From IBM's cybercrime data to Gallup's trust survey and insider selling at Nvidia and CoreWeave, the warning signs are real — even if one firm thinks the market is misreading the numbers
The artificial intelligence boom has produced a week's worth of contradictory evidence from strong corporate revenues, surging cybercrime costs, eroding public trust, and structured insider selling by the people who built the infrastructure.
Wall Street is trying to price all of it simultaneously, and financial advisors are fielding client questions that don't have clean answers yet.
On July 29, 2026, Meta Platforms, Inc. (Nasdaq: META) reported second-quarter revenue of $60.8 billion, a 28% year-over-year increase, according to the company's earnings release. Net income fell 14% to $15.8 billion, as expenses grew significantly faster than revenue. Operating margin compressed to 31%, down from 43% in Q2 2025. Full-year 2026 capital expenditure guidance stands at $125 billion to $145 billion, with second-quarter spending expected to roughly double first-quarter levels.
Mark Zuckerberg, Meta's founder and CEO, said in the earnings statement that "AI is accelerating our core business today, powering our next generation of products, and opening the door to entirely new enterprise opportunities."
Analysts remain broadly constructive, but pre-earnings analysis from Tickeron, published this week noted that Meta stock had declined approximately 9.7% year-to-date and sat roughly 25% below its 52-week high, reflecting persistent investor unease over whether infrastructure outlays will generate proportionate returns.
The public is turning skeptical — and clients are watching
Before advisors can address client questions about AI, they need to understand how those clients feel about the technology in the first place. The answer, as of mid-2026, is ‘cautious and increasingly doubtful.’
A newly released Bentley University-Gallup survey conducted in May with 3,270 US adults found that 70% of Americans now describe themselves as somewhat or extremely knowledgeable about AI, up from 64% in 2024. But growing familiarity has not translated into greater comfort with the percentage of respondents who believe AI does more harm than good increasing from 31% in 2025 to 39% in 2026, nearly matching the 40% recorded before the current AI wave began in 2023. Only 27% of Americans say they trust businesses "a lot" or "some" to use AI responsibly, down from 31% the prior year — halting two years of gradual improvement.
Among those aged 18 to 29, the share expressing at least some trust in businesses' use of AI fell from 30% to 20%, while the percentage saying they have no trust at all rose from 29% to 41%. Concern about job displacement intensified too: 79% of Americans said AI will reduce the number of US jobs over the next 10 years, up from 73% in 2025.
For financial advisors integrating AI into client-facing workflows, this is the environment in which those conversations take place. Firms that deploy AI tools without communicating transparently about where and how they are used and where human judgment remains, face a trust deficit that no efficiency gain will automatically resolve.
The cyberattack cost that changes the AI risk calculation
The negatives of AI are also highlighted in IBM’s newly released 2026 Cost of a Data Breach Report which reveals that one in four malicious breaches were AI-enabled in the period covered by the study, a 56% increase over the prior year.
Those AI-enabled breaches (primarily deepfake impersonation and AI-enabled malware) cost an average of $6 million per incident, roughly $1 million more than the global average breach cost of $4.99 million. More than 62% of AI-driven attacks targeted critical infrastructure sectors. Financial services breaches cost an average of $6.3 million; energy sector breaches averaged $5.2 million.
"What's changing is the economics of cyberattacks,” said Suja Viswesan, VP of IBM Security Software. “AI is making attacks faster and cheaper, while breaches keep getting more expensive. When organizations have an extended gap between discovery and remediation, that imbalance shows up directly in breach costs."
Companies that deployed AI and automation in their own security operations cut breach costs by an average of almost $2 million but one in four organizations had still not adopted those tools. More than 20% of organizations in the study reported a breach targeting AI models or applications directly, with compromised APIs and cloud misconfigurations the most common entry points.
The IBM data sharpens an already complicated conversation: AI is simultaneously the threat vector and a key tool for mitigation. That duality makes simplistic positioning difficult.
The insiders are quietly cashing out
Alongside the macro-level concerns, a more granular signal emerged from Bloomberg's reporting on July 29, which shows that the people closest to the AI infrastructure buildout have been reducing their personal exposure in size.
CoreWeave co-founder and chief strategy officer Brian Venturo led all insider sellers by market value from January through June 2026, according to Washington Service data cited by Bloomberg, having sold more than $1 billion in CoreWeave (Nasdaq: CRWV) shares under pre-arranged 10b5-1 plans since the Livingston, New Jersey-based company's IPO. CoreWeave shares traded as high as $138 earlier in 2026 and had fallen to approximately $67 at the time of Bloomberg's July 29 report. A CoreWeave representative told Bloomberg the co-founders are "making customary and prudent wealth-management decisions" and that their "long-term conviction in CoreWeave is unchanged."
Harvey Stevens, a director at Nvidia Corp. and managing partner of his family office, who first joined Nvidia's board in 1993, unloaded shares worth more than $400 million in the second quarter. Pre-arranged 10b5-1 plans are a standard tool for executives with concentrated equity positions, and structured selling does not in itself indicate a change in outlook. The scale of activity, however, across multiple AI infrastructure insiders in the same quarter, is the kind of signal advisors should be prepared to address when clients raise it.
A counterpoint worth hearing — but not overstating
Against the weight of those concerns, a July 2026 paper from Mercer offers a counter-argument.
Authored by Dmitri Smolansky, Head of Derivative Overlays at Mercer, and Max Becker, CFA, Global Multi-Asset Principal at Mercer, the paper titled "The cost paradox of AI adoption" argues that rising corporate AI bills are being widely misread as evidence that AI is becoming less economical. The authors contend the opposite is true: the price of a fixed level of intelligence is falling sharply, even as total AI spending rises, because newer reasoning and agentic systems accomplish far more work per task.
"There is no contradiction between rapidly increasing AI spending and falling cost of intelligence: AI is becoming more economical, letting companies squeeze more intelligence out of every dollar they spend," said Smolansky.
Mercer's February 2026 Artificial Intelligence in Asset Management Survey, covering 131 managers globally, found that more than half of firms reported live AI integration in at least one strategy, and that 91% of participants planned to increase AI-related activity over the following 12 months. A Morgan Stanley survey cited in the paper found an average productivity improvement of approximately 11.5% among companies using AI. Stanford University's 2026 AI Index reported gains of 14 to 26% in well-defined areas such as customer support and software development.
What advisors should take from July 29, 2026
The data from this single day does not offer a verdict on AI as an investment theme. It offers something more useful: a set of tensions that advisors can help clients navigate with clarity rather than hype.
The Gallup survey found one data point worth sitting with: the share of Americans who believe AI performs worse than a person when giving financial advice fell six percentage points between 2023 and 2026, from 49% to 43%, with more people now viewing AI as roughly comparable to human performance on that task, not better, but comparable.
It is a signal that the public's expectations of AI in financial services are evolving cautiously, and that advisors who can demonstrate where their human judgment adds value will be in a stronger position than those who simply adopt AI tools and assume clients will follow.
