$ALAB Dropped 55% From Its High. Is This the AI Stock Everyone's Sleeping On?

The AI Trade Is Shifting... and Most Investors Are Looking at the Wrong Layer

Everyone's obsessed with GPU stocks. Nvidia, AMD, Broadcom... the usual suspects. But here's what most people miss: GPUs are useless if they can't talk to each other.

Think about it. You've got thousands of chips crammed into a data center rack and they all need to move data between each other at insane speeds. That's the bottleneck nobody talks about. And one company basically owns the solution.

I found $ALAB while researching the AI infrastructure supply chain... and the numbers stopped me cold. Revenue up 115% in a year. Profitable. Debt-free. And Amazon just locked them into a $6.5 billion deal.

So I ran it through my 10x Stock Checklist: My Exact 47-Point Analysis Framework to see if the hype holds up. Here's what I found.

Astera Labs makes the connectivity chips that move data inside AI data centers. PCIe retimers, CXL memory controllers, Ethernet smart cables, fabric switches... basically the plumbing that ties everything together.

If Nvidia makes the brain of AI... Astera makes the nervous system.

And unlike most "AI plays" that are still burning cash and hoping, $ALAB is already printing money.

The Numbers:

• Price: $118 (as of April 2026)
• Market Cap: ~$20 billion
• FY2025 Revenue: $853 million (+115% YoY)
• FY2024 Revenue: $396 million (+242% YoY)
• FY2023 Revenue: $116 million
• 3-Year Revenue CAGR: ~120%
• Gross Margin: 75.7%
• Free Cash Flow: $282 million (+175% YoY)
• FCF Margin: 33%
• Cash on Hand: $1.2 billion
• Total Debt: $31 million (basically zero)
• Forward P/E: 48x
• P/S: 24x

That FCF number is what caught my eye. $282 million in free cash flow on $853 million in revenue... that's a 33% margin. For a company that was basically pre-revenue three years ago. That's rare.

Why does this matter now?

Three things happening at once.

First... the entire data center industry is upgrading from PCIe Gen 5 to Gen 6. This isn't optional. Gen 5 is already saturating under AI workloads. Astera has the most complete Gen 6 portfolio in the market. Every hyperscaler upgrading their racks needs what $ALAB sells.

Second... Amazon signed a $6.5 billion multi-year deal to use Astera's Scorpio switches for their Trainium AI chip clusters. That single deal is 7x Astera's current annual revenue. It's the kind of anchor contract that transforms a company.

Third... the AI connectivity market is projected to grow from about $2.5 billion today to $25 billion in five years. That's a 10x TAM expansion. And Astera is the market leader in merchant connectivity solutions.

What could go wrong?

Here's the catch... Broadcom is coming.

Broadcom ($1.6 trillion market cap) is building competitive PCIe 6 products. Marvell already launched the first 260-lane PCIe 6.0 switch. These companies have 50-80x Astera's resources. If they decide connectivity is strategic enough to prioritize, $ALAB's moat gets a lot thinner.

And the insider selling is hard to ignore. $550 million in insider sales over the last 12 months. Zero insider buying. CEO, COO, CFO... everyone selling. Some of this is post-IPO vesting. But zero buying at a stock that's down 55%? That bugs me.

The Amazon deal is also a double-edged sword. It compresses gross margins by about 200 basis points and creates massive customer concentration. When one customer represents 7x your annual revenue... they have all the leverage.

📚 Recommended Resources

10x Stock Checklist: My 47-point System

6 Simple Steps Spotting Undervalued Stocks

Earn $500/Month with Dividend Stocks

3 Swing Trading Strategies for Predictable Gains

The biggest one: Scorpio X-Series moving to high-volume production in H2 2026. Scorpio fabric switches are Astera's highest-value product and they address a $20 billion market by 2030. Right now they're in initial production. If the ramp goes well... this is where the real revenue acceleration happens.

