysis
Energy16 August 2026

AI and Electricity: When Productive Capital Outruns Social Overhead Capital

The IEA projects $3.4 trillion in global energy investment for 2026, of which only $550 billion goes to electricity grids, while data center demand is set to double by 2030. An economist's reading: a textbook case of unbalanced growth (Hirschman, 1958), where productive capital outpaces the social overhead capital that makes it deliverable.

Read the analysis
Aerospace & Defense15 August 2026

US Federal AI Spending: Market Growth Is Also a Concentration of Buying Power

Between 2024 and 2026, the potential value of US federal AI contracts rose from $4.6B to $91.8B — but the share captured by the Department of Defense alone rose from 93.5% to 98.9%. An economist's reading: a textbook case of deepening monopsony, with a hold-up risk for suppliers investing in relationship-specific assets.

Read the analysis
AI Market & Investment14 August 2026

The AI-in-O&G market: when forecaster disagreement is itself the signal

Two research firms publish, weeks apart, CAGRs of 14.2% and 19.47% for the same upstream oil-and-gas AI market. An economist's read: forecast dispersion is a direct measure of market uncertainty, and should govern the pace of capital commitment more than the midpoint of the forecasts.

Read the analysis
AI Regulation13 August 2026

Buying AI from outside the EU: why the European buyer inherits a risk it cannot verify

The EU AI Act doesn't only cover non-EU providers: any business that imports or distributes an AI system embedded in a product sold in the EU must comply, whether or not it developed the system. An economist's read: a textbook adverse selection problem (Akerlof, 1970), where the buyer cannot verify a non-EU system's compliance quality before purchase.

Read the analysis
Luxury & Cosmetics12 August 2026

Luxury: when customer trust in AI outpaces brand maturity, who captures the relationship?

BCG x Altagamma's 2026 study shows 87% weekly AI usage among luxury consumers and 62% expecting AI assistance from brands themselves, versus only 40% of houses past the emerging AI maturity stage. An economist's read: this gap is not just an execution problem — it's a disintermediation risk on the trust layer.

Read the analysis
Energy11 August 2026

Energy: the AI productivity paradox is back — believing in the catalyst without measuring it

EY's 2026 US AI Pulse Survey puts energy at the top of strategic AI conviction (78% vs 55% cross-industry) and, simultaneously, at the top of the inability to measure its gains (56% vs 35%). Economist's read: this is not an inconsistency — it's Solow's paradox resurfacing in a new form.

Read the analysis
Aerospace & Defense10 August 2026

Aerospace & Defense: the real bottleneck isn't AI adoption, it's the capacity to absorb it

PwC puts AI usage in A&D design workflows at 57%, 16 points above the cross-industry average — yet 42% of executives already expect their infrastructure won't keep up. An economist's read: the bottleneck has shifted from adoption to organizational absorptive capacity.

Read the analysis
Aerospace & Defense9 August 2026

Aerospace & Defense: Does the Record $45B Capex Lock In the Wrong Capacity?

AIA puts 2025 US aerospace & defense capex at $45B, up 13%. An economist's read: a record capex figure tells you the industry intends to invest and nothing about where — and "putty-clay" investment theory shows why that choice, once capital is installed, locks the sector in for a decade.

Read the analysis
Energy8 August 2026

The Oil & Gas AI Market: A $25B Projection Says Nothing About Who Will Capture the Value

Precedence Research projects the global upstream O&G AI software market from $7.64B (2026) to $25.24B (2034), a 14.2% CAGR. An economist's read: a market-size extrapolation is silent on competitive structure — and therefore on who, between specialized vendors and supermajors internalizing the capability, actually captures the margin.

Read the analysis
Aerospace & Defense7 August 2026

Aerospace & Defense: Why Commercial AI Is Accelerating While Military Decision AI Waits

Engine aftermarket revenue grew 20-40% in Q1 2026 on predictive-maintenance AI, while the Department of War's FY27 budget names trusted deployment — not model capability — as the real constraint on military autonomy. An economist's read: two AI speeds reflect two radically different verification-cost structures.

Read the analysis
AI Regulation6 August 2026

The AI Act and Critical Energy/O&G Infrastructure: The Asymmetric Bet of Waiting for Clarity

The AI Act's 2027 deferral, celebrated as blanket relief, does not explicitly name critical infrastructure or energy among the postponed categories. An economist's read: under this ambiguity, real-options logic argues for complying now rather than waiting.

Read the analysis
Aéro & Défense4 August 2026

Capital found defense AI. The constraint is that software doesn't melt titanium.

Defense-tech VC put $19.8B to work in one quarter of 2026 — more than the entire 2025 record — yet the binding constraint is the qualified physical base, not capital. An economist's read on where the returns actually accrue.

Read the analysis
Energy3 August 2026

AI's real bottleneck is no longer the chip — it's the firm electron, and it's being repriced

The scarce input in AI has moved from silicon to firm, 24/7 power. With data-centre demand doubling past 1,000 TWh while the US interconnection queue tops 1,500 GW and ~85–90 GW of needed new nuclear is <10% available, price theory is playing out: buyers pay 5–10× gas capex for guaranteed electrons. For industrial buyers the lesson is to optimise around secured power and proprietary data, not GPUs; for energy and O&G incumbents it is a demand-side windfall.

