The Lee family of South Korea, which controls Samsung, has doubled its wealth in twelve months. Bloomberg’s Billionaires Index now values the dynasty’s holdings at $45.5 billion, up from $22.7 billion a year ago, propelling the Lees from tenth to third among Asia’s richest families. The catalyst is not a new product or a management breakthrough. It is a 186 per cent surge in Samsung Electronics’ share price, driven almost entirely by global demand for the high-bandwidth memory chips that power artificial intelligence data centres. Samsung’s first-quarter operating profit reached 57.2 trillion won, roughly eight times what it earned in the same period last year. The Lee family did not build the AI industry. But the AI industry cannot function without what Samsung builds, and for the moment, that dependency is worth $22.8 billion in new wealth for a single family in a single year.

The chip

Samsung’s financial turnaround rests on one product category: high-bandwidth memory, or HBM, the specialised DRAM chips that sit inside the GPU modules used to train and run large AI models. Nvidia’s next-generation B300 server systems require HBM4 chips, and Samsung has entered mass production of HBM4 ahead of SK Hynix, its primary rival, after years of trailing in the technology. The shift matters because HBM commands margins that conventional memory chips cannot match. When Samsung reported first-quarter results, the semiconductor division accounted for the overwhelming majority of the profit swing, converting what had been a cyclical downturn into the most profitable quarter in the company’s recent history. Nvidia’s B300 servers, which can cost more than $1 million each, are shipping to hyperscalers and sovereign AI programmes worldwide, and Samsung is now one of the primary suppliers of the memory those systems require.

The concentration of value in a single product line is both Samsung’s opportunity and its vulnerability. HBM4 is a generational leap in memory architecture, moving the chip from a stacked DRAM design to a logic-integrated base die that allows higher bandwidth and lower power consumption. Samsung’s ability to reach volume production on HBM4 before its competitors gave it a pricing advantage that flowed directly into the first-quarter numbers. But the AI chip supply chain is notoriously volatile. Nvidia’s own product cycles, the pace of data centre buildouts by Amazon, Google, Meta, and Microsoft, and the geopolitical restrictions on chip exports to China all determine how much HBM Samsung can sell and at what price. The stock’s 186 per cent gain in twelve months prices in a sustained AI infrastructure boom. If that boom slows, the same leverage that doubled the Lee family’s wealth can reverse it.

The inheritance

The wealth surge arrives at a consequential moment for the Lee family’s finances. The heirs of the late Samsung chairman Lee Kun-hee, who died in October 2020, have been paying the largest inheritance tax bill in South Korean history. The total obligation is approximately 12 trillion won, roughly $9 billion at current exchange rates, which the family agreed to pay in six annual instalments. The final instalment came due in April 2026. The tax was assessed on the estate’s value at the time of Lee Kun-hee’s death, when Samsung’s share price was substantially lower than it is today. The family has funded the payments through a combination of dividends, share sales, and loans against their Samsung holdings. The timing of the stock rally means the inheritance tax, once seen as a potential threat to the family’s controlling stake, has been absorbed without forcing a dilutive restructuring of the group’s cross-shareholding structure. The dynasty’s grip on the Samsung conglomerate remains intact.

That grip is unusual by the standards of global technology companies. Samsung is not a founder-led startup or a publicly traded corporation with dispersed ownership. It is a chaebol, a family-controlled industrial conglomerate in which the founding family maintains control through a web of cross-shareholdings across dozens of subsidiaries. The Lee family’s direct equity stakes in Samsung Electronics are relatively modest, around 5 per cent of outstanding shares, but their control is exercised through Samsung C&T, Samsung Life Insurance, and other group entities that collectively hold enough voting power to determine the company’s direction. The AI-driven rally in technology stocks has inflated the value of every entity in this chain, amplifying the family’s paper wealth far beyond what their direct Samsung Electronics holdings alone would suggest.

The workers

The wealth transfer to the Lee family has not gone unnoticed inside Samsung. In March, approximately 30,000 members of the National Samsung Electronics Union rallied outside the company’s Hwaseong semiconductor campus, the largest labour demonstration in the company’s history. The union is demanding that workers receive a share of the profits their labour produces, specifically a bonus tied to 15 per cent of the semiconductor division’s operating profit. Samsung has historically resisted union demands, and the company’s labour relations remain more adversarial than those of most large technology employers. The union has threatened an 18-day strike beginning 21 May if its demands are not met. Technology companies cutting thousands of workers while reporting record profits is a pattern that extends well beyond Samsung, but the dynamic is sharper in a chaebol structure where the controlling family’s wealth is publicly tracked and the connection between labour and capital is unusually direct.

The workers’ grievance has a specific arithmetic. Samsung’s semiconductor division generated 57.2 trillion won in operating profit in the first quarter alone. Fifteen per cent of that figure is approximately 8.6 trillion won, or $6.3 billion, for a single quarter. The union argues that the HBM4 chips driving Samsung’s profits are manufactured by workers operating in cleanrooms under demanding conditions, and that the value those chips create should be distributed more broadly than the current compensation structure allows. Samsung’s management has not publicly responded to the specific profit-sharing demand, but the company’s annual wage negotiations have historically ended with increases well below what the union requests. The tension is a microcosm of a broader question that the AI boom is raising across the technology industry: when a single product category generates windfall profits because of macroeconomic conditions beyond any individual worker’s control, who is entitled to the upside?

The dependency

The Lee family’s wealth is a proxy for a structural shift in the global economy. The $22.8 billion they gained in twelve months did not come from Samsung selling more phones, televisions, or appliances. It came from the world’s largest technology companies spending hundreds of billions of dollars on AI infrastructure that requires the specific type of memory chip Samsung manufactures. Alphabet, Amazon, and Meta alone guided for more than $650 billion in combined AI capital expenditure in their most recent earnings, and a significant share of that spending flows through the semiconductor supply chain to companies like Samsung, SK Hynix, and Micron. The concentration is extreme: three memory manufacturers supply virtually all the HBM chips the AI industry needs, and Samsung’s ability to reach HBM4 production at scale has shifted its market share in the highest-margin segment at exactly the moment demand is peaking.

That dependency runs in both directions. Samsung needs the AI boom to sustain the share price that doubled the Lee family’s wealth. The AI industry needs Samsung to produce enough HBM4 chips to keep Nvidia’s server shipments on schedule. If Samsung’s HBM4 yields falter, data centre buildouts slow. If data centre buildouts slow, Samsung’s margins compress. The venture capital ecosystem that has emerged around AI infrastructure is built on the assumption that compute will keep scaling, which requires memory to keep scaling, which requires Samsung and its competitors to keep investing in fabrication capacity at a pace that matches demand. The Lee family’s $45.5 billion fortune is not a static asset. It is a real-time readout of the market’s confidence that the AI infrastructure cycle has years to run. The family’s position, third-richest in Asia, is held at the pleasure of a supply chain that did not exist in its current form eighteen months ago. Dynasties are supposed to be durable. This one’s value is a function of how many GPUs Nvidia can ship next quarter.