(SMCI), (NVDA), (AMD), (MSFT)
One of my hedge fund buddies called me last week about Super Micro Computer (SMCI), sounding more worried than I've heard him since the 2008 crash.
"What's the real story behind these missing financials?" he demanded.
I had to smile. After decades of watching companies navigate regulatory waters, I've learned that sometimes the best opportunities come disguised as paperwork problems.
And SMCI's numbers tell a fascinating story.
Let's dive into what matters. Their preliminary Q2 FY25 earnings show revenue of $5.6-5.7 billion – a staggering 54% growth from last year.
Yes, gross margins have dipped to 11.8-11.9% from 13.3%, but here's what the market is missing: their capacity utilization is only 55% in the US, 60% in Taiwan, and a mere 1% in Malaysia.
That's not a company struggling to meet demand – that's a coiled spring waiting to launch.
The regulatory drama? They finally found a new auditor, confirmed no restatement of prior financials is needed, and committed to filing their delayed reports – the FY24 10-K and Q1/Q2 FY25 10-Qs – by February 25, 2025.
The market's initial relief sent the stock up 6% in mid-February, but there's more upside here.
Follow the technology trail. SMCI just started full-scale production of NVIDIA's (NVDA) Blackwell Rack-Scale Solutions for the HGX B200 system. Their servers are already certified for NVIDIA H100 and H200 GPUs.
But here's the game-changer: liquid cooling technology.
With over 30% of new data centers expected to adopt liquid-cooling systems in the next twelve months, SMCI is positioned perfectly.
They've smartly raised $700 million through 2.25% convertible senior notes due 2028 to expand these capabilities.
I've seen plenty of tech companies come and go in my years covering the market, but SMCI's approach to the AI infrastructure challenge is different.
When you're running advanced AI workloads, traditional air cooling is like trying to cool a blast furnace with a desk fan. Their direct-liquid cooling technology gives them a significant edge as data centers struggle with power density challenges.
The numbers tell the story: while traditional air-cooled data centers typically support 15-20 kW per rack, liquid cooling can handle upwards of 100 kW.
That's the difference between hosting basic enterprise applications and running complex AI workloads.
Despite this technology, the valuation disconnect is striking.
SMCI trades at 12x forward earnings while NVIDIA commands 30x and Advanced Micro Devices (AMD) sits at 18x.
Yes, management revised down their FY25 revenue guidance from $26-30 billion to $23.5-25 billion, but they're still targeting $40 billion by FY26.
Having watched tech cycles for decades, I know ambitious targets when I see them – and these are actually achievable.
Why? The shift from AI training to inference workloads changes everything.
While training demands massive computing power, inference needs efficient, scalable solutions – exactly what SMCI provides.
Management projects 65% annual revenue growth through FY26, moderating to 30% as the market matures, then settling around 10% long-term.
Margins might only see 10 basis points of expansion due to competition, but the volume growth more than compensates.
For those wondering about timing, here's my take: The Special Committee found no accounting fraud or inappropriate revenue recognition.
This isn't an Enron with imaginary Nigerian barges and "special purpose entities" – it's just a filing delay from a company that actually makes real products that actually work.
Once SMCI meets the Nasdaq deadline – and my sources suggest they will – we would most likely see a significant re-rating of the stock.
So, I recommend that you keep a close watch of the next few weeks. In this market, companies solving real AI infrastructure problems don't stay discounted for long.
Just keep your position size reasonable – even the best tech plays require disciplined risk management.
As for me, I'll be tracking this one closely and will alert you the moment I see a clear entry point.
After all, in tech, today's paperwork problem often becomes tomorrow's profit engine. Just ask my old friend who panic-sold Microsoft (MSFT) in 1987 over their messy IPO filing.
