Q2 Earnings Preview:

Heading into Q2 earnings season, the backdrop is unusually strong. Analysts raised S&P 500 EPS estimates by 3.4% during the quarter, reversing the typical pattern of estimate cuts. Companies have also been far more optimistic than normal, with 57% of firms issuing guidance delivering positive EPS outlooks versus a historical average of 41%. As a result, expected Q2 earnings growth has climbed to 23.3%, up from 18.8% at the start of the quarter, while expected revenue growth has increased to 12.2% from 9.5%.

The strength, however, has been concentrated rather than broad-based. Energy and Information Technology accounted for most of the upward earnings revisions, with Energy alone seeing a 61.5% jump in Q2 EPS estimates. If estimates hold, this would mark the second consecutive quarter of earnings growth above 20%, the seventh straight quarter of double-digit earnings growth, and the strongest revenue growth since Q2 2022. Analysts remain constructive beyond this quarter as well, forecasting earnings growth of 26.8% in Q3, 24.4% in Q4, and 24.1% for full-year 2026.

The setup is genuinely strong and genuinely narrow at the same time. A bar raised mid-quarter — against a decade of downward-drift habit — leaves little room for merely-good results: the index needs beats, not meets, to hold up. Revenue breadth is the healthiest part of the story, with all eleven sectors projected to grow; the EPS concentration in Energy and Tech is where a miss would do structural damage. Only three companies report in the coming week — the real test begins when the banks open the season.

Broadcom's Anthropic Opportunity:

Anthropic's annualized revenue run-rate has surged from roughly $9 billion at the end of 2025 to more than $30 billion just four months later. It has rapidly emerged as the enterprise AI leader, with Claude overtaking OpenAI across several enterprise adoption metrics. Based on my own experience, it also remains the best model for research, writing, and coding.

The obvious trade is to buy the companies powering Anthropic's AI infrastructure. But not all beneficiaries are equal.

Amazon, Alphabet, Microsoft, Nvidia, Broadcom, and SpaceX all stand to benefit from Anthropic's growth. The difference is how they benefit.

Amazon, Alphabet, Microsoft, and Nvidia are both strategic investors and infrastructure partners. They benefit through some combination of cloud revenue, hardware sales, and equity ownership. Those are meaningful tailwinds, but they are spread across businesses generating hundreds of billions of dollars in annual revenue and multiple AI growth vectors.

Broadcom offers a different proposition. According to Mizuho, Anthropic could contribute roughly $21 billion of Broadcom's AI-related revenue in 2026, rising to around $42 billion in 2027. If those estimates prove accurate, Anthropic could account for nearly one-quarter of Broadcom's revenue next year. More importantly, Broadcom isn't simply supplying compute. It is the custom silicon partner behind Google's TPUs, powering Anthropic's next-generation AI models. Once a hyperscaler commits to a custom silicon platform, those design wins become deeply embedded and increasingly difficult to replace.

Google is likely the second-largest beneficiary. It earns cloud revenue, owns a meaningful equity stake in Anthropic, and every new Claude deployment further strengthens its TPU ecosystem. But within a company generating hundreds of billions in annual revenue, Anthropic is incremental. For Broadcom, it could be transformational.

SpaceX also provides relatively direct exposure through its Colossus GPU lease, but the economics are very different. GPU capacity can be reallocated, and lease agreements renegotiated. Custom silicon is a long-duration commitment embedded deep within the AI stack, making switching costs significantly higher.

Broadcom's own results reinforce the opportunity. AI semiconductor revenue reached $8.4 billion in the latest quarter and is expected to reach $10.7 billion this quarter, with management expecting AI semiconductor revenue to exceed $100 billion annually by 2027. The company also exited the quarter with a $73 billion AI backlog, providing strong visibility into future demand.

The biggest risk is customer concentration. If Anthropic becomes one of Broadcom's largest customers, its success increasingly becomes Broadcom's success. It's also important to remember that Broadcom does not disclose customer-level revenue, so the Anthropic revenue figures are based on analyst estimates rather than company filings.

Bottom line: Anthropic's rapid rise is creating winners across the AI infrastructure stack. Amazon, Alphabet, Microsoft, and Nvidia all benefit through diversified AI ecosystems. Google is likely the second-largest beneficiary given its cloud business, TPU platform, and equity stake. But Broadcom stands apart. It offers the most direct, durable, and financially meaningful public-market exposure to Anthropic's continued growth. If Claude continues gaining enterprise share, Broadcom could be one of the biggest winners.

Key Takeaways From This Week's Earnings:

The market finally paid for beats outside of AI

This was a different week. General Mills, FactSet, MSC Industrial, and Constellation all beat, and all four rose. For the last month, only the AI names got rewarded for good results. This week the money went into food, financial data, industrials, and beer. That is the rotation everyone has been talking about, and it is now showing up in how earnings trade.

Nike beat, but read the fine print

Nike’s $0.72 looks like a blowout until you see that about $0.52 of it was a one-time tariff refund. Strip that out and the number is closer to $0.20. Revenue still fell. North America is getting better and China was less bad than feared, but the turnaround is showing green shoots, not a harvest, and management guided the next two quarters to flattish. That is why the stock dropped first and recovered later. The quality of a beat matters, not just the size.

Cost cutting is the new growth

General Mills jumped on its earnings day, but not because it is growing. Organic sales were flat and next year’s EPS is guided lower. The stock rose because the company laid out a plan to cut $3 billion in costs by 2030. You saw a version of the same thing at MSC Industrial, where margin expansion did the heavy lifting. When the top line will not cooperate, the market will pay for efficiency.

AI keeps compounding in the boring places

FactSet is not a name people think of as an AI story. But its subscription growth just accelerated for the fifth quarter in a row, as clients adopt the AI tools it has built into its platform. This is the quiet side of the AI trade. It is not only the chipmakers and the hyperscalers. The companies that sell data and software into finance are monetizing it too.

A broader, healthier week

In short, most of these names beat, and most of them went up, and the leadership was consumer, financials, and industrials, not tech. After a month where only AI worked, that is a healthy change. If this rotation holds into the second half, the market will look very different from the one we had in the first.