"AI infrastructure" is one of the most-searched investing categories in 2026. It is also one of the most misused. Investors pile into it as though it is a single trade, but the stocks inside it live in completely different financial realities.
JPMorgan projects that more than $5 trillion will be spent on global data center and AI infrastructure over the next five years, with annual data center funding needs in 2026 running around $700 billion, a figure confirmed across Q1 2026 hyperscaler earnings calls. That spending is real. What is not real is the assumption that every company tagged "AI infrastructure" benefits from it equally.
Stoxcraft tracks 3,487 stocks across 156 industries. When you score every name in this category using health, risk, and performance metrics, the group splits into four distinct profiles. Each one demands a completely different way of thinking.
The four archetypes inside "AI infrastructure"
Before looking at individual stocks, it helps to understand how Stoxcraft's scoring system organizes them. Three scores do the heavy lifting.
The Health Score measures fundamental quality relative to sector peers. A score of 8.3 means the company sits in the top tier for balance sheet strength, profitability, and cash generation. The Performance Score captures price performance across multiple time horizons compared to the full Stoxcraft universe. The Risk Score uses inverse logic: a score of 10.0 means maximum risk, not maximum quality. High volatility, large drawdowns, and elevated beta all push it upward.
The combination of these three scores produces recognizable patterns. Here is how the AI infrastructure category maps to four of them:
- Quality Compounder: High Health, strong Performance, Low Risk. The business earns its price over time.
- Picks-and-shovels play: High Health, strong Performance, sector-validated revenue. Profits, not promises.
- Turnaround Candidate: Improving Health, explosive revenue growth, negative net income. A bet on trajectory, not today's numbers.
- Momentum Play: Near-maximum Risk Score, extreme valuation, genuine fundamental story underneath. The trade is real; so is the danger.
NVIDIA: the quality compounder that set the floor
NVIDIA (NVDA) is the clearest Quality Compounder in the AI infrastructure space. Its Health Score of 8.3 reflects a business that generates cash at scale, carries manageable leverage, and sustains margins most technology companies will never reach.
NVIDIA's fundamental profile in plain numbers
The figures behind the score are not ambiguous:
- Net profit margin of 53%, placing NVDA in the top percentile of its sector
- Return on assets of 61.6%, reflecting exceptional capital efficiency
- Altman Z-Score of 69.5, which signals zero financial distress risk
NVIDIA reported record revenue of $81.6 billion for Q1 fiscal 2027, up 85% year over year, with data center revenue reaching $75.2 billion for a 92% increase from the prior year, and guided Q2 revenue to $91 billion. Those are not growth rates manufactured by narrative. They are driven by real hardware orders from hyperscalers converting $700 billion capital plans into chips, racks, and cables.
Why NVIDIA's Health Score held through every pullback
NVDA has sold off sharply at multiple points over the past two years. Its Health Score barely moved during any of them. That is the point of the metric. It measures the underlying business, not the stock's mood. A company with 53% net margins and a 61.6% return on assets does not suddenly become a weak business because sentiment shifts for six weeks.
The floor is supported by fundamentals that most competitors will not reach in a decade. Investors who understand that distinction tend to behave differently during drawdowns.
Applied Materials: the picks-and-shovels play with proven numbers
Applied Materials (AMAT) does not build AI chips. It builds the machines that make AI chips possible. That distinction matters when reading scores.
AMAT carries a Health Score of 8.9 and an overall rating that places it inside Stoxcraft's Top 25. It is the textbook picks-and-shovels position: selling into the AI capex cycle without concentrating all exposure in a single product line.
AMAT's Q2 2026 results confirm the thesis
Applied Materials reported record Q2 2026 revenue of $7.91 billion, up 11% year over year, with a GAAP gross margin of 49.9% and record GAAP EPS of $3.51, its fourth consecutive earnings beat. Management then raised its calendar 2026 semiconductor equipment growth forecast to more than 30%. This is not a company riding a theme. This is a company translating the AI infrastructure buildout directly into its income statement.
The Health Score of 8.9 reflects that. AMAT ranks near the top of its sector peer group for financial strength. Its Risk Score is moderate, consistent with a business that has a diversified customer base and does not rely on any single hyperscaler's quarterly capex decision.
Why AMAT's position differs from pure-play chip stocks
The structural difference matters. AMAT sells capital equipment to chip fabs, and those fabs are expanding across advanced logic, DRAM, and advanced packaging simultaneously. When one segment softens, the others provide a buffer. That diversification is part of what keeps the Health Score elevated through cycle fluctuations, and it is why AMAT belongs in a separate risk conversation from chip designers.