Beyond that I'm watching for:

• A second hyperscaler mega-deal. If Google or Microsoft signs something similar to Amazon's $6.5 billion contract, the stock re-rates instantly
• PCIe Gen 6 adoption inflection across the industry in 2026-2027
• CXL 3.0 memory pooling going mainstream via Astera's Leo product... this could be an entirely new multi-billion revenue stream
• Continued beat-and-raise quarters. Q1 2026 guidance already beat consensus by $30 million. If that pattern holds, forward estimates keep climbing

Let's be honest. $ALAB isn't cheap.

• Trailing P/E: 97x (sector range: Nvidia 46x, Broadcom 60x, AMD 78x)
• Forward P/E: 48x (in line with Broadcom at 57x... cheaper than AMD)
• PEG Ratio: ~1.67 (above the ideal 1.5 but not crazy for this growth rate)
• P/S: 24x (similar to Nvidia at ~25x)
• ROE: 18.8% (was negative two years ago... rapidly improving)

The PEG ratio tells the real story. At 1.67 it's saying the market expects earnings growth to sustain around 29% annually. Given they just grew revenue 115% and guided above consensus... that growth expectation feels conservative if anything.

But if growth slows to 20%? This stock gets punished hard at these multiples. That's the risk you're taking.

Analyst consensus:

• 15 Buy / 6 Hold / 1 Sell (68% Buy)
• Average price target: $202-212 (~70% upside)
• High target: $275 (Citi)
• Low target: $155

Sector trend: Strong uptrend. Data center capex heading to $1.7 trillion by 2030. PCIe switch market growing at 10.6% CAGR. No signs of slowdown in AI infrastructure spend... though tariff uncertainty adds near-term noise.

I'm setting clear tripwires:

• Gross margin drops below 72%. That means the Amazon deal margin compression is structural, not temporary. Trim.
• Scorpio production delays. If high-volume ramp slips past 2026... the platform thesis breaks. Trim aggressively.
• No second hyperscaler deal by mid-2027. One anchor customer is a feature. Only one anchor customer is a risk.
• Broadcom or Marvell wins a major PCIe 6 design win at a top hyperscaler. That proves the moat is penetrable. Re-evaluate.
• Revenue guidance misses consensus. The beat-and-raise pattern is what supports the premium. If it breaks, exit.
• Hyperscalers cut AI capex guidance. $ALAB is a picks-and-shovels play. No shovels needed if nobody's digging.

$ALAB scored a 7.6/10 on my DVR framework. That puts it solidly in "strong watchlist" territory but not an immediate all-in.

Here's my thinking. The stock is down 55% from highs... so the entry is way more attractive than it was six months ago. The growth is real. The products are real. The Amazon deal provides a revenue floor. But the insider selling and competitive threats keep me from pounding the table.

I'm starting a small position around $115-120 and adding if:

• Scorpio ramp confirms in Q3/Q4 earnings
• Insider selling slows or reverses
• A second hyperscaler deal materializes

Position sizing: small to start. This is a high-conviction sector bet but the individual company risk from Broadcom/Marvell competition is real. I'd size it at 3-5% of a growth portfolio max until the competitive picture clarifies.

If you want to see exactly how I scored this one and run the same analysis on your own picks, grab my 10x Stock Checklist. It's the same 47-point framework I use for every stock on the watchlist.

$ALAB is one of the most interesting AI infrastructure plays I've found this year. The 55% drawdown created a real entry point for a company growing revenue at 120% CAGR with 33% free cash flow margins. The Amazon deal provides a revenue floor most small caps would kill for.

But I'm not blind to the risks. Insider selling at these levels is a yellow flag. Broadcom entering the space is a real competitive threat. And the valuation still assumes near-perfect execution.

My target: 3-5x in 3 years if Scorpio ramps and they land a second mega-deal. 10x requires them to become a $5 billion+ revenue platform company while fending off trillion-dollar competitors. Possible... but I'm not betting the farm on it.

Watch the Scorpio ramp. Watch the margins. Watch what insiders do, not what analysts say.

Original Tweet 👉