Renowned PE Firm Completes Exit

marsbitPublished on 2026-07-29Last updated on 2026-07-29

Abstract

A leading global private equity firm has completely divested its stake in a Japanese memory storage giant. Bain Capital has sold all its shares in Kioxia, formerly Toshiba Memory, realizing approximately $17 billion in total proceeds. This transaction is reportedly one of the most successful exits in global private equity history. The divestment concludes a nearly decade-long investment that began in 2018. Bain led a consortium to acquire the then-Toshiba Memory business for $18 billion during a period of financial distress for its parent company, Toshiba. The investment initially faced challenges, including multiple postponed IPO attempts and a failed merger plan with Western Digital's flash memory business. The investment's fortunes reversed dramatically with the surge in demand for memory chips driven by the AI boom, particularly for High Bandwidth Memory (HBM). This led to a sharp increase in prices for DRAM and NAND flash memory. Consequently, Kioxia's stock price soared nearly 50-fold from its 2024 IPO price, briefly making it Japan's most valuable company in mid-2025. Bain began significantly reducing its stake in late 2025 and finalized the complete exit by July of this year. The report draws a parallel to the success of Changxin Memory Technologies (CXMT) in China, whose backers also demonstrated long-term patience. CXMT recently debuted on China's STAR Market with a market capitalization exceeding 3 trillion yuan ($412 billion), underscoring the massive returns p...

Across the Pacific, a leading storage company was fully divested by a private equity firm.

The protagonist is the Japanese memory giant Kioxia—riding the AI tailwind, its stock price once soared nearly 50-fold. As circumstances shifted dramatically, its major shareholder, Bain Capital, revealed that it had completely divested its holdings by July of this year, with total proceeds reaching $17 billion (approximately 115.1 billion RMB).

In other words, Bain Capital realized a profit of 110 billion RMB, setting a global record for returns in private equity history.

"An opportunity like Kioxia is rare and unpredictable; it has been an outstanding outcome for all parties involved," said an executive partner at Bain Capital. Market observers noted that the successful completion of such a large-scale share transaction indicates that demand from buying institutions remains robust.

Like other major memory manufacturers, but different from Changxin Technology which focuses on the DRAM (runtime memory) track, Kioxia primarily produces NAND flash memory. Looking domestically, it's Wuhan-based Yangtze Memory Technologies that is truly advancing in this field.

Looking back to last night, overseas memory giants experienced a sudden shock, with stocks retreating 30% to over 50% from their June highs. On the other hand, Changxin Technology, the new stock king, topped the market with a market cap exceeding 3 trillion RMB. Amidst the calm and frenzy, everyone is watching the future landscape of memory in the AI era.

A Decade Ago, Contrarian Heavy Investment

To sell well, one must also buy at the right time.

Looking back to 2016, Japanese diversified electronics manufacturer Toshiba, due to massive losses incurred by its nuclear subsidiary Westinghouse Electric and compounded by an accounting scandal, fell into insolvency and faced delisting.

That year, the century-old giant Toshiba was in dire straits, forced to sell assets to cover its losses. Its memory business was placed on the divestiture list at that time—however, in the 2017 fiscal year, Toshiba's Memory Division's operating profit was as high as 500 billion yen, arguably its most profitable business then.

In the view of many, given the profitability of Toshiba Memory at the time, waiting a bit longer could have allowed Toshiba to recover without a sale. But as "one can never earn money beyond their cognition," constrained by Toshiba's internal long-term decision-making culture favoring heavy electrical sectors, the memory division requiring continuous high investment was ultimately abandoned.

Spin-off, fundraising, crisis resolution—once the news broke, various bidders flocked.

Throughout 2017, it's reported that at least ten companies participated in the bidding, including Micron, SK Hynix, Western Digital, Broadcom (together with Silver Lake), Amazon, Google, Apple, Foxconn Group, and others.

Finally, the real buyer emerged. Bain Capital, together with SK Hynix, Apple, Dell, Seagate, and others, formed a consortium and in 2018 acquired approximately 55% of Toshiba Memory for $18 billion, with Bain Capital contributing over $1 billion. Later in 2019, Toshiba Memory was renamed Kioxia, establishing its positioning as a memory value service provider.

