Summary

Blankfein’s core risk-management call is that investors must take risk while planning and buying cheap mitigants in advance. Forecasting matters less than asking what could happen, what the portfolio would do, and which protections are available before “the hurricane is coming.” Thorough contingency planning lets a firm move so quickly that it appears prescient: “I want everybody to be called for a false start.”
Finance will adopt AI aggressively, but its near-zero error budget produces a different adoption curve from Silicon Valley’s. Goldman kept the trusted system running while testing its replacement, sometimes needing “50 times” and perfection on “the last 49” before switching. Technology therefore increased costs initially; regulated institutions could not rely on apologies after a rollout failed.
The underappreciated AI danger is opaque, massively leveraged execution rather than machine supremacy. Software could execute 70,000 transactions without the human intuition or visible reasoning trail that once stopped a trading room cold. Regulation might need to slow deployment “not because it’s smarter than us and it’s going to turn us into pets,” but because institutions cannot yet test whether outputs are right.
AI is “going to be very, very important,” yet that does not make every model or company a winner. Blankfein sees genuine conviction in founder-dominated hyperscalers risking their own wealth and egos, but conviction is not correctness. He suggests the world might need four large language models rather than 10, with two becoming very large winners and the field possibly reducing to two.
Goldman preserved partnership behavior after its IPO by aligning people to the whole firm, then reshaped its earnings for public-market math. Glass-Steagall’s repeal made a larger balance sheet necessary, but shareholders valued consistency: “In a private company, you care about the E; in a public company, you care about P/E.” Moving principal risk into funds converted “100-cent dollars” into lower-risk “20-cent dollars,” requiring more volume but supporting higher P/E and return on equity.
Goldman’s financial-crisis edge came from forcing marks into reality before reality forced the firm’s hand. A separate valuation group told traders to “go out and sell something—sell a fraction,” exposing vanished bids in purportedly AAA assets and embedding losses before positions had to be sold. Goldman also fully hedged AIG exposure, demanded collateral, and honored commitments such as Chrysler’s—though “not for more” and “not sooner” than agreed.
Systemically important AI companies should establish public legitimacy before backlash arrives. Goldman learned that it had become “too important, too influential, too big to be anonymous,” yet had no consumer relationship anchoring its reputation when the crisis hit. Blankfein’s advice to leaders at OpenAI, Anthropic and similar institutions is to explain their economic function early, because being modest and invisible becomes a liability when the public decides something went wrong.

Deep dive

1. Crisis leadership begins by refusing to submit to chaos

During an active-shooter scare, Blankfein watched armed security arrive, was advised to get down under the desk, then asked a nearby guest, “Are you going to finish your salad?” The joke was not hunger or bravado; he instinctively tries to be “disarming” in moments of crisis.
His “normal resting state is to not be resting,” but crises make events seem to slow down. After encountering a supposed “crisis of the century” roughly every four or five years, his operating priority became simple: get people to do their jobs, keep them from freezing, and “don’t submit to the chaos.”
Résumés and physical bearing proved unreliable predictors of performance. During the financial crisis, a rodeo-riding “real man’s man” struggled, while people who looked unable to climb a flight of stairs excelled. In the board-selection discussion, the advice is to favor someone demonstrably tested by crisis over someone who merely looks composed.

2. Low expectations and an accidental acquisition opened Goldman’s door

Blankfein grew up in NYCHA public housing in a “two-fare zone,” where earning more than roughly $90 a week disqualified a family from his building. Manhattan, though visible, “might as well have been 5,000 miles away”; before college, he recalls going about three times, including once for his Harvard interview.
He calls low expectations an “advantage” because he avoided their psychological burden, while conceding he knew little of the wider world. At a failing high school, “I don’t think I’d read a book”; his verbal scores were low, his math score about 790, and his ambition extended only to attending college out of town.
After law school and four or five years practicing, Goldman rejected his application. His sole offer came from J. Aron, a small commodity-trading firm that hired him in precious-metals sales and was then acquired by Goldman. The deal resembled Columbus seeking the Indies and finding America: amid the inflation-era push into commodities, Goldman unexpectedly acquired an entrepreneurial, “streety” culture where driving for a trader had once been a prized entry-level job.

