2Q26 earnings are the strongest in a while

U.S. earnings just keep getting stronger. After reaching 27% last quarter, S&P 500 2Q26 EPS growth is on track to hit 38% y/y, bringing the estimate for the full year up to 28%. This would mark the strongest year since COVID and the third year of double-digit earnings growth for the first time since the 2000s. Profit margins too are at a record high.

But that strength is concentrated in two themes: AI and oil. Just 10 companies are responsible for 77% of the earnings growth expected in 2Q26. The hyperscaler numbers are also being skewed by enormous gains from their investments in private AI companies. Excluding those, 2Q26 EPS growth would still be strong at 19%, in line with last quarter.

There are industries outside of AI seeing genuinely strong growth: aerospace & defense, luxury apparel, personal care products and medical equipment, to name a few. But their impact on the S&P 500 index is obscured by their lower weight and the magnitude of the AI and energy gains.

• Tech is once again behind the strength in S&P 500 earnings, but it’s not coming from the hyperscalers. After adjusting for investment gains, the hyperscalers are contributing just 8% of the EPS growth this quarter, down from 21% in 2025 and 41% in 2024, as capex continues to pressure profitability. Indeed, for the first time ever, their collective free cash flow turned negative this quarter. But a lot of that cash has flowed right into the semiconductor industry, which is on track to grow EPS by 134% y/y, driving 61% of the index-level growth. But narrative, not fundamentals, is driving returns. Semis earnings have come in 20% above expectations, but the industry is down 8% since the first report.
• Energy earnings growth estimates shot up from 0% before the war to 128% currently, thanks to the increase in oil prices. Brent rose an average of 53% y/y over the quarter, but rather than adding capacity to chase a short-term spike, management teams remain focused on shareholder returns over and the long-term cycle.
• Financials upside continues to be driven by AI. The boom in IPOs, equity & debt issuance, and stock gains is driving fee revenues in investment banking and asset & wealth management. Investment banking fees were up 26% y/y, driven by equity issuance, IPOs and M&A. Elsewhere, tax refunds and elevated gas prices are boosting consumer spending and credit card revenues. According to Chase data, credit card spending rose an average of 5.3% y/y in 2Q and a still strong 4.7% excluding gas stations.

Eventually, earnings growth will need to broaden out. The continued profitability of the semi industry depends on the continued profitability of the hyperscalers, which depends on monetizing consumer and enterprise AI. 

Consumer’s are rapidly adopting AI

Everybody loves generative AI. In less than four years since ChatGPT’s launch, 1 out of every 10 people in the world use it each week. It’s contributing over a third of U.S. GDP growth and has driven the equity market up 100% over the past three years, creating $30 trillion in wealth. 

While its impact is undeniable, AI’s value remains the subject of much debate. Everyone’s using it, but not everyone is paying for it. Even those who are might not be paying for all their usage. Only 6% of users are paying for ChatGPT, and the $20/month plan gives users access to about $700 worth of inference, or for just $200/month, you get a whopping $14,000 in inference.^[1] So, while OpenAI’s revenue more than tripled between 2023 and 2025, paying for inference eats up more than half of it. Throw in the exorbitant cost of training new models, and management doesn’t expect to make a dime until 2030^.[2]

But ultimately, companies, not consumers, will probably foot the bill.

In their 2025 report on enterprise AI, OpenAI wrote, “The majority of economically valuable activity takes place inside organizations…The revenue generated from solving these problems can help fund broad, free access to powerful AI for hundreds of millions of people worldwide.”^[3]

If the buck indeed stops with companies, the future of the AI trade rests heavily on their adoption. 

Enterprise adoption is much slower

Unfortunately, it takes a lot more than a snappy response to automate real work. Enterprise AI agents need access to companies’ proprietary data, documents, code, workflows and operational systems. All that information needs to be migrated onto the cloud and organized into databases that models can understand. Agents need to be connected to those databases, as well as operating systems like Windows, ERP and CRM. They also need to be provisioned with the correct data access, protected with cybersecurity and monitored constantly. In short, the real challenge in enterprise AI isn’t the model’s capabilities, it’s the system around it.

Microsoft’s CEO argued, “Models are an input, not some extraction of the knowledge of the enterprise…you got to keep your harness separate from the model. Now, when the harness will ensure that your memory, your context, all of that is external. That means, any given model at any given time is swappable.”

