The Art of Doing Financial Engineering

In the midst of every economic boom, there arises a commensurate increase in financial engineering. This phenomenon is no exception in the present context. Forecasts for AI infrastructure capex through 2030 span between $7.6 trillion and $8.0 trillion, notwithstanding the fact that bond markets, while extensive, are not infinite.

Analysts at KKR project that traditional US public investment-grade investors could contribute $1.7 trillion in financing; however, this still leaves a deficit of $6.3 trillion. They anticipate this funding gap to be filled through a combination of financing instruments: private investment-grade debt, asset-backed finance, infrastructure debt, structured solutions, and potentially novel financial structures.

This isn't a novel occurrence. AI capex could potentially surge to 3.63% of GDP by 2032, but the past offers a similar precedent. From 1870 to 1890, as rail infrastructure expanded at a rapid pace, associated capital expenditure constituted 2.24% of GDP. Richard White, in his book Railroaded, elucidates this scenario: "Investors proceeded from government bonds to government secured railroad bonds, to convertible bonds, to mortgage bonds vouched for by the same entities that issued the government bonds, to a myriad of financial instruments, and potentially into hazardous territory...

There were first and second mortgage bonds; there were mortgages on the main line and on the branches. There were land grant bonds and income bonds. There were bonds secured by anything and everything that investors deemed acceptable collateral. The bankers and railroads they represented extolled the security of these investments, yet they were essentially carnival barkers."

Although measuring financial engineering is inherently challenging, its fluctuations serve as an insightful indicator of the current economic cycle. Like other engineering disciplines, it can be wielded for beneficial or detrimental purposes. It tends to proliferate when conventional balance sheets cannot keep up with an investment surge.

Yet, the structures that sustain the boom may also obfuscate the accumulation of associated risks. One method to scrutinize it is through the perspective of the financial engineers who orchestrate it. Their mandate is to navigate the requirements of companies, investors, regulators, auditors, and credit rating agencies to create structures that apportion cash flows and risk among the entities most equipped and prepared to shoulder them. To gain insight into their operations, let's examine some of the structures engineered this year.

Case 1: Hyperion

In December 2024, Meta inaugurated a new data center complex in Richland Parish, Louisiana, marking the beginning of Mark Zuckerberg's foray into artificial intelligence. Spanning 2,250 acres, this site was intended to be a linchpin of this push. Zuckerberg heralded its potential, declaring it "capable of scaling up to 5GW over several years."

Although not as expansive as Manhattan, the scale of the project posed significant funding challenges. Meta had allocated $10 billion for initial costs, yet the magnitude of the project threatened to surpass its balance sheet. Consequently, it sought alternative financing arrangements. Morgan Stanley's financial engineers stepped in to fill this gap.