On a single Monday in July, three deals crossed the wire that tell you more about where partnership strategy is heading than a year of conference keynotes.
Prysmian, the world’s largest cable maker, signed a supply agreement with Molex worth up to €5.5 billion over ten years — optical fiber for AI data centers. Honeywell Aerospace announced the largest equipment win in its history: avionics and power systems for 810 of IndiGo’s incoming Airbus A320neos, aftermarket support attached. And Aeromexico became launch customer for Honeywell’s runway-safety system across more than 100 Boeing jets, ahead of certification that won’t even begin until late this year.
Three sectors. Cables, avionics, safety tech. One pattern — and one detail that gives the game away.
Molex paid Prysmian €550 million in advance. Before a single cable shipped.
The advance payment is the story
A supply contract with a half-billion-euro prepayment is not procurement. It is capacity finance. Molex is not buying cable; it is underwriting the construction of the factories that will make the cable — funding Prysmian to double its US fiber capacity, extend production down to the glass preform stage, and add hundreds of manufacturing jobs. The buyer has become the banker.
Why would a customer finance its supplier’s capex? Because AI demand has inverted the scarcity. For two decades, industrial buyers held the leverage: capacity was abundant, suppliers competed on price, and contracts were annual knife-fights over basis points. Data-center buildout has flipped that. The scarce asset is no longer the customer’s order book — it is the supplier’s future capacity. And when capacity is the scarce asset, the deal architecture changes shape entirely.
These are capex partnerships: long-duration agreements in which the buyer’s balance sheet and the supplier’s production roadmap are deliberately fused. The contract’s real subject is not this year’s delivery. It is a shared bet on a ten-year demand curve.
Aviation is running the same play
The Honeywell deals look different on the surface — no prepayment headlines — but the architecture rhymes. The IndiGo agreement bundles auxiliary power units, weather radar, collision avoidance, and flight management systems across 810 aircraft with aftermarket support attached. The equipment sale is the entry ticket; the economics live in a service relationship that will outlast the executives who signed it. Honeywell has supplied IndiGo since 2015 — this deal converts a vendor relationship into something closer to a shared fleet-economics venture.
Aeromexico goes further into partnership territory: it is adopting a runway-safety system that regulators have not yet certified. A launch customer for pre-certification technology is not a buyer. It is a development partner absorbing regulatory timing risk in exchange for first-mover advantage — a risk allocation you negotiate, not a product you purchase.
What the book calls economic engineering
In The New Rules of Partnerships, the argument is that most partnership failures are designed in at signing — and that the antidote is economic engineering before governance. Get the incentive structure right and governance becomes light-touch; get it wrong and no steering committee can save you.
The Prysmian–Molex structure is economic engineering in its purest industrial form. A ten-year commitment only survives if both sides’ upside is locked to the same demand curve — and the advance payment does exactly that. Prysmian cannot take the €550 million and under-invest, because the capacity it builds is contractually spoken for. Molex cannot walk away from a decade of committed volume, because it has already paid for the factory. Each side has posted a bond against its own future defection. That is incentive alignment made physical — poured into concrete and drawn into glass.
Compare that with the classic failure modes: incentive inversion, where partners profit from each other’s weakness; the governance vacuum, where nobody owns the joint outcome; motivational decay, where deal sponsors move on and the partnership drifts. A prepaid, capacity-linked, decade-long structure is engineered against all three. The commitment mechanism is the governance.
The ecosystem logic underneath
Step back far enough and the three deals resolve into a single picture. AI demand does not flow through one company; it pulls through an ecosystem — chips to data centers to fiber to power to the aircraft moving the engineers who build it all. No firm in that chain can capture the demand alone, and no firm can scale for it alone. The companies moving fastest are not asking “what should we buy?” They are asking “whose capacity do we need to exist, and what will we commit to make sure it does?”
That is the thesis of the book operating at industrial scale: ecosystems outcompete individual firms. The winners of the AI buildout will not be determined solely by who has the best technology. They will be determined by who engineered the most durable commitments across their supply ecosystem — while their competitors were still running annual RFPs.
Five questions for your own deal pipeline
For operators looking at their next major supply or platform agreement, the July 20 deals suggest a diagnostic. Is the scarce asset in your category shifting from demand to capacity? Would prepayment or co-investment buy you priority that pricing never will? Does your agreement’s duration match the capex cycle it depends on? Is aftermarket or service economics where the real value sits? And if your key supplier’s roadmap failed, what would it cost you — versus the cost of underwriting it now?
Companies that answer those questions deliberately are building capex partnerships. Companies that don’t will discover, sometime around 2028, that the capacity they assumed would be available was committed to someone else — years ago, on a Monday in July.
Randy McGraw is the founder of M2 Ventures and author of “The New Rules of Partnerships.” Over 30 years in APAC he has closed $2.3B in partnership and JV value, including Amazon Prime’s two largest telco bundling deals globally and the HOOQ streaming JV at SingTel. m2-ventures.com