Companies need to walk before they can run

This complexity helps explain why enterprise adoption hasn’t progressed as quickly as consumer. According to Section AI, “56% of Americans say that they use AI, yet 85% of the workforce does not have a value-driving AI use case.”^[4] An MIT labs paper found that despite an over $30 billion investment in enterprise AI, 95% of companies were seeing zero return.^[5] Last year, McKinsey surveyed 1,993 companies globally, and while 88% reported using AI in at least one business function, less than 40% were seeing it translate to operating income.^[6]

Management teams have no shortage of ideas for AI use cases, but many don’t have tech stacks modern enough to implement them. For one, AI workflows typically require cloud infrastructure, but according to Amazon’s CEO, over 85% of global IT spending is still on-premises.

The CEO of Centene put it this way: “There is…a risk of spending a lot of money on AI and getting no return for it…that's not something that we can afford to do as a Medicaid-first company in a margin compressed environment…the ability to maximize AI is really predicated on the command you have of your data, the ability to build differentiated data products and to maintain and own the context layer…we aren't just going to deploy AI to talk about AI, we're going to deploy it where there is very clear, tangible return on that investment…we think the way to do that is by this foundational focus on data and context, in the short term.”

Enterprise software companies could benefit

Software companies could very much benefit from that foundational investment in areas like cybersecurity, data governance, workflow orchestration, identity management, error tracing and compliance. Even with their newfound coding efficiency, non-tech companies are unlikely to recreate all of these systems in-house. Indeed, despite being left for dead by the rest of the market, the software industry is seeing earnings growth hold steady at 19% y/y.

ServiceNow reported annual AI contract revenue over $1bn, on track to beat 2026 guidance of $1.5bn, and a 45% increase in deal value with first-time AI customers, driving an 8% rally the next day.

According to the CEO, “The path to value isn't just making AI, it's deploying AI securely across the enterprise. IDC forecasts spending on AI software is going to grow 53% this year, 17% faster than AI hardware. Whichever chip wins, whichever lab wins, whichever price per token regime prevails, the enterprise needs one governance layer of record for work, and ServiceNow offers needed certainty in an uncertain stack. Our platform is optionality on all AI outcomes, not a bet on anyone.”

Snowflake said 50% of new customers won in 2Q26 were tied to AI: “once they are on our platform, AI becomes a cornerstone of their strategy, powering 25% of all-deployed use cases with over 6,100 accounts using Snowflake’s AI every week.”

AI is also creating the need for new software systems altogether. Microsoft has developed more than four new products to orchestrate AI agents, which are being used by tens of thousands of customers including almost 90% of the Fortune 500:

“We are building an IQ layer that combines data with model capabilities to deliver the right context at the right time…[customers] are already grounding their agents in enterprise context with Foundry, Fabric and Work IQ…”

This quarter, they launched a new tool to give agents access to the web, which is even being used by ChatGPT itself. They also have another platform called Agent 365 to help companies monitor and control all their agents, which has almost 40 million registered so far. 

How companies are using AI

AI’s main impact for now has been on cutting costs

Even though companies aren’t ready to scale AI broadly, they are using it to reduce operational costs. S&P Global is targeting a 20% reduction in costs by 2027 through both AI and traditional productivity measures and is already 60% of the way there. Equifax’s AI productivity initiative is going so well they doubled the cost savings target from $75 to $150 million by 2028, or 5% of their current cost of sales. AI coding tools deserve a lot of the credit, generating anywhere from 10% of new code at Uber to 60% at Airbnb and 90% at General Motors.

It’s also boosting productivity in areas like customer service and document processing. Booking.com cut customer service costs by 10%, and Citi is saving over 1,700 hours per month using AI to process documents like trade confirmations. Some companies have even connected these improvements to job cuts.

Visa’s CEO explained, “We now have more than 150 AI-powered applications, and over the last 12 months, we have shipped more than 300 major product releases. Changes in the way we work and where we invest also impact how we operate the company. Today, we announced that we are eliminating roles with the majority being in our technology and product teams to ensure that we are continuing to position Visa for future growth.”

Importantly, a lot of these AI savings are being reinvested back into the businesses, rather than boosting margins in the short-term.

