In late December 2017, Anthony Noto slipped into a San Francisco law firm’s office for a confidential meeting. At the time, he was Twitter’s chief operating officer and looking to level-up to become CEO of San Francisco’s SoFi , a fast-growing student loan startup. Noto was there to pitch the company’s board of directors in the wake of its founder, Mike Cagney , resigning in disgrace after having a relationship with an employee.

The West Point graduate and former Goldman Sachs banker launched into a PowerPoint laying out his plan for SoFi, including what he would do in his first 45, 60, 90 and 180 days. The tech startup, he argued, should act like a bank, offer a supermarket of financial services and help customers “from high school graduation to [the] grave.” Success, he said, meant growing SoFi’s market value to more than $20 billion.

Noto also came with baggage: Earlier that year, shareholders had accused him and Twitter CEO Dick Costolo of concealing a decline in the company’s daily active users, one of management’s primary engagement metrics. The lawsuit was later settled for $810 million, with Twitter, Noto and Costolo denying wrongdoing.

Yet that day in December, Noto’s pitch wowed the room. “It was the most thorough thing I had ever seen,” says Pete Hartigan, a former venture capitalist and SoFi board member. Noto was named SoFi’s CEO in February 2018, and over the past eight years, the 58-year-old has followed a classic banking playbook: sell more financial services to more customers and become central to their financial lives. He has expanded SoFi from three product categories to twelve, pushing beyond student and personal loans into budgeting tips, bank accounts, credit cards, crypto investing, home equity lines of credit, small business loans and technology services.

His results have been impressive. When he took over in 2018, the startup had $240 million in revenues, 650,000 customers and a valuation of $4 billion. By the end of 2025, the stock market was valuing SoFi at nearly $40 billion, it had 13.7 million customers and reached revenues of $3.6 billion. Among the nation’s banks, SoFi now ranks as the 50th-largest, with about $50 billion in assets, according to the Federal Reserve. Noto has been richly rewarded: He earned $30 million in 2025 and $103 million in 2021, the year SoFi went public. Forbes estimates his net worth at about $400 million.

But in recent months, investors have begun to question SoFi’s success story. Since November 2025, its stock has fallen by nearly 50%, reflecting concerns about AI-driven job displacement, turmoil in private credit markets and the company’s growth prospects. SoFi had been priced as though it were a high-growth tech company with a forward price-earnings ratio of nearly 70 times, while major banks like JPMorgan Chase tended to trade at 15 times’ earnings or less. Wall Street was suddenly re-evaluating the fintech darling.

Then, just as SoFi’s stock was reeling from its Wall Street drubbing, short seller Muddy Waters came out with a scathing, 28-page report in March, essentially accusing the company of cooking its books in order to inflate its earnings, calling it a “financial engineering treadmill.” Mainstream stock analysts had already turned against the company, with some accusing SoFi of “aggressive” accounting practices, a claim the company denies.

Today, Wall Street analysts rate the one-time fintech darling as a “hold” on average, according to FactSet, putting it in the doghouse among brand-name financial technology peers like Affirm and Robinhood .

“Investors are starting to realize SoFi is really just a lending business,” says KBW analyst Tim Switzer.

F ormerly known as Social Finance, SoFi began life at Stanford Business School in 2011 , in a famed class known as “Startup Garage” where Mike Cagney and his cofounders dreamed up a peer-to-peer student loan specialist organized around top 50 elite colleges. Federal loans had always carried the same interest rate regardless of a borrower’s credit profile. Cagney realized he could offer lower refinancing rates to alumni of top schools. He also figured that prosperous alumni from places like Harvard, MIT and Stanford would happily lend to borrowers from their alma maters, earning good returns at lower risk of default.

One of Cagney’s standout skills as a founder is that he’s a “gifted communicator and synthesizer,” Pete Hartigan says. “Back then, doing fintech was not normal, and doing student loans was definitely not normal.” According to PitchBook, from 2011 through 2015, Cagney raised $1.4 billion in equity financing for SoFi, a stunning sum in that era for such a short time period and young company.

Beyond equity capital, SoFi needed a mountain of debt financing to fund its loans. For that, former Wells Fargo banker Cagney had a secret weapon: Nino Fanlo, an old colleague who had become a private credit executive at KKR before joining SoFi as CFO in 2012. Fanlo called his friends in finance and got many of them to buy SoFi’s loans.

SoFi’s success was short-lived. A scandal involving a relationship between Cagney–who was married–and an employee, plus allegations of inappropriate in-office behavior by CFO Fanlo threw the executive suite into turmoil. (Fanlo says media depictions were overblown.) By the fall of 2017, both Cagney and Fanlo were out, and SoFi, which depended on Wall Street’s confidence for funding, was in search of new leadership.

Enter Noto, who, like Cagney, had solid Wall Street credentials, but, unlike the cofounder, was known as an operator rather than an innovator. Noto grew up in Poughkeepsie, New York. His parents divorced when he was three, and his mother was a hairdresser. His family was on welfare and food stamps for a period of time. At West Point, Noto became the top-ranked mechanical engineer in his class and a star linebacker. At Goldman, he was named the number-one analyst for Internet stocks by Institutional Investor for five years in a row, from 2003 to 2007. He later did a stint as the NFL’s CFO.