The story of breaking through fierce competition seemed wonderful, but the deal was not without controversy.

As is well known, the memory industry is highly cyclical. When Toshiba decided to sell, memory had just passed a super cycle driven by smartphone development. By the time Bain Capital's acquisition was finalized, with the smartphone market beginning to hit saturation and replacement cycles lengthening, demand for memory chips softened, and prices for Toshiba Memory's NAND Flash slid into a downturn.

Therefore, many voices at the time considered Bain Capital to have "bought at the peak." No one anticipated that the AI wave and the subsequent explosion in memory demand would completely rewrite its fate.

Landscape Shifts, Selling Amidst the Roar

The epic reversal takes the stage.

Starting in the second half of 2025, explosive growth in AI computing power made High Bandwidth Memory (HBM) a hot commodity. Memory manufacturers like Samsung Electronics, SK Hynix, and Micron adjusted capacity towards high-end products, significantly squeezing supply for consumer-grade memory. Consequently, memory product prices rose across the board: at the beginning of 2026, DRAM and NAND Flash product prices hit their highest levels in nearly a decade, with cumulative increases for some models even exceeding 300%.

The frenzy first ignited in the capital markets. Memory giants like Samsung, SK Hynix, and Micron, which collectively hold over 90% of the global DRAM market, saw their stock prices soar by 5 to over 10 times from their lows since June last year.

The same goes for NAND producers like Kioxia and SanDisk. Over the past year, Kioxia's stock price surged nearly 50-fold at one point, with its market cap even reaching 56 trillion yen (approximately $345 billion) in mid-June this year, surpassing Toyota to become Japan's most valuable company. SanDisk followed a similar trajectory, with its stock price rising over 50-fold from June 2025 to June this year.

Looking domestically, similar fervor unfolded. A typical example is the "Shenzhen Memory Four Little Dragons"—Longsys, Tech Star, BIWIN, and Dapu Micro—whose stock prices have doubled over the past year, with their combined market cap once nearing 900 billion RMB in mid-June this year.

It was amidst this roar.

Starting in November 2025, Bain Capital began large-scale divestment of its Kioxia holdings. By mid-June this year, its stake had decreased from about 44% in December 2025 to about 14%, until early July this year when Bain Capital completed its full exit.

"We no longer hold any Kioxia shares," David Gross, an executive partner at Bain Capital, formally stated in an interview earlier in July. The latest news, according to Nikkei estimates, is that Bain Capital's sale of its shares realized gains of approximately 2.5 trillion yen (about $17 billion).

Calculated, the return is nearly 20 times the initial investment, creating one of the most classic private equity cases in the global semiconductor industry over the past decade.

Patience, A Little More Patience

Once again proving the victory of patient capital in the hard tech field.

What outsiders may not know is that before Kioxia's stock price surge began in 2025, Bain Capital endured pressure from various sides for several consecutive years:

Due to weak consumer electronics demand and plunging memory prices, Kioxia's IPO plans failed successively in 2020, 2021, 2022, and 2024. With no exit via listing, news emerged in 2023 of a merger between Kioxia and Western Digital's NAND business. If successful, this would have created the world's largest NAND manufacturer, surpassing Samsung. However, due to strong opposition from SK Hynix to the merger plan, this strategy to enlarge the pie ultimately fell through.

It wasn't until the end of 2024 that Kioxia finally listed on the Tokyo Stock Exchange, but its debut performance fell short of expectations. Its overall valuation was only 750-780 billion yen (about $5 billion), significantly lower than the total $18 billion acquisition price paid by Bain Capital years earlier.

Accompanying and waiting during the trough, deserving of praise at the peak.

A similar scene unfolded with investors in Changxin Technology. In 2016, when Changxin Technology hadn't even officially started production on its first production line, Hefei State-owned Capital invested 14.4 billion RMB in Changxin. Against a backdrop where domestic DRAM technology, equipment, and talent were virtually nonexistent, this investment was undoubtedly a bold gamble. Similarly, in 2021, when Changxin's products were not yet mass-produced on a large scale and its performance was still in the red, while most were still watching and hesitating, institutions like Walden International and Cornerstone Capital firmly made heavy investments, enduring five years of solitude.

