The Proof of AI Growth: Why Order-Backed Stocks Like Amphenol Are Outperforming Usage-Backed Giants Like Datadog

Datadog beat its earnings in the second quarter of 2026. Revenue rose 36% year-over-year to $1.12 billion. EPS came in above estimates. The stock fell roughly 19% in a single session.

The same quarter, Amphenol posted $8.8 billion in sales — up 55% year over year — and its stock kept climbing. Both companies are AI growth stories. Both delivered above expectations. The market treated them as opposite plays.

The difference has nothing to do with which company is growing faster. It's about where the growth proof lives.

Datadog's revenue comes from usage. Customers consume its monitoring and observability platform, and Datadog bills them for what they use. That model is flexible and scales well — until it doesn't. In the Q2 earnings call, management disclosed that its largest customer would reduce its usage, and folded that into third-quarter guidance implying 28–29% year-over-year growth, down from the 36% the quarter just delivered.

The market didn't sell off because Datadog failed. It sold off because the stock had been priced for a company whose growth would not slow. At roughly 20 times trailing revenue and 450 times trailing earnings — with a negative 0.67% operating margin — the multiple assumed perpetual acceleration. The moment that assumption cracked, the math caught up with the price.

Amphenol's revenue comes from orders. The company booked $10.732 billion in orders, up 94% year-over-year. Its book-to-bill ratio was 1.23:1. That number tells you something usage-based revenue can't: the next quarter's growth has already been contracted. Virtually all of Amphenol's IT datacom segment growth — 43% of total sales — is tied to AI infrastructure interconnects: copper and optical cables that connect the servers powering AI training and inference. You can't unorder a cable assembly the way you can throttle cloud usage.

The market rewards that certainty. Amphenol trades at roughly 55 times forward earnings — rich on paper — but the operating margin expanded 420 basis points year over year to 29.8%, driven by volume leverage and the CommScope acquisition that's performing better than expected. Management raised its full-year EPS accretion estimate from $0.15 to $0.30, doubling the upside from a deal it closed only two quarters ago.

This distinction between order-backed and usage-backed growth is the one that matters for anyone holding AI-related stocks. The market has been willing to pay extreme multiples for growth — but only when that growth looks irreversible. A backlog is irreversible. A customer reducing spend on a monitoring platform is not.

This explains why concentrated growth funds are being selective. Baron Fifth Avenue Growth Fund added Amphenol to its portfolio, citing the company's decentralized structure and compounding acquisition model. The fund still holds Datadog and Snowflake as core long-term positions — but it's the order-backed names, not the usage-backed ones, that drew fresh capital. The pattern is clear: even growth-focused capital allocators are distinguishing between growth that's contracted and growth that's hoped for.

Datadog and Snowflake have no such floor. Both are usage-based SaaS companies with rich multiples and no contracted backlog. Datadog's largest customer pulling back is a structural risk the market now has to price in. Snowflake, while still a core holding for the fund, carries a negative GAAP P/E and roughly 20 times revenue — a multiple that only makes sense if growth stays above 30% with margin expansion on track. One quarter of deceleration and that thesis needs to find a bottom.

The practical takeaway for a retail investor is this: when a growth stock trades at a triple-digit earnings multiple or 20 times revenue, the market is charging you for a future that hasn't happened yet. The question isn't whether the company is growing. The question is whether that growth is already on the books or still a hope. Amphenol's orders are on the books. Datadog's are not — and when the one customer who matters decides to spend less, the multiple collapses.

Datadog's recent decline has already been about 25% from its highs, down from an all-time high above $292. Whether that's a buying opportunity depends on one variable: whether the growth rate stabilizes above the pace the remaining multiple requires. At today's levels, the math is tighter, but the usage-based risk remains. A company that can't contract its next quarter's revenue is always one customer decision away from a re-rating.

Amphenol, by contrast, keeps raising guidance because the orders don't stop coming. The communications networks segment — wireless operators and equipment makers — is the one soft spot, with organic sales slipping. But even there, the AI infrastructure buildout is absorbing the weakness. The stock trades at a premium because the proof is visible. You can see it in the backlog, the margins, and the quarterly guidance that keeps getting lifted rather than lowered.

Growth alone doesn't justify a multiple. The proof behind that growth does.