Intel’s $9 Billion Regret: The Business Sale Haunting It Today

The semiconductor industry has a funny way of changing the rules just when companies think they have figured them out. A few years ago, the hottest chip businesses were processors and foundries. Memory was viewed as a cyclical commodity that could swing from shortage to glut in a matter of quarters.

Fast forward to 2026, and artificial intelligence has turned memory into one of the most strategically valuable pieces of the AI supply chain. That shift helps explain why Intel‘s (NASDAQ:INTC | INTC Price Prediction) biggest strategic headache today traces back to a decision the company made more than four years ago.

Intel’s Stunning Turnaround Changed Everything

Just a year ago, Intel looked like a company fighting for survival. Rivals including Broadcom (NASDAQ:AVGO), Taiwan Semiconductor Manufacturing (NYSE:TSM), Advanced Micro Devices (NASDAQ:AMD), and NVIDIA (NASDAQ:NVDA) were reportedly exploring ways to acquire pieces of Intel’s business or potentially pursue larger transactions. The company’s foundry ambitions were burning cash, market share losses had mounted, and investors questioned whether Intel could remain a leading semiconductor manufacturer.

Then the script flipped. In August 2025, the Trump administration converted $5.7 billion in unpaid CHIPS Act grants and $3.2 billion from the Secure Enclave defense program into a 9.9% equity stake in Intel, paying $20.47 per share. That government investment, totaling $8.9 billion, dramatically altered perceptions of Intel’s future. Instead of becoming a breakup candidate, Intel became a national technology priority.

The result was one of the market’s most striking recoveries. Intel shares climbed more than 170% in the first half of 2026, hitting an all-time high near $142 in late June, as investors priced in a genuine turnaround rather than a restructuring scenario. The stock has since pulled back from those heights, but the broader narrative shift remains intact.

Second-quarter 2026 results confirmed the momentum. Intel posted revenue of $16.1 billion, up 25% year over year, the fastest quarterly growth rate since 2011. Its data center and AI segment surged 59% to $6.3 billion, and gross margin recovered to 42% from roughly 3% a year earlier. CEO Lip-Bu Tan called the results “our strongest revenue growth in more than fifteen years.” Intel also disclosed it is now supply-constrained in its data center division, a problem it has not faced in years.

But that very success has revealed a new problem.

The $9 Billion Sale That Looks Different Today

The deal originated in October 2020, when Intel agreed to sell its NAND memory business to SK hynix for approximately $9 billion across two phases. The first phase closed in December 2021, transferring Intel’s NAND solid-state drive business and its Dalian, China manufacturing facility for roughly $7 billion. The second and final phase closed in March 2025, when SK hynix acquired the remaining NAND intellectual property and R&D employees for approximately $1.9 billion, completing the divestiture.

At the time of that first closing, the logic was difficult to argue with. Memory had long been one of the semiconductor industry’s most cyclical markets: prices swung violently on supply and demand shifts, margins could evaporate in a single quarter, and Intel badly needed capital to fund its foundry ambitions. Selling the NAND business allowed management to focus on CPUs, foundry services, and higher-priority growth initiatives.

Today, the AI boom has rewritten the economics of memory. A global shortage of DRAM and NAND is now squeezing the entire tech supply chain, driven by hyperscalers redirecting memory production toward high-bandwidth memory (HBM) for AI accelerators. Micron’s CEO stated in June 2026 that the shortage is expected to persist through 2027. Intel itself said it expects flat PC sales in the third quarter because of the memory crunch, a direct, tangible consequence of not controlling its own supply.

That creates a compounding problem for Intel. Rather than producing its own memory, Intel has to compete for constrained supply. As CEO Lip-Bu Tan recently told analysts:

“We used to have a memory business. We sold it to SK hynix. So right now, we have to figure out a way to really secure some of the memory requirement to serve our customer.”

That is not the kind of statement investors want to hear when memory is the bottleneck throttling growth. The situation appears serious enough that, in August 2026, Tan hinted that Intel is investigating a new memory architecture and 3D stacking approach, signaling the company may eventually seek to re-enter the space it spent years exiting.

What It Means for Intel Going Forward

The irony is hard to miss. Intel spent years and billions of dollars trying to become a world-class foundry. It now has government backing, renewed foundry momentum, and 10 long-term customer agreements signed with outside manufacturers. Yet one of the industry’s fastest-growing profit pools sits entirely outside its walls.

Selling the NAND business in 2021 was not irrational. Intel needed capital, focus, and a credible path forward, and the company’s financial position at the time left little room for discretionary bets on cyclical commodities. The regret is not rooted in bad decision-making; it comes from how dramatically the market changed afterward.

AI has transformed memory from a cyclical side business into a strategic necessity. Industry leaders such as SK hynix, Samsung, and Micron (NASDAQ:MU) are benefiting from shortages that have left portions of their AI-related memory capacity effectively spoken for years in advance. SK hynix, which now owns the NAND assets Intel sold, has seen its NAND market share rise to roughly 20% following the completed acquisition, making it the world’s second-largest NAND supplier behind Samsung.

Intel now finds itself on the opposite side of that equation, competing for allocations of the very product it once manufactured.

Key Takeaway

Intel’s sale of its NAND business was not a mistake at the time it happened. The company needed focus, capital, and a path forward, and on those terms the deal delivered. The regret stems from how profoundly the market transformed afterward.

Intel has engineered a remarkable turnaround and regained credibility with investors. Its Q2 2026 revenue growth, data center expansion, and foundry progress are all genuine achievements. But the AI era has elevated memory from a cyclical commodity to a strategic resource, and Intel now relies on competitors for a product its own business increasingly cannot function without.

The company’s future still looks considerably brighter than it did a year ago. One of its biggest remaining constraints, though, may be a business it willingly sold for $9 billion a few years back.

Editor’s note: This article has been updated to reflect Intel’s two-phase NAND divestiture structure (Phase 1 closed December 2021, Phase 2 closed March 2025 for a combined ~$9 billion), Intel’s Q2 2026 financial results including 25% revenue growth to $16.1 billion and a 59% surge in the data center and AI segment, the government’s 9.9% equity stake acquired at $20.47 per share for $8.9 billion, Intel CEO Lip-Bu Tan’s August 2026 hints about re-entering memory, and Micron’s CEO forecast that the memory shortage will persist through 2027.

Contact [email protected] for any questions or corrections.