3. Investors must alternate between taking risk and restraining it

Blankfein’s foundational split applies to any investor: make money by taking risk, then “bifurcate yourself” into a risk manager asking whether the portfolio is diversified, overly committed, or poorly managed. “You have to do both”—neither risk avoidance nor unconstrained conviction fulfills the mandate.
The management paradox is asymmetric. Restraining eager risk-takers may be the harder challenge, but it occupies only about one-third of the cycle; much of the time, leaders must exhort or even shame recently singed investors into deploying again because “we’re paid to put out money in the right place.”
His own wiring helps him “find the cloud around any silver lining,” yet he also discovered an appetite for remaining inside risky situations without shrinking. That combination matters: nervousness generates downside questions, while tolerance for uncertainty prevents risk management from degenerating into permanent inactivity.
Contingency meetings should bracket probability and ask, “What will you do if it does happen?” Cheap protection is available in winter; insurance becomes expensive when the hurricane is approaching oceanfront property. The exercise also sensitizes people to triggers, turning preparedness into apparent prediction.

4. Good judgment reconstructs the fog rather than rewarding hindsight

Blankfein never replied “I already know” when junior employees raised concerns or opportunities. Listening to redundant reports revealed information about both the event and the messenger, while removing an excuse for future self-censorship: nobody could assume the hierarchy had already carried the warning upward.
Losses require distinguishing stupidity from error. Smart people are often wrong, but managers let “after-acquired information seep into their judgment” and retroactively treat uncertainty as obvious. Evaluation must reconstruct what was knowable “in the fog,” because “none of us know the future” and most people cannot even sort out the present.
His answer to confident pundits is: “If you were so prescient, tell me what happens next.” Once the present becomes the past, “everybody’s a genius”; risk management is therefore less forecasting than rehearsing multiple outcomes and reacting to the starting gun faster than others.

5. Finance adopts technology early but only after parallel proof

Financial markets have long rewarded technological advantage on a winner-take-all basis. If an execution computer sat half a block closer to an exchange, milliseconds could determine who captured the entire bid or offer while everyone else was “left looking at your dust.”
Regulated finance could not emulate companies that rolled out mistakes and apologized. Blankfein cites Robinhood’s early claim that certain accounts were government-insured when they were not; Goldman instead ran the proven and experimental systems simultaneously, sometimes needing 50 runs and perfection on “the last 49.” New technology initially augmented cost, then improved efficiency as the firm crossed from “one lily pad to another.”
SecDB embodied the payoff from durable architecture: its modular risk framework remained adaptable while rival systems were rigid, leaving a core from a system roughly 25 or 30 years old still in use. Blankfein compares it to his 40-year-old HP 12C calculator, whose battery lasted about 22 years and whose design still looks current.

6. Partnership culture survived the IPO by preserving ownership behavior

A legal partnership makes senior colleagues co-owners rather than subordinates: they care about the entire enterprise, expect extensive information and influence, and can place personal fortunes—including, historically, their homes—behind firmwide decisions. That unlimited liability “focuses your attention” on risk more effectively than investing only clients’ capital.
Major moves were socialized in advance, which could slow or table a preferred decision and gave owners influence. Blankfein says the process enlisted support from otherwise neutral colleagues and honored their status as co-owners; Goldman’s alumni office still serves people decades after departure, reinforcing why former employees continue to self-identify with the firm.
Going public became necessary after Glass-Steagall’s repeal allowed commercial lenders to finance the advice they provided. If J.P. Morgan could become an adviser, Goldman needed to become a lender with a larger balance sheet—something impermanent partnership capital could not support. The legal conversion happened instantly; culturally, Blankfein says it took about 25 years.
Public ownership changed the objective from earnings to valuation: “In a private company, you care about the E; in a public company, you care about P/E.” Shifting principal investments into funds meant earning “20-cent dollars” rather than “100-cent dollars,” but with lower risk, higher return on equity and a potentially higher multiple—provided Goldman did more business.

7. Institutions compound when powerful individuals defer to the platform

Blankfein wanted to be “not so much liked as appreciated”—the leader who made people better, not the commanding officer who merely juggled or told jokes. His test for newly promoted managers was intimate: employees discuss their boss at home every night, so “what do you want them saying about you?”
Whole-firm compensation and partnership elections taught bankers to resolve conflicts collectively, even when several teams wanted opposing sides of one transaction. The metaphor is not one 800-pound gorilla but 20; nineteen must periodically say, “Excuse me, after you.” Pay can reward exceptional performance, but management must “mute the effects of the cycle” enough to sustain cooperation.
Principal investing also let Goldman approach clients as peers rather than “supplicants looking for business.” That brought understanding and swagger, while partnership culture kept talented investors attached through cycles when gains tempted them to leave and losses tempted the firm to disconnect from them.
The sharpest example of ignoring the organization chart came when titleless Blankfein proposed an S&P 500 cash-and-carry structure for Middle Eastern clients who could earn investment returns but not interest. He approached Bob Rubin directly, equity traders were assigned to help, and the first order was $100 million—by far the biggest trade ever.

