The Second Supplier Is Not Waste. It Is a Premium You Already Owe.

Writing for the Forbes Tech Council, Dr. Aditya V Kashyap argues that modern management has spent 50 years waging a largely successful war on slack: inventories cut to days of demand, vendor lists consolidated, infrastructure centralised, capital pushed toward its highest expected return ^[1]. The consequence for operators is not moral but budgetary, because the items that campaign removed are the same items that determine whether a firm can still act once its assumptions break ^[2].

The framing is borrowed from finance, and it survives the borrowing. An option has value because it confers the right, not the obligation, to act after uncertainty resolves, and the premium is paid up front for the freedom to decide later ^[3]. On that reading, optionality is the preservation of decision rights: a second supplier preserves the right to redirect supply, unused borrowing capacity preserves the right to liquidity when conditions turn, and a modular stack preserves the right to swap a component after an assumption fails ^[4]. Redundancy, diversification, modularity and liquidity are the mechanisms, not the option itself, and none of them produces much under normal conditions, which is precisely why conventional analysis marks them down ^[5].

This is not an argument against efficiency. Efficiency maximises performance conditional on assumptions holding; optionality preserves performance when they fail ^[6]. The failure mode worth fearing is not tolerated inefficiency but the decision that closes off future choices quickly and permanently: the proprietary platform that makes migration prohibitive, or the capability that takes a decade to rebuild once it has been disbanded ^[7].

Architecture is where the trade becomes legible, because architecture sets the cost of changing your mind ^[8]. Tight coupling is cheaper to build and faster to run, while modularity, open standards, data portability and clean API boundaries carry a premium paid in integration overhead and forgone optimisation ^[8]. Kashyap is explicit that this does not mean flexibility everywhere: duplicating every workload across clouds usually costs more than the option is worth, so the discipline is asking, interface by interface, whether the switching cost being created is one the organisation could afford to pay ^[9].

Two data points give the argument teeth. A 2021 study by the Semiconductor Industry Association and Boston Consulting Group found that as of 2019, 92% of global fabrication capacity for advanced logic chips below 10 nanometres sat in Taiwan ^[10], leaving 8% of that capacity everywhere else ^[14]. Most of the individual decisions behind that concentration were economically rational, and the public money now going into geographic redundancy is, on Kashyap's account, the deferred price of options nobody bought ^[11]. The second point is that these options do get exercised: in March 2020, Federal Reserve economists documented commercial and industrial lending rising by nearly $480 billion in a single month, the largest monthly increase in the history of the series, chiefly because firms drew down revolving credit lines that had sat unused for years as a fee paid for nothing visible ^[12][13].

What to watch inside your own numbers: whether the dual-source line, the idle headroom and the modular boundary appear in planning as premiums with a stated payoff state, or as variances to be closed by the next cost review ^[5][9]. The switching costs a firm cannot afford to pay are the ones it will discover only when it needs to change ^[7].