After describing their cutdown in coding cycle time, American Expresses CEO added, “Now, that's really not a savings because what that does is allows us to do more. And so that's what we're doing…we have a large backlog of technology projects.”

Similarly, AI is cutting down on Netflix’s content creation costs. Their new documentary series “American Experiment” contained 17 minutes of AI-enhanced footage that were produced twice as fast and at half the cost of previous options. But management emphasized that cost savings like these “will likely be reinvested into more content on the service, which fuels high-quality engagement and that whole kind of revenue-profit flywheel.”

It’s early days for AI products, but there are opportunities across sectors

• Consumer: No matter how much technology has evolved, the U.S. economy still runs on consumption, which drives 68% of GDP. So, it’s no surprise that companies are focused on monetizing AI through ad recommendations and agentic shopping. Meta’s AI-enhanced ad-recommendation models drove a 14% increase in ad impressions and a 12% increase in the average price per ad, while their new content generation algorithm increased time spent on Instagram by over 10%. Amazon’s AI assistant Rufus can search products, track prices and auto-purchase items when they hit a set price, with engagement up around 400% y/y. The number of customers using Walmart’s AI assistant Sparky increased by 100% in 1Q26 alone, which is lucky because they spend an average of 35% more than non-users. Google is working with retailers to enable Gemini to check out directly on their sites. Target and Steve Madden went live this quarter, and Sephora, Ulta, Macy’s and more are on their way.
• Financials: Data companies are enhancing their existing products with AI features. NASDAQ embedded AI into their compliance software, and management said a quarter of new bookings in 2Q26 were associated with AI use cases, adding, “We're in the very early innings of monetizing our AI capabilities, but we're very encouraged by the way that the clients are moving from free to a paid subscription.” Similarly, the number of customers on S&P Global’s LLM-ready APIs grew 70% in 2Q26 and call volume is up 5x. The CFO said they’re starting to charge for these additional features and “feel quite comfortable that the revenues are beginning to come through along with higher retention and higher sales.”
• Health care: Pharmaceutical companies will be able to use AI to exponentially increase the pace of drug development. The CEO of Revvity, which creates research and diagnostics tools, said the industry was still in the infrastructure buildout phase and compared it to laying fiber optic cables before the internet boom. Like in financials, companies are having success incorporating AI into existing products. GE HealthCare now has 100 FDA-listed AI-enabled devices and says these products are “in high demand,” contributing to “orders, revenues, and growth,” though management didn’t provide any numbers around that impact. Intuitive added “AI-driven case insights” to their da Vinci 5 surgical robots, and Thermo Fischer similarly added “AI-driven capabilities” for data interpretation to their Orbitrap mass spectrometry machine.

Almost every company talks about the ways they’re incorporating AI into products. The early results are always very promising, but it’s still too early to see tangible impacts on earnings. 

Investment implications

• Much of the AI’s ROI depends on enterprise adoption, which is still in the early stages.
• Enterprise adoption is less about the models and more about the data layers around them, which creates an opportunity for enterprise software companies.
• Companies are realizing some cost savings through AI automation but re-investing most of it back into their businesses.
• These cost savings applications are likely low-hanging fruit in the medium-term, so their benefit to margins will likely be competed away. S&P 500 margins are already at 20-year highs, and companies can’t keep accreting operating leverage forever. Some of the benefits will need to accrue to consumers.
• Agentic commerce may be the first AI use case driving material revenue growth.
• investing in AI is a risk, and thus a view on each company’s management team and strategy is essential. 

¹ SemiAnalysis. “The Math Doesn’t Work.” June 11, 2026.

² Berber Jin and Nate Rattner. “An Inside Look at OpenAI and Anthropic’s Finances Ahead of Their IPOs.” Wall Street Journal. April 5, 2026.

³ OpenAI. “The state of enterprise AI.” December 17, 2025.

⁴ Section AI. “The AI Proficiency Report.” January 2026.

⁵ Challapally, Aditya, Pease, Chris, Raskar, Ramesh, Chari, Pradyumna. (Jul 2025). The GenAI Divide, State of AI in Business 2025. MIT Nanda

⁶ McKinsey & Company. “The State of AI in 2025: Agents, Innovation, and Transformation.”

By Meera Pandit and Katie Korngiebel - July 31, 2026

ccf33131-8b8f-11f1-9510-d5001bff916f