A recent Silicon Valley expedition by Goldman Sachs analysts, spearheaded by George Tong, reveals a significant evolution in artificial intelligence implementation across corporate America. The team’s meetings with AI startups, established companies, venture capitalists, and academic experts from Stanford, UCSF, and UC Berkeley provided crucial insights into the industry’s trajectory.
The investigation comes at a time when AI-related capital expenditure is driving unprecedented data center construction nationwide, while investors carefully monitor signs of practical adoption beyond infrastructure development.
According to Tong’s Friday client note, artificial intelligence development is expanding beyond basic infrastructure into practical applications, with notable decreases in Large Language Model (LLM) costs, despite ongoing capital expenditure growth driven by increasing adoption and usage. The analysis suggests that university research efforts could further reduce operational costs, while noting that although software development expenses are declining, this creates both competitive challenges and pricing pressures.
For software companies, competitive advantages are emerging through various channels, including broad user distribution, engagement with power users that enables learning from feedback, workflow integration, and utilization of proprietary data sets.
The findings suggest that widespread generative AI adoption will accelerate beginning in 2026. Key trends identified include a shift from infrastructure-focused development (such as chips and cloud computing) toward end-user applications and industry-specific software solutions. While LLM costs continue to decrease, overall capital expenditure is expected to rise as usage expands, with academic research potentially accelerating efficiency improvements in AI models.
The transformation of the software development landscape presents both opportunities and challenges, as reduced costs enable faster development but intensify market competition and pricing pressures.
The research indicates positive implications for several major companies, including S&P Global, Moody’s, Iron Mountain, Verisk Analytics, and Thomson Reuters. Additionally, Goldman’s team has launched coverage of McGraw-Hill with a “Buy” recommendation and a $27 price target for the next 12 months, citing the company’s digital transformation efforts in education.
The analysis raises questions about whether current AI adoption rates can justify the substantial capital expenditure by hyperscalers, suggesting potential market corrections if expectations aren’t met.
The broader market context shows increasing complexity, with NASDAQ and NVIDIA testing crucial support levels and possible double top formation emerging. Morgan Stanley’s AI portfolio recently experienced its most significant daily decline since April, while volatility indicators show concerning signals, including elevated VVIX levels and widening technology sector volatility.
Market concentration in mega-cap stocks has reached extreme levels, with NVIDIA’s market capitalization now exceeding that of the entire Russell 2000 index. This has sparked debates about sustainability and potential risks in the current market structure.
Despite ongoing buying activity, market sentiment surveys and hedge fund positioning data remain subdued, creating a complex market environment. Goldman Sachs has recommended implementing hedging strategies through the remainder of the summer period, highlighting the uncertain nature of current market conditions.
The situation reflects a broader transformation in the technology sector, as companies navigate the shift from infrastructure
development to practical applications while managing market
expectations and competitive pressures in an increasingly AI-driven business landscape.