At SoFi, the former banker worked quickly to restore order. He released a new set of corporate values, including “embrace diversity” and “do the right thing.” He brought in an entirely new executive team, pulled the company away from its Silicon Valley bro culture and focused on unit economics, giving division heads responsibility for their own unit’s profits and losses.

Noto went to work transforming SoFi from student loan specialist into a technology-first digital bank with a diverse mix of offerings. He also understood the value of branding. In 2019, SoFi agreed to pay more than $25 million annually in a 20-year deal to rename the Los Angeles Rams’ home field SoFi Stadium.

Then the Covid-19 pandemic struck, and the government paused student loan repayments and suspended interest accruals. This caused student loan refinancing demand to crater. Noto’s diversification into higher-margin personal loans eventually helped the company survive the lending drought. In 2021, Noto raised $2.4 billion and took the company public as SoFi Technologies through a SPAC, securing a bank charter about six months later. The bank charter allowed the company to take in deposits and offer high-yield savings accounts, credit cards and mortgages.

Over the next few years as interest rates rose, Noto rode out the fintech winter and launched new products, including home equity loans and small business loans, at a rapid clip. But beneath the rapid expansion and optimism, a more complicated question emerged on Wall Street: whether SoFi’s meteoric growth was being propelled as much by financial engineering as by financial innovation.

W hen it comes to the vast accounting rules governing publicly traded U.S. businesses, any seasoned auditor or C-suite executive will tell you that they leave enormous room for interpretation. “You can drive a truck through accounting rules,” quipped JPMorgan Chase CEO Jamie Dimon on a podcast last year.

SoFi has exploited this flexibility–a half-dozen people we spoke with, including equity research analysts and fintech executives, call SoFi’s accounting practices aggressive, a characterization the company denies.

Earlier this year, its bookkeeping drew the interest of well-known Austin, Texas-based short seller Muddy Waters. On March 17, in a 12,000-word diatribe entitled “SOFI: A Financial Engineering Treadmill Leaving Management Fat, Shareholders the Biggest Loser,” Muddy Waters alleged that SoFi used “Enron-esque” accounting tricks.

It made an array of detailed accusations, including that SoFi was under-reporting how many customers weren’t paying back their loans and that it was providing hidden financing to its loan investors. Experts debated the claims’ accuracy. The stock barely budged, which analysts concluded was because investors were already aware of the questions raised by the report and because SoFi’s share price had already fallen steeply earlier in the year.

In a statement released a few days later, SoFi called Muddy Waters’ analysis inaccurate, saying its claims “demonstrate a fundamental lack of understanding of our financial statements and business.” The statement continued, “SoFi maintains strong confidence in the integrity of our financial reporting.”

The dispute resurfaced a debate about SoFi’s books that some analysts had flagged as early as 2023. To value its loans, SoFi uses fair-value accounting–a complex approach permitted under U.S. Generally Accepted Accounting Principles (GAAP) and also used by SoFi competitor LendingClub–where estimates can be used when liquid markets don’t exist. One of the key inputs into a fair-value assessment is a loan’s discount rate, which represents the interest rate debt buyers would realistically pay to purchase a loan and take on the associated credit risks, such as borrowers paying off their loans early or not making their payments.

For SoFi’s personal loans, which its customers often use for consolidating credit card debt or home improvement projects, the fintech applies an unusually low discount rate of 4.6%. That’s within one percentage point of the so-called risk-free rate, or the return expected from high-quality, short-term government securities like U.S. Treasurys. LendingClub, whose personal-loan borrowers have lower credit scores on average than SoFi’s customers (720 FICO compared with 745 for SoFi) and carry higher credit risk, sets its discount rate at 7.2%.

Why would a lender want to use a low discount rate? Because it boosts the value of the loan and the company’s profits when portfolios are marked to market. The lower the rate, the higher the present value of the loan’s expected cash flows—and the higher its estimated fair value. For lenders that use fair-value accounting, this matters: When they originate a loan, they can record more upfront income if the discount rate is lower, because it implies investors will pay more for the loan.

At the end of 2025, SoFi had $20.2 billion in unsecured personal loans outstanding, making it the top digital consumer lender in the nation. Forbes estimates that valuing that loan book with a more conservative 5.5% discount rate, instead of the 4.6% rate SoFi uses, would have reduced its cumulative pretax profit by roughly $275 million over the life of its personal loan business. In 2025, the company had $526 million in pretax profits. It had $233 million in 2024 and negative pretax income for each of the three years prior.

Industry insiders say that, despite SoFi’s good track record of keeping defaults low, it’s unrealistic that institutional debt investors would buy its loans at the rock-bottom discount rates the company uses in its fair-value calculations. They say that for super-prime consumer credit, investors typically require at least 1.5 to two percentage points above the risk-free rate to feel comfortable with the risks they’re taking. One expert tells us, “I haven’t found a single person who invests in this asset class and believes that SoFi’s loans transact at yields that match the discount rate it’s using.”