Read the analysis
AI Regulation2 August 2026

The EU AI Act deadline everyone braced for just moved — and deferral is not cancellation

As of today, 2 August 2026, the EU AI Act's transparency duties and GPAI enforcement are live — with fines up to €35M/7%, above the GDPR ceiling — but the Digital Omnibus deferred the heavy high-risk obligations to 2 December 2027. Read as an economist, deferral is not cancellation: it is a discount on preparation, not on the obligation. The sixteen-month window rewards regulated sectors already fluent in conformity and punishes those who treat it as a holiday.

Read the analysis
AI Economics1 August 2026

The AI capex bubble is the supplier's problem — not the buyer's

Hyperscalers will spend ~$660-690B on AI infrastructure in 2026 against ~$600-650B of revenue needed to justify it and only ~$50-150B earned — a 4-13x gap. Read as an economist, that bet sits on the supplier's balance sheet, not the buyer's. For regulated, capital-intensive sectors, ROI is decoupled from the frontier race: buy commoditised capability against data you already own, not the capex.

Read the analysis
Luxe & Cosmétique31 July 2026

Luxury: counterfeiting is a $467B tax on brand equity — and most brands are buying the wrong AI

Counterfeiting drains $467B a year and hits fashion hardest, with over half of some brands' resale footwear now fake. The economist's move isn't better fake-detection downstream — it's brand-owned provenance at the source, which doubles as EU Digital Product Passport compliance.

Read the analysis
Énergie30 July 2026

Energy: the grid's real shortage isn't megawatts — it's flexibility, already installed and idle

The IEA's Electricity 2026 quietly reframes the grid's binding constraint from generation to flexibility. Read as an economist: the world uses ~100 GW of demand response while single load categories run into the hundreds of GW. The scarce resource is the AI layer that turns loads you already own into dispatchable capacity.

Read the analysis
Aéro & Défense29 July 2026

Aerospace & defense: AI's first job is procurement, not the cockpit

85% of aerospace & defense leaders now prioritize AI for sourcing and procurement — 22 points above general industry — because the binding constraint on program growth is qualified, compliant supply, not model performance. Read as an economist, the highest-ROI AI project shortens supplier qualification and automates traceability, and the sector's compliance discipline is the very moat that makes it deployable.

Read the analysis
Energy & O&G28 July 2026

Upstream oil & gas: a $500bn AI prize, and one lock — and it is not the technology

Rystad Energy values the AI and digitalization opportunity in upstream oil & gas at ~$500bn through 2030 — and names deployment at scale, not technology, as the binding constraint. Read as an economist, that turns ROI from a model choice into an organizational one.

Read the analysis
AI Governance27 July 2026

The EU AI Act's two-speed calendar: one deadline holds, the rest slipped — don't misprice the slip

On 2 August 2026 the EU AI Act's transparency and GPAI-enforcement obligations go live — with fines up to €15M or 3% of turnover — while the Digital Omnibus pushes the high-risk deadlines out to 2027-2028. Read as an economist, that postponement is a mispriced option, not relief — and regulated sectors already own the discipline to exploit it.

Read the analysis
Luxury & Cosmetics26 July 2026

L'Oréal × NVIDIA: in beauty AI, the moat is proprietary data — not compute

L'Oréal's NVIDIA ALCHEMI deal promises ~100× faster formulation discovery. Read as an economist, the lesson for every specialty manufacturer is that the advantage sits in proprietary data no competitor can rent — not in the GPU.

Read the analysis
Économie de l'IA25 July 2026

The AI ROI paradox: output is up, returns are flat

93% of enterprises say AI improved production, only 43% say the return outpaced spend. The economist's read of the 2026 Domino / BARC report: the discriminant is not more AI budget but integrated governance — and Europe is falling behind.

Read the analysis
Aéro & Défense24 July 2026

Autonomy re-prices force: the Pentagon's swarm milestone, read by an economist

The Pentagon just put Shield AI's Hivemind autonomy inside the LUCAS strike drone — one operator, a whole swarm, no GPS. The capability headline hides a pricing event: an autonomous 'loyal wingman' costs a quarter to a third of an F-35, and the same edge-autonomy cost curve is bending across aerospace, energy and O&G asset operations.

Read the analysis
Energy23 July 2026

AI and energy: the data-centre power panic has the economics backwards

The headline says AI will strain the grid. The data says data centres account for less than 10% of total electricity-demand growth to 2030 — and that the scarce resource was never the megawatt.

Read the analysis
Aero & Defence22 July 2026

Aerospace & Defence: the AI market compounds at 13.5% a year — certification cadence does not

The AI-in-A&D market grows toward $55B by 2030. The binding constraint is not budget — it is the internal capacity to evaluate, version and re-qualify frontier models that ship almost monthly.

Read the analysis
Energy & O&G21 July 2026

The utilization gap: your plants use less than 8% of the data they already collect

The real bottleneck in O&G and heavy industry is not technological — it is utilization. The marginal cost of exploiting data you already own is near zero, which makes the ROI a decision measured in hours, not weeks.

Read the analysis
AI Sovereignty20 July 2026

AI sovereignty in Europe: the market has already answered

When private capital offers roughly 11x the public stake, the question of whether Europe needs sovereign AI is already settled. The real discriminator is data-sensitivity mapping, not model performance.

Read the analysis
Luxury & Cosmetics19 July 2026

AI in cosmetics and luxury: the business case that pays first isn't the glamour

The AI-in-beauty market is glamorous on the consumer side — virtual try-on, skin diagnostics — but for a French manufacturing SME the first profitable project is industrial vision on packaging lines, with payback under a year.

Read the analysis