One failed drone venture tells a tale of SVB’s eccentric lending

Had it ever left the ground, the self-piloting camera drone that Antoine Balaresque dreamt of building when he was an undergraduate at Berkeley would have been a marvel of the modern world. The loan from Silicon Valley Bank that sustained his doomed venture arouses a different kind of wonder.

It is unlikely that many banks would have considered Lily Robotics for a loan in late 2015. The most visible achievement of Balaresque’s company, which he set up soon after graduating, had been to produce one of that year’s most-watched YouTube advertisements. In the video, a kayaker throws a Lily camera aloft before plunging down whitewater rapids; the device automatically pursues her, shooting cinematic footage as it goes.

Unfortunately, Lily never made a device capable of doing all that, according to a lawsuit that was later filed by the San Francisco district attorney. While the company did raise $14mn in venture capital to try to turn its vision into reality, it spent a significant portion of the cash without manufacturing a single drone.

As the self-described “financial partner of the innovation economy”, SVB specialised in lending to such hard cases. An arresting sentence in SVB’s final set of accounts, published weeks before the bank’s failure in March, declared that “many of our loans . . . are made to companies with modest or negative cash flows [and] no established record of profitable operations”. In December 2015, SVB agreed to lend Lily $4mn with interest of just 1 per cent above its “prime rate”. At the time, Lily was burning through $1mn a month. Repayment was due over four years.

As it turned out, unconventional lending practices do not even register among the causes of America’s second-biggest bank failure. Instead, SVB’s leadership blew themselves up by ploughing flighty customer cash into long-dated government bonds, which mechanically lose value when interest rates rise. Still, the banality of the bank’s demise should not preclude a reappraisal of its unusual $74bn loan book, which points to a kind of risk-taking that may have gone unnoticed elsewhere in the financial system.

By SVB’s own tally, 70 per cent of its loans consisted of “low credit loss lending” to vineyards, private equity funds and “innovation economy influencers”. The rest was more dicey. About 9 per cent of the total went to companies that the bank considered unlikely to be able to pay back what they owed unless they could find a buyer or raise new money. According to one academic estimate, three-quarters of venture capital-backed companies eventually fail.

This all sounds like an excellent way to lose money. Executives at First Citizens Bank, which bought the loan book in a fire sale organised by regulators, praised SVB’s underwriting practices and added that the bank’s business model was “unique and strong”. Even so, they gave themselves a large cushion. The $16.5bn discount that First Citizens negotiated at auction suggests that some bidders were hesitant.

It is tempting to dismiss SVB’s lending to risky start-ups as just another aberration by a bank that ultimately failed. Yet there are other possibilities.

One suggestion is that SVB may have been able to profit from venture lending where other banks could not. Some start-up veterans say privately that VC firms cared enough about their standing with SVB that they would use their own money to shield the bank from losses on start-ups. Shortly before Lily Robotics entered bankruptcy in 2017, SVB sold its loan to one of the company’s VC backers for an undisclosed price. It is unclear whether such deals were common, or whether this one reduced the losses on SVB’s loan. Regardless, no bank can afford to rely too heavily on the sympathy of its customers; after all, it was a stampede of SVB’s depositors that caused the bank to fall.

More tantalising is the possibility that SVB, while functioning as a bank, had found a way to share in the more lucrative financial returns of its VC clients. As a condition of its loans to start-ups, SVB typically received warrants, allowing it to participate in increasing equity value. If just a few clients became unicorns, those warrants could yield staggering profits.

In the aftermath of a $42bn bank run that wiped out one-quarter of SVB’s deposits, its unusual lending practices may seem like a harmless curiosity. Yet a lesson of past crises is that such anomalies deserve study.

Low interest rates make it difficult for regulated financial institutions to make money on safe but low-yielding assets. At the same time, economic growth and placid markets can make some bets seem less risky than they really are.

In Silicon Valley, the trauma of the pandemic was tempered by a period of extraordinary economic optimism. Long after SVB is gone, we may be reckoning with other consequences of that euphoria.

mark.vandevelde@ft.com