Who Is Charles Liang and How Super Micro's Founder Bet $39 Billion on AI

Charles Liang watched his company lose roughly a third of its market value in one session on June 10, 2026. Super Micro Computer announced an equity raise of about $7 billion to fund a $39 billion AI-server order backlog, and SMCI crashed to $29.27 on 184 million shares traded, more than 300% above average volume. Liang founded this company in 1993 and has run it for over three decades. He has now bet the whole thing three separate times.

Most traders know the ticker and the crash. Almost none know the man whose signature is on all three bets. Here is the breakdown.

From Taiwan to Texas to the 1993 Founding With Sara Liu

Liang was born in Taiwan in 1958 and trained as an electrical engineer, completing his masters degree at the University of Texas at Arlington before moving to Silicon Valley. The path matters because it is the standard route of the Taiwanese-American hardware generation that built the modern server industry. Engineers educated in Taiwan, credentialed in Texas or California, and then concentrated within a few square miles of San Jose. Jensen Huang followed a version of it. So did half the founding layer of the AI supply chain.

Through the early 1990s Liang worked as a chipset and motherboard design engineer in the Valley, the unglamorous plumbing layer of the PC era. In September 1993 he founded Super Micro Computer in San Jose together with his wife Sara Liu, who has served as a senior executive at the company ever since. The official history at Super Micro's company profile is corporate-clean, but the founding reality was a small motherboard shop competing against established Taiwanese manufacturers on one variable only. Speed.

That founding choice hardened into the company's permanent identity over the next three decades. Super Micro never tried to out-invent Intel or out-scale the contract manufacturers. It built a modular "building block" design system that let it ship a server configured around any new chip faster than anyone else could. The company IPO'd on Nasdaq in 2007, and Liang never handed the CEO seat to anyone. Thirty-two years in, he is one of the longest-tenured founder-CEOs in Silicon Valley hardware, a category where founders usually get replaced by operations executives within a decade.

The NVIDIA Symbiosis That Made Super Micro the Fastest Server Builder of the AI Boom

Super Micro's AI-era business model reduces to a single sentence. When NVIDIA ships a new GPU generation, Super Micro is first to market with a complete rack-scale server built around it. That was true with the H100, true again with the Blackwell generation, and it is the entire reason a motherboard shop from 1993 ended up with a $39 billion order backlog.

The speed edge is structural, not accidental. The building-block architecture means Super Micro designs hundreds of interchangeable server modules in advance, so a new GPU drops into a validated chassis in weeks rather than quarters. Liang and Jensen Huang have known each other for decades, both Taiwanese-born engineers who built their companies a short drive apart, and NVIDIA has repeatedly used Super Micro as the launch partner that proves a new platform works at rack scale. The dependency runs both ways but not equally. NVIDIA's allocation decisions can make or break Super Micro's quarter, which is why SMCI traders should read the Phemex NVIDIA stock outlook alongside anything written about Liang.

The ride was violent in both directions. Super Micro stock gained over 240% in 2023, entered the S&P 500 in March 2024, and briefly traded as the single best performer in the index. A company that spent twenty years as boring infrastructure became a momentum trade, and momentum trades attract short sellers.

The Liquid-Cooling Bet That Looked Eccentric Until It Printed Money

Years before AI racks made it mandatory, Liang spent capital on direct liquid cooling while the rest of the industry treated water near servers as a liability. He told investors repeatedly that liquid cooling would jump from a rounding error to as much as 30% of new data center deployments, a claim that sounded promotional until Blackwell-class GPUs started pulling so much power that air cooling physically stopped working.

When the inflection arrived, Super Micro was one of the only vendors that could ship liquid-cooled AI racks at volume from day one. The bet gave the company a genuine differentiator in a business with notoriously thin hardware margins, typically in the single digits to low teens on gross margin, far below the chip designers it assembles for. Competitors closed the gap by 2025, and the cooling edge is narrower today than the bulls claim. But the pattern is what matters for understanding Liang. He spends years early, looks wrong publicly, and then collects when physics forces the industry to his position.

The same supply chain context explains why his customers keep paying. Memory is the binding constraint on AI servers right now, and Micron's pricing power, covered in the Phemex Micron analysis, feeds directly into Super Micro's input costs. The custom-silicon wave tracked in the Phemex Marvell outlook determines whose chips fill his racks next.

The 2024 Crisis That Nearly Ended Super Micro

Honest profiles of Liang have to sit with this part, because it is the second accounting crisis of his tenure, not the first. In 2018 Super Micro was delisted from Nasdaq after delayed filings tied to prematurely recognized revenue, relisted in early 2020, and settled with the SEC for $17.5 million in August 2020. Liang personally returned roughly $2.1 millionin stock-sale profits as part of the resolution.

