Why AI concentration matters more than AI credibility
The setup is straightforward: the market is not at risk because AI is imaginary. It is at risk because too much of the index now depends on investors getting the AI payoff right, too soon. AI-infrastructure beneficiaries are expected to account for roughly half of 2026 S&P 500 earnings growth. That turns the market into a concentration trade. If AI profits prove durable, this concentration pays. If they do not, the correction does not need to dispute AI's long-term value. It only needs to question how much of this year's earnings power should be attributed to AI right now.
Momentum has reset sentiment quickly
That confidence has built fast. After a rough start that included a 9% decline from the start of the year through the end of March, the Nasdaq reversed course and rose sharply from early April. Semis led the rebound, while beaten-down software and cybersecurity names also staged a relief rally after Microsoft was down close to 20% for the year before surging 13% in a week. The behavioral signal is clear: investors are rotating back into tech narratives as soon as momentum returns, even though the deeper debate over AI monetization is still unresolved.
That is where the risk sits. AI may be transformative, but the market is already acting as if the earnings stream is more certain and more widespread than the evidence yet shows. With sharp increases in momentum and narrow market breadth flagged as cautionary signals, the next earnings season matters. Investors are not just pricing AI adoption; they are pricing minimal execution risk.
Three AI-linked buckets where expectations look ahead of the receipts
The market is no longer debating whether AI matters. It is debating who gets paid first, and that is where expectations get dangerous. Investors tend to buy the story before the income statement proves durable, especially when momentum resets sentiment quickly.
1) Software after the "SaaSpocalypse" scare
What the crowd sees: A sector that was feared would be disrupted by AI is suddenly posting better numbers, which feels like proof that software companies can monetize the wave rather than suffer under it. On June 1, the software sector had one of its strongest trading sessions of the year. MongoDB jumped 20.4%, and the move was tied to results that challenged fears of a "SaaSpocalypse."
The bias: This looks like relief rallying mixed with confirmation bias. Investors were primed for an AI-led earnings collapse in software. When a few names beat, the market overcorrects from "AI will destroy software" to "AI will lift software." That emotional swing can make investors treat a few good quarters like broad, lasting immunity.
What is still missing: Evidence that AI-related revenue is material enough to support richer multiples across the category, not just in headline beats. Recent results helped, but the key question is whether AI features are creating new demand or simply helping firms defend pricing. Bears can still argue the rebound is sentiment-led for now.
Cybersecurity and enterprise software stocks surged last week after a brutal start to 2026, which shows how quickly fear can flip into FOMO. The controversy now is simple: are investors buying a turnaround in monetization, or buying back into names that were merely hated for too long?
2) Semis after a massive run
What the crowd sees: If AI is real, chips are the pick-and-shovel winner. The market is acting like that conclusion needs no further validation. After the SOX was up more than 80% year-to-date, investors were clearly pricing continued tight demand and aggressive spending.
The bias: Recency bias and herd behavior do the heavy lifting here. When a trade compounds this fast, investors start treating acceleration as permanence. Anyone who questioned semis earlier in the year looks late, so the crowd keeps leaning harder into the same narrative instead of asking what is already priced in.
What is still missing: A cleaner line of sight from chip demand into sustained customer returns. Semis can stay expensive if hyperscalers keep building. But they also remain vulnerable to one bad signal on utilization, capex discipline, or demand durability. Broadcom's warning pumped the brakes on the AI rally mattered because it reminded everyone that expectations were stretched. Semis do not need weaker long-term demand to rerate lower; they only need slower optimism.
3) "Must-own" AI buzz lists
What the crowd sees: A long list of familiar AI-linked names getting pushed together in forums and social feeds. In the last 10 days of tracked recommendations, the top 30 included AMZN, GOOGL, AMD, NVDA, AVGO, and PLTR, among others. The appeal is obvious: if you cannot buy the whole AI theme, at least buy what everyone else is watching.
The bias: This is social proof turning into herd behavior. When the same cluster of names keeps showing up, investors start treating attention as validation. That creates a feedback loop where popularity feels like proof of monetization.
What is still missing: Evidence that every widely mentioned name will convert AI spend into durable earnings. Some spend is productive. Some may be performative. The broader concern is not that AI lacks value; it is that excessive spending, hype, and unrealistic expectations can create bubble conditions even in a genuinely transformative industry. The risk is not sector disruption so much as widely dispersed outcomes.
How to separate AI leadership from AI overpaying
The next earnings cycle is a filter, not a final verdict on AI. With AI-infrastructure beneficiaries are expected to account for roughly half of 2026 S&P 500 earnings growth, the market is still leaning on a narrow profit base. That creates a tight positioning window: if leaders show durable customer returns, the rally can keep winning; if not, the market can quickly stop rewarding narrative and start punishing overpaying.
The trigger to watch
Treat the next guidance check as the trigger. Broadcom already pumped the brakes on the AI rally, a reminder that expectations are stretched. The real danger is not necessarily an immediate collapse in demand. It is a crack in confidence caused by heavy spending without clear monetization.
What to monitor
- Software: Whether AI-related demand proves recurring rather than confined to a strong reporting quarter.
- Semis: Whether customer economics keep supporting aggressive capex after a huge run.
- Buzz-list platforms: Whether heavy AI spend is producing clearer productivity or monetization outcomes.
Positioning lens
Stay selective where index returns still depend most on AI-infrastructure beneficiaries and where narrow market breadth has pushed momentum too far, too fast. Favor names showing earnings quality over names that merely sit on the buzz list.
What would invalidate this skepticism
This cautious view would be wrong if the next round of reports shows broadening proof: software keeps converting AI into recurring demand, semis see customer spending translate into durable returns, and platform names begin to justify heavy AI spend with clearer monetization outcomes. If that happens, the market may simply be getting ahead of the curve rather than overreaching.