All sails have passed. On July 27th, Changxin Technology officially listed on the STAR Market with a market cap exceeding 3 trillion RMB, becoming the first hard tech IPO in A-share history to surpass a trillion-yuan market cap at its debut.

Excellent technology companies need time to mature, and rational investment likewise requires patient companionship. On the path of stealthy progress over 5, 8, or even 10 years, "perhaps the pessimists are correct, but the optimists always reap the rewards."

This article is from the WeChat public account "PEdaily" (ID: pedaily2012), author: Feng Yuchen

Related Questions

QWhat did Bain Capital achieve through its investment in Kioxia, and how does this compare to the original investment?

ABain Capital achieved a total return of approximately $17 billion by completely divesting its stake in Kioxia by July of this year. This represents a return of nearly 20 times its initial investment of over $1 billion in 2018, setting a record for a global private equity transaction.

QWhy did Toshiba decide to sell its profitable memory business in 2017 despite it being a major profit driver?

AToshiba faced a severe financial crisis in 2016 due to massive losses from its Westinghouse nuclear subsidiary and an accounting scandal, pushing it to the brink of insolvency and delisting. To raise funds and survive, Toshiba was forced to sell assets, including its highly profitable memory unit, despite its strong performance at the time.

QWhat major market shift created the opportunity for Kioxia's dramatic valuation increase and Bain Capital's successful exit?

AThe explosive growth in AI computing power starting in the second half of 2025 led to a surge in demand for high-bandwidth memory (HBM). This caused major memory manufacturers to shift capacity to high-end products, squeezing the supply of consumer-grade memory and triggering a significant price increase for DRAM and NAND Flash products, with some models rising over 300%.

QWhat challenges did Bain Capital and Kioxia face before the AI-driven market boom in 2025?

ABefore the 2025 boom, Kioxia faced a prolonged downturn in the memory chip market due to weak consumer electronics demand and falling prices. Its plans for an IPO failed four times (2020, 2021, 2022, 2024). A proposed merger with Western Digital's NAND business in 2023, which could have made it the world's top NAND producer, also collapsed due to opposition from SK Hynix. When it finally IPO'd in late 2024, its valuation was significantly lower than Bain's original acquisition cost.

QHow does the article compare the investment stories of Bain Capital in Kioxia and Chinese investors in ChangXin Memory Technologies (CXMT)?

AThe article draws a parallel between the two, highlighting the theme of 'patient capital' in hard tech. Both investments required long-term commitment through difficult periods. Bain Capital endured years of pressure and low valuations before Kioxia's value skyrocketed. Similarly, early investors in CXMT, like the Hefei municipal government in 2016 and firms like Walden and Cornerstone Capital in 2021, made bold bets when the company had no established production lines or was still unprofitable, patiently waiting for its eventual success and massive IPO valuation.

Related Reads

A7 Discusses Accumulated Experience in Using Stablecoins

The Russian State Duma and Federation Council have passed the "Law on Digital Currencies and Digital Rights," which is set to expand the use of stablecoins in cross-border trade settlements starting September 1, 2026. According to the Bank of Russia, exporters and importers will be able to use cryptocurrency in international payments, both directly and through intermediaries, while domestic crypto payments will remain largely prohibited. Company A7, which facilitates operations with the A7A5 stablecoin—Russia's first ruble-pegged stablecoin classified as a Digital Financial Asset (DFA)—commented on the development. Oleg Ogienko, Director of Government Relations and International Affairs for the A7A5 project, stated that while digital asset settlements are still a new practice for many firms, A7 has already accumulated significant expertise in legal documentation, compliance, currency control, and working with infrastructure participants. He added that A7 will continue to apply this expertise for client operations after the law takes effect and will adapt its business processes as the Central Bank issues further regulations. A7 is a Russian international settlement system established in 2024 with participation from PSB Bank. It facilitates cross-border payments and supports foreign trade operations for Russian businesses. Its key instrument, the A7A5 stablecoin, operates on the Tron and Ethereum networks with a market capitalization exceeding $567 million and is circulated in Russia as a DFA via the "Token" platform.

