Beyond Chips: Why These 3 "Second Derivative" AI Stocks Could Soar in 2026

For the past few years, the AI boom has been a semiconductor story—Nvidia, TSMC, and other chip makers hogged the spotlight. But here’s the thing: while everyone’s fixated on the infrastructure layer, a whole new wave of opportunities is brewing in AI software. And that’s where the real alpha could come from in 2026.

Think of it this way—first derivative was chips and computing power. Second derivative? That’s the software and applications actually doing something with all that power. These three companies are positioned at the intersection of data, agents, and enterprise needs.

SoundHound AI: The Voice-AI Play That’s Going Mainstream

SoundHound AI (NASDAQ: SOUN) started as a voice recognition company, but it’s evolved into something way more interesting—a platform for voice-powered AI agents. The bet here is simple: AI agents need to understand natural language to work properly, and voice is the most natural interface we have.

The execution has been solid. Revenue more than doubled in the first nine months of 2025. The company has built real traction in two massive verticals: automotive (where voice assistants in cars are becoming table stakes) and restaurants (voice-ordering is actually a solved problem now).

Then there’s the acquisition of Amelia—this wasn’t just a tech grab, it was a customer grab. Suddenly SoundHound had relationships in healthcare, financial services, and retail. Its new Amelia 7 agentic AI platform is still rolling out, and gross margins are improving. The company expects to hit positive EBITDA soon, which would be a huge inflection point.

Salesforce: The “AI Loser” That’s Actually Winning

Salesforce (NYSE: CRM) got written off as an AI laggard. That take was always wrong. Here’s why: as enterprises scramble to deploy AI agents, they immediately run into a wall—dirty, siloed, unorganized data. And Salesforce owns the system of record for how most enterprises manage customer relationships, sales, and marketing.

The Informatica acquisition (a data integration platform) was the key move. Now Salesforce isn’t just storing data; it’s positioning itself as the truth layer that makes enterprise AI actually work.

But the real momentum driver is Agentforce, Salesforce’s AI agent offering. Last quarter, its annual recurring revenue hit $540 million with a mind-bending 330% year-over-year surge. That’s not hype—that’s real customer adoption. The company smartly introduced flexible pricing (seat-based or consumption), which removed friction for deals.

The valuation looks interesting too: sub-5.5x forward price-to-sales, around 20x forward P/E, and a PEG ratio under 0.65 (undervalued territory for growth stocks). It’s one of the few mega-cap SaaS names that doesn’t look stretched after its recent run.

Snowflake: The Data Warehouse That’s Becoming an AI Hub

Snowflake (NYSE: SNOW) operates a cloud data warehouse with a clever architecture—compute and storage are separated, which means you’re not locked into one cloud provider. Your data lives in Snowflake; you access it where you want, when you want.

The stickiness factor is huge. Once data is in Snowflake, switching costs are brutal. But that’s not even the main play anymore.

Snowflake Intelligence is the new game. Customers can now build their own AI agents that securely query Snowflake data. At the end of last quarter, over 1,200 customers were actively using this. The company is already running at a $100 million AI revenue annualized rate from this single product line.

The company that once looked like an AI afterthought is now firing on all cylinders. Last quarter revenue jumped 29%, and net revenue retention held steady at 125% over the trailing twelve months. Customer adds hit record levels. Those are the metrics of a company riding the right wave at the right time.

The Macro Thesis

All three of these companies share a common thread: they’re solving the actual, unglamorous infrastructure that makes enterprise AI work. Voice interfaces, data integration, secure querying—not flashy, but absolutely essential.

The market’s still pricing in semiconductor dominance. But in 2026, don’t be surprised when the second derivative—the software layer—starts to outperform. These three are worth watching closely.