Micron ($MU): AI has raised the floor but has it killed the memory cycle?
Hi guys, I'd appreciate it if you shared your thoughts about the valuation of Micron. I know it's not a value stock right now but I hope the subreddit is suitable for a general valuation discussion, reverse DCF models etc... as well. Can be deleted if not.
# Intro
Micron is perhaps in the strongest position in its history and indeed the stock has been priced in that way.
Memory stocks have historically created a dangerous optical illusion where near the bottom of the cycle, earnings disappear and the stocks look expensive and near the top, pricing rises faster than costs, margins explode, and the stocks suddenly trade at seemingly irresistible multiples.
Historically, that low multiple has often been a warning that earnings are near a cyclical peak and not really proof of undervaluation.
The difference in the field today, Micron included, is that the bull case is real.
AI has created enormous demand for high-bandwidth memory where Micron is now volume-shipping HBM4 for NVIDIA, expanding data-centre products, and becoming more deeply integrated into customer roadmaps. The business has clearly improved but the question is whether it has improved enough to justify the current valuation...
# The key problem: Most of the current boom is pricing
Micron’s recent results are extraordinary, but the acceleration came mainly from higher selling prices.
Sequential DRAM prices increased in the mid-60% range, while NAND prices rose in the high-70% range. Bit shipments grew much more modestly. I thibk that's important because Micron has been here before.
In fiscal 2018, it generated $30.4 billion of revenue and $9.2 billion of adjusted free cash flow. Then by fiscal 2023, revenue had almost halved, operating margin was negative 37%, and adjusted FCF had fallen to negative $5.5 billion.
The products still remained essential but supply, demand and pricing changed.
# What the valuation requires
At the valuation used in my analysis, Micron’s enterprise value was approximately $1.36 trillion.
Using a 9% discount rate and 2% perpetual growth, that corresponds to roughly:
**$95 billion of sustainable annual FCFF**
For context:
• Previous full-year adjusted-FCF record: $9.2 billion
• 10-year average through fiscal 2025: about $1.9 billion
• Annualised fiscal Q2 2026 adjusted FCF: about $27.6 billion
At a normalised 30% FCFF margin, $95 billion of annual cash flow would require approximately:
$317 billion of annual revenue
That is more than 8 times Micron’s fiscal 2025 revenue.
The market is clearly pricing in a fundamentally different company, not just another strong memory cycle.
# Why Bulls aren't crazy but Bears are not crazy either
HBM is harder to manufacture than ordinary DRAM. It requires advanced packaging, stacked dies, high yields lengthy customer qualification,...
It also consumes disproportionate wafer capacity, which can tighten the broader DRAM market.
Micron is more technologically capable, more exposed to data centres and more strategically important than during previous cycles.
On the other hand, bear case doesn't require AI demand to collapse but rather "only" supply to respond.
Micron, Samsung and SK hynix are all expanding capacity. New supply takes years to arrive, so the shortage may continue, but eventually more capacity could restore customer choice and weaken pricing.
We've seen that story before and just because the floor raised, doesn't mean it's not the same room anymore.
Micron’s exceptional margins are also not limited to HBM. Its Mobile and Client segment recently produced a 79% non-GAAP gross margin, suggesting a broad scarcity premium across conventional memory as well.
# My 4 scenarios the shortage could end
Ny scenarios all assume that AI demand remains strong.
What changes is how much of today’s pricing power survives, how much supply enters the market, and how long elevated growth continues beyond fiscal 2030.
(illustrative operating scenarios, not precise forecasts)
**Scenario A: The traditional cycle returns**
Micron remains relevant to AI, but new capacity eventually weakens pricing across HBM, conventional DRAM, and NAND.
Revenue falls from its peak while depreciation and capital intensity remain elevated.
Illustrative assumptions:
Fiscal 2030 revenue: approximately $95 billion
Fiscal 2030 FCFF: approximately $8 billion
Fiscal 2035 revenue: approximately $130 billion
Fiscal 2035 FCFF: approximately $17 billion
Estimated enterprise value: approximately $200 billion
Estimated equity value: roughly $183 per share
This is a scenario in which investors paid a permanently different valuation for what ultimately remained a cyclical scarcity business, not an “AI was fake” scenario.
This scenario supports approximately 15% of the June 22 enterprise value.
**Scenario B: Micron establishes a much higher floor**
HBM consumes enough capacity to keep conventional DRAM structurally tighter. Industry discipline improves, and Micron becomes significantly more oriented toward data-centre products.
Assumptions:
Fiscal 2030 revenue: approximately $165 billion
Fiscal 2030 FCFF: $40 billion
Fiscal 2035 revenue: $207 billion
Fiscal 2035 FCFF: $52 billion
Estimated enterprise value: $616 billion
Estimated equity value: roughly $552 per share
This would represent an extraordinary improvement over Micron’s historical economics yet it would still support only around 45% of the June 22 enterprise value.
**Scenario C: A genuinely new Micron**
Micron becomes a major long-term AI-memory supplier with durable HBM exposure.
Advanced server memory and data-centre storage support structurally higher scale and margins, although the business retains some cyclicality.
Assumptions:
Fiscal 2030 revenue: $235 billion
Fiscal 2030 FCFF: $71 billion
Fiscal 2035 revenue: $349 billion
Fiscal 2035 FCFF: $94 billion
Estimated enterprise value: $1.10 trillion
Estimated equity value: roughly $978 per share
This is already an aggressively bullish operating outcome. Even so, it supports approximately 81% of the June 22 enterprise value.
**Scenario D: The market-implied AI-memory champion**
This is one potential operating path that closes the valuation gap. Micron becomes a technological peer or leader across major HBM generations. AI infrastructure spending remains exceptional and HBM continues consuming disproportionate wafer capacity. Here, Micron sustains elevated growth well beyond 2030.
Assumptions:
Fiscal 2030 revenue: $270 billion
Fiscal 2030 FCFF: $84 billion
Fiscal 2035 revenue: $438 billion
Fiscal 2035 FCFF: $118 billion
This scenario supports the full enterprise value of approximately $1.36 trillion.
The market doesn't necessarily require Micron to maintain peak margins forever, but it does require the company to become radically larger, generate exceptional cash flows for most of the next decade, and avoid anything resembling its historical cycle.
# My view
I think AI and HBM have materially raised Micron’s through-cycle earnings power but i am less convinced they have abolished the memory cycle.
Micron may remain a much stronger company than it has ever been while still falling short of what the valuation requires.
The company has probably raised the floor yet the market appears to be pricing it as though the company has removed the ceiling. Personally I think that based on the valuation and my analysis, Micron will end up somewhere between the B and C scenario, probably closer to the B one.
Today's results can test execution, pricing and cash conversion but rhey cannot settle the larger question of whether Micron’s current economics survive once supply catches up and memory pricing normalises...
So, the question for the discussion is: has HBM permanently changed Micron’s economics, or are investors once again valuing a memory producer near the top of a shortage?
What do you think?
\-------------------------------------
(i don't want to promote, nor you need to follow me anywhere but in the previous posts, people talked about me pulling final/end numbers out of my bottom, so here's the longer version of the analysis with models in greater detail and visuals for anyone interested. No paywall, not essential to read for the purpose of the discussion here: https://open.substack.com/pub/hatedmoats/p/micron-what-happens-after-the-shortage)