Nebius Group: the turnaround candidate with explosive revenue
Nebius Group (NBIS) is a completely different type of stock. It is not a mature compounder. It is a company in the middle of a transformation, and the scores reflect that honestly.
Its Health Score of 5.1 represents a meaningful improvement from a low of 3.5 in recent periods. That level is not strong in absolute terms, but the direction matters as much as the current number. A rising Health Score on a company with this kind of revenue trajectory is a signal worth watching carefully.
The NBIS growth numbers and the risk they carry
The growth data is dramatic:
- Revenue grew over 770% across three years
- Three-year CAGR above 106%
- Current Ratio of 9.6, indicating strong short-term liquidity
- Total debt of approximately $50 million, conservative for a company at this stage
The risk is equally clear. Net income sits at negative $394 million. Nebius Group targets $7 billion to $9 billion in annualized revenue by end of 2026, implying growth of 521% to 900% from its recent starting point, with analysts attaching upside targets above 90% from earlier 2026 price levels. That is a target, not a guarantee. The gap between today's losses and tomorrow's scale is exactly where the risk lives.
What the Turnaround Candidate profile means in practice
A Turnaround Candidate is not a broken stock. It is a stock where the health trajectory is more important than the current health level. NBIS has the liquidity to fund its growth phase without near-term balance sheet stress. The test is execution. Revenue growth of this magnitude is real. Sustaining it while narrowing the net loss is the question that separates successful turnarounds from expensive ones. Tracking the Health Score each quarter is the most direct way to monitor whether the story is converting.
Astera Labs and AST SpaceMobile: where risk scores reach the ceiling
The final group is where the scores deliver their most direct warning. Both Astera Labs (ALAB) and AST SpaceMobile (ASTS) carry Risk Scores at or near the maximum. That is the system flagging extreme volatility, massive drawdowns, and elevated beta in a single number.
What the two companies share ends there. They are risk-maximum for entirely different reasons.
Astera Labs: proven fundamentals inside extreme price risk
ALAB has a Health Score of 7.2, which is a genuinely solid number. The business designs semiconductor connectivity solutions for AI infrastructure, and the underlying financials are real. Q1 2026 revenue hit $308 million, up 93% year over year, and the company is profitable.
The Risk Score of 10.0 is not about the business model. It is about the stock price. ALAB fell roughly 60% from its late-2025 peak before recovering sharply in mid-2026. Its Altman Z-Score of 135.5 signals an exceptionally strong balance sheet with no insolvency risk. But at a forward P/E above 260 times, any execution miss, guidance trim, or macro shift creates instant and severe drawdowns. According to 26 analysts polled by S&P Global, Astera Labs carries a consensus rating of Buy with an average 12-month price target of $244.97, well above where the stock traded through much of early 2026. Analyst confidence in the long-term story is real. The path to that target will not be smooth.
AST SpaceMobile: a pre-commercial network with a long runway ahead
ASTS is a different category of risk entirely. The company is building a space-based cellular broadband network that connects directly to ordinary smartphones. The technology ambition is extraordinary. The financials reflect how early in that journey the company still is.
ASTS carries a Risk Score of 9.9 and a net loss exceeding $487 million on trailing twelve-month revenue of approximately $85 million. AST SpaceMobile guided for full-year 2026 revenue of $150 million to $200 million, with approximately half of that coming from existing contracted backlog, backed by $3.5 billion in cash and over $1.2 billion in contracted commitments from mobile network partners. That represents genuine commercial progress from a near-zero starting point.
But solvency is not the same thing as commercial proof. An investor in ASTS is making a long-duration bet on a network that does not yet exist at full scale. The Risk Score of 9.9 is the system's honest, direct description of that position.
What these profiles say that headlines do not
Headlines describe a supercycle. Scores describe what each company is doing inside it.
NVDA has the margins and balance sheet to sustain any drawdown and recover. AMAT is converting the macro trend into record quarterly results with diversified revenue. NBIS has the liquidity to fund its transformation and the growth rate to justify optimism, but not yet the profitability to call the turnaround complete. ALAB has proven fundamentals wrapped inside a Risk Score that reflects genuinely extreme valuation risk. ASTS is a long-duration infrastructure bet backed by cash and contracted revenue, but still far from commercially proven at scale.
Every one of those is a different position requiring a different risk tolerance, time horizon, and monitoring discipline. The AI infrastructure category label tells you none of that.
For broader market context, the Stoxcraft news article on what is driving this AI-fueled market expansion and the in-depth look at how NVIDIA and AMD are shaping the next tech decade both add useful context to this category.