cryptonews.ru11m ago

A7 Discusses Accumulated Experience in Using Stablecoins

cryptonews.ru11m ago

Bitcoin Panicking? Mythos Cracks Post-Quantum Cryptography Algorithm in 60 Hours

Claude Mythos Preview, a new AI system from Anthropic, has made significant breakthroughs in fundamental cryptanalysis, targeting the mathematics underlying algorithms rather than just implementation bugs. Its first target was HAWK, a candidate in the NIST post-quantum signature standardization process. Mythos identified a previously unexploited symmetry (a nontrivial automorphism) in the lattice structure of HAWK, effectively halving its effective key strength. It achieved this feat in just 60 hours, a task that had eluded two years of human expert review. Crucially, the human operator was not a lattice cryptography expert, primarily managing the project while AI agents debated and discovered the attack path. Its second, more dramatic achievement was against a simplified 7-round version of AES-128, the world's most widely used symmetric encryption algorithm. Initially reluctant, stating the task was "impossible," Mythos was prompted to "try novel ideas." It then autonomously rewrote its own agent framework and proceeded to discover a novel attack method it named "Möbius Bridge." This technique bypasses a critical 256-guess step in previous "meet-in-the-middle" attacks, resulting in a speedup of 200-800 times. Both findings are currently "harmless"—HAWK is not deployed, and full AES-128 remains secure. However, the process reveals a profound shift. Mythos completed the AES discovery in about a week, while human researchers spent nearly a month verifying its correctness. This highlights a new bottleneck: the rate of AI-driven discovery may soon outpace human capacity for verification. The research concludes with a pressing, unanswered question: what happens when such an AI finds a critical flaw in a widely deployed, real-world cryptosystem?

marsbit21m ago

Bitcoin Panicking? Mythos Cracks Post-Quantum Cryptography Algorithm in 60 Hours

marsbit21m ago

From Computing to Storage: The Two-Dimensional Semiconductor Industry Chain is Taking Shape

The article "From Computing to Storage: The 2D Semiconductor Industry Chain is Taking Shape" discusses the rise of two-dimensional (2D) semiconductors as a pivotal material in the post-Moore era. As silicon-based chips approach physical limits, 2D semiconductors like molybdenum disulfide (MoS2), offer atomic-layer thinness and superior electrical properties, making them a promising non-silicon alternative for advanced nodes beyond 1nm. Globally, industry leaders like TSMC, Intel, and Samsung, along with research institutes like IMEC, are actively investing in this field. Domestically in China, significant progress is being made across the entire supply chain, from material production and equipment R&D to chip integration and ecosystem development. **Key highlights include:** * **Material & Equipment Breakthroughs:** Chinese research teams and companies like Jimo Xin Technology have achieved mass production of 6-inch and 8-inch 2D semiconductor single-crystal wafers using domestically developed equipment like Oxy-MOCVD. This addresses the critical challenge of large-area, high-quality material growth. Breakthroughs have also been made in p-type 2D materials (e.g., MoSi2N4) and wafer-scale ferroelectric thin films, essential for building complementary circuits and low-power transistors. * **Engineering Ecosystem Formation:** A landmark 8-inch 2D semiconductor pilot line in Pudong has become fully operational, bridging the gap between lab research and industrial manufacturing. The release of a Process Design Kit (PDK) compatible with mainstream EDA tools further connects design and fabrication, supporting wafer-level manufacturing of complex circuits. * **Expanding Application Landscape:** Applications are expanding from logic computing to memory. Chinese researchers have developed pioneering 2D semiconductor-based microprocessors, including a 32-bit RISC-V CPU and a higher-density multi-bit parallel processor. In memory, 2D materials' ultra-low leakage current enables DRAM with exceptionally long data retention and advances toward single-electron non-volatile storage. 2D semiconductors also show unique potential in RF circuits, radiation-hardened communications, flexible electronics, and quantum computing. In conclusion, the 2D semiconductor industry in China is rapidly advancing from laboratory research into the engineering validation and small-batch production phase. A complete industrial chain covering materials, equipment, manufacturing, design, and application is初步 forming, positioning the sector as a new competitive dimension in the global semiconductor landscape.

marsbit26m ago

From Computing to Storage: The Two-Dimensional Semiconductor Industry Chain is Taking Shape

marsbit26m ago

Trading

Spot
活动图片