**SMCI DD:**
**Position**
Long SMCI.
Not financial advice.
I made hundreds of thousands on position then got nuked in 7 days
Gpt 5.5 analysis confirming my thesis citing Charles keynote at computex
**The Setup**
For the last year the market has been treating Super Micro Computer (SMCI) like a glorified server assembler.
The thesis goes something like this:
Nvidia gets all the profits.
SMCI screws GPUs into boxes.
Margins compress.
Competition increases.
Earnings disappoint.
Reasonable thesis.
The problem is that after spending the last few weeks digging through customer quotes, earnings calls, industry reports, and Charles Liang’s Computex 2026 keynote, I’m increasingly convinced the market is looking at the wrong business.
The question isn’t:
How many servers does SMCI sell?
The question is:
Is SMCI becoming the deployment layer for AI factories?
If the answer is yes, the stock is materially undervalued.
**The Computex Clue Nobody Is Talking About**
Everyone focused on the rack shipment number.
Charles said:
“In the last nine months, last three quarters, we shipped more than 8,000 racks so far.”
People immediately argued whether those were:
GB300 NVL72 racks
HGX racks
mixed AI racks
Honestly I think that’s missing the point.
Go read the keynote.
He barely talks about servers.
Instead he spends 45 minutes talking about:
deployment speed
liquid cooling
rack integration
power infrastructure
software
networking
lifecycle support
turnkey deployment
He repeats the phrase:
time to online
about twenty times.
Why?
Because the customer isn’t buying hardware.
The customer is buying:
faster monetization of AI capex.
**The Most Important Quote**
Not the 8,000 racks.
Not Rubin.
Not GB300.
This:
“This year can be 40 billion and next year much more than 40 billion.”
Read that again.
Much more than $40B.
The CEO is standing on stage publicly telling investors that next year’s revenue could be substantially larger than $40B.
Most people ignored it.
**The Real Bull Thesis**
Look at what SMCI is selling now.
Old SMCI:
Motherboards
Servers
Storage
New SMCI:
Rack-scale integration
Liquid cooling
CDUs
Power shelves
Battery backup
Networking
SONiC software
Cluster software
Deployment services
Lifecycle support
That is a completely different business.
A server company gets valued like a server company.
An AI infrastructure company gets valued differently.
**Why Margins Matter**
Everybody is obsessed with revenue.
Wrong metric.
The entire investment case comes down to one thing:
Gross margin.
The market currently assumes SMCI remains a low-margin pass-through vehicle for Nvidia.
But if SMCI is capturing value from:
cooling
integration
deployment
software
support
then margins should rise.
That’s exactly what we’ve started seeing with AI infrastructure businesses like Dell and HPE.
The next earnings report is not about revenue.
It’s about margins.
Here’s my framework:
10-11% gross margin:
Market is right.
12-13% gross margin:
Interesting.
14-15% gross margin:
The thesis is working.
16%+ gross margin:
The market is dramatically wrong.
**Rubin Is More Important Than People Realize**
Charles also said:
“We already shipped some samples to certain customers.”
This matters.
It means:
qualification underway
customer testing underway
production preparation underway
The market constantly worries about transition risk.
This suggests SMCI is already deeply embedded in the Rubin deployment cycle.
**Then Everything Changed Today**
Today SMCI announced up to $7B in equity and equity-linked financing.
The stock got obliterated.
Everyone saw:
DILUTION.
And yes, dilution is real.
But almost nobody read the reason.
Management disclosed approximately:
$39 billion of AI server orders from more than 20 customers in recent weeks.
That’s the important number.
Not $7B.
$39B.
Think about how absurd that is.
A company talking about $40B annual revenue just disclosed roughly $39B of recent orders and is raising capital because they need inventory and components to fulfill them.
Suddenly the Computex comments make sense.
Suddenly the manufacturing expansion makes sense.
Suddenly the urgency around deployment speed makes sense.
**Why The Market Panicked**
The market interpreted the announcement as:
“SMCI needs money.”
I interpret it as:
“Demand is exceeding the balance sheet.”
Those are not the same thing.
A bad company raises money because nobody wants its product.
A great company sometimes raises money because everybody wants its product.
The question is whether these orders are profitable.
Which brings us back to margins.
Again.
Margins.
Margins.
Margins.
**What I Think Happens Next**
Scenario 1: Bear
The orders are real but margins stay terrible.
Gross margin remains around 10%.
SMCI becomes a low-margin pass-through business.
Stock struggles.
Scenario 2: Base
Revenue ramps.
Gross margin reaches 13-14%.
Investors begin to realize AI factory deployments are more profitable than expected.
Stock rerates.
Scenario 3: Bull
Revenue exceeds $45B.
Gross margins push into the mid-teens.
Rubin ramps successfully.
Investors stop viewing SMCI as a server assembler.
They start viewing it as critical AI infrastructure.
Stock goes much higher than consensus expects.
**Final Take**
The market spent today asking:
How much dilution are we getting?
I think the more important question is:
Why did management feel comfortable raising $7B immediately after telling investors demand is exploding?
The answer may be that SMCI sees a once-in-a-generation AI infrastructure buildout and is willing to dilute shareholders to capture it.
If margins don’t improve, the bear case wins.
If margins do improve, today’s selloff may eventually look like the market confusing a financing event with a deterioration in demand.
The next earnings report will decide which side is right.
Until then, I think the most important number in the entire SMCI story is not revenue, rack count, or backlog.
It’s gross margin.