8. Mark-to-market turned accounting discipline into crisis detection

Haber’s pushback was that a crisis centered in private equity might have been harder for Goldman; Blankfein conceded it would, because illiquid assets are difficult to mark. Goldman nevertheless empowered a separate valuation group—partners paid well to oppose traders—and sided with that group in disputes unless an investor could prove a different price by selling part of the position.
That process exposed purportedly AAA securities whose bids vanished and then came back much lower. Blankfein personally suspected an opportunity to accumulate them, but belief could not override the market: marks kept falling until a sale was possible. Because losses were already embedded in books, subsequent disposals became easier.
AIG illustrated pre-committed protection. Goldman had fully hedged its exposure with credit protection and demanded a collateral agreement despite AIG’s AAA rating; Haber recalled that perhaps only “five or seven” companies had the temerity to ask. Goldman would not otherwise have transacted.
Relationships constrained crisis behavior as much as contracts. Blankfein promised Chrysler its committed financing, but “not for more” and “not sooner” than agreed. Today’s junior cohort would run important institutions 20, 30 or 35 years later; crisis-era grudges and goodwill are correspondingly “sticky.”

9. Important institutions cannot wait for crisis to explain themselves

Goldman’s wholesale model left it without branches, checking accounts or mortgages connecting it to the public; its PR department had historically kept its name out of newspapers. After Lehman and Bear Stearns disappeared and commercial banks lost amounts such as $50 billion, Goldman remained visible and successful—“too important, too influential, too big to be anonymous”—so the official sector filled its reputational vacuum.
Blankfein’s advice to AI leaders is to explain their public function before defensive communications become necessary. Goldman linked capital with businesses and took risks such as bringing Tesla public when companies were generally expected to be profitable first. “Being modest and understated carries a lot of disadvantages”; the moment people are trying to “kill you” is a poor time to begin making friends.

10. AI will matter enormously, but investability still turns on reliability

Asked about potentially enormous SpaceX, OpenAI and Anthropic IPOs, Blankfein refuses a cycle forecast. AI could be as significant as electrification or the internet—or bigger—but “I don’t think anybody knows”; meanwhile, someone in a basement could be building “OpenAI 7” unnoticed, creating upside surprise alongside overenthusiasm.
Founder-dominated hyperscalers are committing their own money and egos, signaling deeply held conviction without guaranteeing correctness. The world may not need 10 large language models: maybe it needs four, with two becoming very large winners, two getting by, and the field possibly reducing to two. A tech-bubble-like washout is also left open, but Amazon itself once looked extravagantly speculative.
Reliability separates approximation from institutional use: “If you’re in the business of horseshoes or throwing hand grenades, you don’t have to be precise.” Google supplied a bibliography users could check; large language models can obscure their reasoning. A trading room once stopped when someone quoted a wrong price, whereas hidden software can execute 70,000 transactions.
Technological leverage scales tail consequences. Blankfein contrasts Bhopal’s single-digit-thousands death toll with Fukushima’s potential, had the wind shifted, to affect tens of millions. Regulation may rightly slow systems because outputs cannot be tested—not because AI will make humans “pets.” Yet the knowledge cannot be unlearned, so debating whether progress should happen wastes time needed to manage it.

11. Automation makes historical range more valuable, not less

Blankfein rejects mournfulness about AI-enabled productivity. More than half the country once worked in agriculture; today it is a single-digit percentage, and people found other work. Greater wealth might enable a three-day week, six-hour days and afternoons spent as poets, hunters or fishermen: “I’m for all this stuff.”
With deference to Peter Thiel’s success, his advice to young people is to become “complete people.” Humanities, history and varied activities build appreciation, commercial resilience and the kind of interesting personality colleagues and investors want to engage; extreme early specialization may pay in the “first game” while narrowing the rest of life.
History corrects present-tense catastrophism. Blankfein invokes the Civil War, late-1960s campus shootings and political violence, the draft exodus to Canada, Soviet tanks entering Czechoslovakia in 1968, and DEFCON 2 during the Cuban Missile Crisis. Even amid what he calls a regional war in Iran, “knowing that something has been done” should provide confidence it can be done again.
Today’s dominant geography and skill may not persist: professionals once rushed to learn Japanese, Blankfein and his predecessor spent substantial time going to China, and “Silicon Valley” once meant Route 128 around Harvard and MIT rather than Stanford. Longer lives make the rush stranger; Blankfein does not believe productive years end at 24, and argues that broad foundations make later reinvention easier.

How We Grew Koch Inc. to $150 Billion Without Going Public: Charles & Chase Koch 31. “26Q1”: Has the China-U.S. Foundation-Model Gap Widened or Narrowed Over the Past Year? — Hugging Face | 王铁震