Do not call $9B to $150B a simple 16x increase

In 2020, SK hynix agreed to acquire Intel’s NAND memory and storage business for $9 billion. The transaction included the NAND SSD business, NAND component and wafer operations, and the Dalian NAND manufacturing facility in China. After the first-phase closing in 2021, the SSD business became the foundation of the U.S.-based Solidigm entity.

On September 25, 2026, Reuters reported that Solidigm was considering a U.S. IPO as early as 2027 that could value the company at up to $150 billion and raise up to $15 billion. The report also said the company had held a bank bake-off for potential IPO roles.

Those numbers are not like-for-like. The 2020 transaction covered a specific asset package; the $150 billion figure is a preliminary upper-end valuation discussed for a later standalone company. Treating this as a simple 16x return would overstate what the evidence proves.

The useful signal is not the valuation — it is Storage moving up the stack

GPU was the most visible AI infrastructure bottleneck. Then HBM, networking, power and cooling became increasingly strategic as compute scaled.

Storage fits the same logic. Training data, checkpoints, inference data, enterprise datasets and generated outputs all grow with AI usage. The ability to store and feed that data at the required scale starts to affect system performance, power use and economics.

Solidigm’s 2026 outlook argues that more powerful GPUs are not enough and that storage efficiency is becoming a defining factor in next-generation system performance. That is a vendor claim, so it should not be treated as a market-wide conclusion by itself. The stronger test is whether products and customers behave as if storage has become more strategic.

A 122TB SSD targets power and space, not just capacity

Solidigm introduced the 122.88TB D5-P5336 data-center SSD, doubling the capacity of its earlier 61.44TB version and positioning the drive for AI and other data-intensive workloads.

The important issue is not simply storing more data. AI data centers have constrained rack space and power budgets, while GPUs compete for an increasing share of available power. Solidigm positions high-density SSDs as a way to reduce storage power and footprint so more of those resources can be allocated elsewhere in the system.

That moves storage competition from ‘how much data can a drive hold?’ toward ‘how efficiently can the entire AI system be built within fixed power and space?’

CoreWeave signed a multi-year deal to secure storage capacity in advance

In August 2026, CoreWeave and Solidigm announced a multi-year strategic agreement giving CoreWeave priority access to enterprise SSD capacity for its AI cloud platform.

CoreWeave described AI as an integrated stack of compute, networking, storage and software that must scale together. The announcement explicitly called storage a critical constraint in capacity planning.

That is stronger evidence than a forecast that storage may become important. An AI cloud provider is making a long-term capacity commitment so storage supply can keep pace with its roadmap.

The talent map should treat Storage as its own AI infrastructure layer

If AI talent is reduced to researchers and GPU designers, the workforce requirements created by infrastructure scaling are easy to miss. As storage becomes more system-critical, capabilities around NAND and SSD products, firmware, controller architecture, reliability, system integration and storage performance can become strategically relevant.

The public evidence in this article does not prove a specific hiring growth rate for those roles. What it directly supports is a shift in Solidigm’s product positioning, customer commitments and strategic emphasis.

The next step for talent intelligence is therefore to watch actual job postings for where these storage-infrastructure capabilities begin to expand, rather than counting only job titles that contain the word AI.

BANSEOG VIEW | Track what becomes scarce when GPU capacity grows

No one yet knows what valuation Solidigm would receive in an actual IPO. The reported $150 billion ceiling is preliminary and the transaction itself could change with market conditions.

But the operating sequence is worth retaining: enterprise SSD was already part of the 2020 acquisition logic; Solidigm later pushed 122TB high-density drives; an AI cloud provider signed a multi-year capacity agreement; and the company now presents storage as a core AI infrastructure decision.

The reusable question is not ‘Will Solidigm really be worth $150 billion?’ It is ‘What must scale every time more GPUs are deployed, and what becomes the next constraint?’ That question is more useful for spotting where AI infrastructure value may move next.

BANSEOG VIEW

Banseog View — follow what becomes scarce as GPUs scale

The reported up-to-$150B IPO valuation is preliminary and should not be compared mechanically with the $9B acquisition price.

The 122TB SSD and CoreWeave’s multi-year capacity agreement show storage becoming connected to power, space and supply constraints in AI infrastructure.

Instead of searching only for the next GPU company, track the layer that becomes scarce as GPU deployment expands.

SOURCES

Primary sources and references

The up-to-$150B valuation and up-to-$15B proceeds are preliminary figures reported by Reuters from unnamed sources, not finalized IPO terms. The 2020 $9B transaction and a potential later IPO valuation cover different scopes, times and valuation methods, so this article does not present them as a simple 16x investment return. Product power and footprint claims are based on Solidigm’s own published materials and may vary by system configuration.