A SoFi spokesperson insists the company’s discount rate is appropriate and says any claim that its loans don’t transact at their stated yields is “patently false.” If SoFi’s estimates are too aggressive, he says, future revenue and net income will suffer because upfront benefits are later expensed. He says SoFi’s numbers are scrutinized by internal teams, its board audit committee, third-party auditors and regulators. He also notes that SoFi’s cash interest income since 2024 has been more than double its non-cash, fair-value income over the same time period.

In a recent investor presentation, SoFi reported that it sold bundles of whole, unsecuritized personal loans in 2025 at 105.8% to 106.5% of face value, implying premiums of roughly 6%. Yet those premiums are hard to interpret for multiple reasons. First, it’s unclear how representative the loan sales are because the transactions are relatively small. In the third quarter of 2025, the personal loan sale SoFi cited totaled $176 million, less than 1% of its personal loan book. In the fourth quarter, the cited sale was $100 million, and in the first quarter of 2026, it completed no whole loan sales.

Second, SoFi declines to disclose what the age of the loans were when they sold them or what their values were on the first day they were originated. These are critical factors in determining their value. SoFi declined to answer Forbes’ specific questions on the topic, which analysts say is par for the course with SoFi. Giuliano Bologna, an equity research analyst at Compass Point, wrote in a research brief in March 2026 that, for over two years, SoFi hadn’t answered questions he had asked and had continuously postponed phone calls with him.

In fact, over the past four quarters, SoFi’s management hasn’t taken a single earnings-call question from an analyst with a sell or underweight rating on the stock. A spokesperson says SoFi engages with analysts outside earnings calls and participated in more than 200 one-on-one calls, conferences and events with the research community in 2025.

Mark DeVries, an analyst at Deutsche Bank, says he doesn’t question the integrity of SoFi’s books but that the complexity of its fair-value accounting invites scrutiny and “makes it easy to doubt them.”

Another SoFi accounting move that has raised eyebrows is how it treats marketing costs, an issue Muddy Waters flagged in its short report. Instead of recording all of its spending as an immediate expense, SoFi capitalizes it over time—as much as five-plus years, according to Compass Point. The effect is simple: it boosts profits in the short term. Some brokerages, including Schwab, capitalize certain marketing costs. But the practice is extremely rare among fintechs and traditional banks, in part because accounting rules restrict it for some types of consumer accounts, including deposit and lending accounts.

SoFi chief financial officer Chris Lapointe has said the company complies with the relevant rules because it capitalizes marketing expenses only for revenue tied to debit cards and brokerage services. Yet when Forbes asked SoFi to provide examples of the associated marketing materials to better understand how the ads fit within the accounting rules, a spokesperson declined.

The impact of this capitalization has been significant. In 2024 alone, SoFi capitalized $177 million in marketing expenses; without that treatment, its net income before taxes would have been 76% lower. The company has also capitalized more in marketing costs than those businesses have generated in related revenue. In the years leading up to the end of 2025, SoFi capitalized $408 million in cumulative marketing expenses, more than the $382 million in cumulative brokerage and debit-card interchange revenue it generated from January 2022 through December 2025.

Needless to say, SoFi’s accounting tactics and lofty price-earnings multiple, still more than double that of most banks, have been a turnoff for asset managers. Institutional investors hold just 54% of SoFi’s stock, according to FactSet, well below the levels at other major fintechs and at three-quarters of the 100 largest U.S. banks by assets.

Have a story tip? Contact Jeff Kauflin at jkauflin@forbes.com or on Signal at jeff.273.

A ggressive accounting aside, the 15-year-old company, born in Silicon Valley, is being forced to face the harsh Wall Street reality that at its core, it is little more than a consumer bank, albeit one with no bricks and mortar.

In the first quarter of 2026, lending-based services made up 88% of SoFi’s $1.1 billion in net revenues (see chart). Noto has said SoFi is “no longer just a lending company” and has tried to amp up SoFi’s non-lending tech offerings. In 2020, SoFi acquired Galileo , a back-end banking technology company, for $1.2 billion. Two years later, he took over Miami-based Technisys, a core-banking tech provider, for $915 million.

Both purchases were designed to help SoFi speed up its own product development and sell technology services to other companies, but sales of these tech services have been slowing. The segment’s revenue fell by $28 million in the first quarter of 2026 compared with the year prior after losing Chime as a major customer. Excluding the loss of Chime, it grew just 12%. In April, after SoFi announced first quarter earnings, its stock fell 15% .

SoFi’s sagging stock may be having repercussions beyond Wall Street. Last year, according to people familiar with the matter, Noto made an unsolicited proposal to buy Column, a fast-growing San Francisco fintech led by Plaid cofounder William Hockey . Like Galileo and Technisys, Column is a tech infrastructure firm that enables other fintechs to offer banking services. Noto discussed paying $6 billion for Column, and while the two sides disagree over the exact details, the merger would have strengthened SoFi’s attempt to diversify away from lending and was to be financed by SoFi’s then-inflated stock.

Noto and Hockey met several times to discuss a potential deal, according to sources, but the talks ended earlier this year, just as SoFi’s stock was in a free-fall.