Then it happened again, bigger. In August 2024 Hindenburg Research published a short report alleging accounting manipulation and related-party dealings. Super Micro delayed its annual filing. In October 2024 Ernst and Young resigned as auditor mid-engagement, citing concerns about governance and management representations, one of the most severe public signals an auditor can send. The stock collapsed more than 80% from its March 2024 peak, and delisting looked like a live possibility for the second time in six years.

The resolution was messier than either bulls or bears predicted. An independent special committee reviewed the allegations and reported no evidence of fraud or misconduct and no need for restatements. BDO came in as the new auditor, the delayed filings landed in February 2025, and Nasdaq compliance was restored. The company survived. The discount never fully closed, and the June 10 crash shows why. A company with this filing history gets zero benefit of the doubt the moment it asks shareholders for $7 billion. The full filing record is public in Super Micro's 8-K filings at SEC EDGAR, and the longer corporate history is laid out in the Supermicro Wikipedia entry.

The $39 Billion Backlog and the $7 Billion Dilution Math

Now the third bet. Super Micro disclosed an order backlog of roughly $39 billion spread across more than 20 customers, a staggering number for a company that did about a tenth of that in annual revenue during the pre-AI years. Filling those orders requires buying GPUs, memory, and components upfront, months before customers pay. That working-capital gap is the entire reason for the $7 billion equity raise announced June 10, detailed through the Super Micro investor relations page.

The market's reaction was about arithmetic, not the backlog. If the share count sits near 600 million, raising $7 billionanywhere near current prices means printing well over 200 million new shares, expanding the count by roughly a third. Every existing holder gets diluted, including Liang himself, whose stake of roughly 14% drifts toward 10% if he does not participate in the offering. He is voluntarily shrinking his own slice of a company he has controlled for 32 years because he believes the orders are worth more than the ownership.

The selloff also landed on the worst possible day. Oracle crashed 13% the same session despite beating earnings, punished purely for the scale of its AI capital spending, a setup traders can compare through the Phemex Oracle stock outlook. The market is repricing the entire AI-capex complex at once, and a dilutive raise into that tape was always going to get hit. How the component suppliers split the same spending wave is mapped in the Phemex Samsung versus Broadcom comparison.

Liang's three company-defining bets line up like this.

What Kind of CEO Charles Liang Actually Is

The personal profile reads like an engineering monk. Liang is famously frugal for a CEO of a company this size, has described working 80-hour weeks for decades, and is known internally for reviewing thermal and mechanical design details that most chief executives delegate three layers down. There is no yacht-and-foundation arc here. The company is the entire project, which is exactly what makes the dilution decision so readable. A founder this attached to control does not give up a third of the cap table for a backlog he doubts.

For traders, that profile cuts both ways. The strength is conviction with a long record of being early and right on physical infrastructure. The weakness is that founder-controlled companies with thin governance layers produced both accounting crises on his watch, and an 80-hour-a-week engineer is not the person who builds the boring compliance machinery that prevents a third one. You are not buying a manager. You are buying a believer, with everything that implies in both tails.

Frequently Asked Questions

Who owns Super Micro?

Charles Liang is the largest individual shareholder with a stake of roughly 14% before the June 2026 offering, and his wife and co-founder Sara Liu holds an additional position. The rest of the float sits mostly with institutional index and mutual funds, which grew their share after the company entered the S&P 500 in March 2024.

Is Charles Liang still CEO of Supermicro?

Yes, Liang has been chief executive continuously since founding the company in September 1993, making him one of the longest-serving founder-CEOs in American hardware. He held the seat through the 2018 delisting, the 2024 auditor crisis, and the June 2026 crash.

Why did SMCI stock crash in June 2026?

Super Micro announced an equity raise of roughly $7 billion on June 10, 2026 to fund component purchases for its $39 billion order backlog, and the stock fell about 30% to $29.27. The selling was driven by dilution math rather than demand doubts, since the raise could expand the share count by roughly a third.

Did Super Micro fix its accounting problems?

The independent special committee formed after the 2024 Hindenburg report found no evidence of fraud or misconduct and required no restatements. BDO replaced Ernst and Young as auditor, delayed filings were completed in February 2025, and Nasdaq compliance was restored. The history still costs the stock a persistent governance discount versus other AI hardware names.

Bottom Line

Liang has bet the company three times and is two for two so far, which is precisely why the third bet is tradeable rather than dismissible. The near-term structure is simple. The offering price will set the floor, because $7 billion of new stock does not clear below the level the underwriters defend. Hold $28 through the pricing and a post-dilution base forms with the backlog as the bull case. Reclaim $35-36 and the market has decided the orders outweigh the share count, opening a move back toward the pre-crash close near $41. Lose $26 and the governance discount is back in charge, with the low $20s as the next zone. The man has spent 32 years proving he would rather own less of something enormous than all of something small. The chart will tell you within a quarter if the market agrees one more time.

Disclaimer: This article is for educational purposes only and does not constitute financial advice. Cryptocurrency and stock trading carries significant risk. Always do your own research and consult a qualified advisor.