MSCI Inc. (MSCI)

MSCI is a leading provider of investment-decision support tools worldwide. Its line of products and services includes indices, analytical tools, and ESG and climate research products. The stock is a component of the S&P 500. The company has approximately 6,300 employees.

10 Questions About Artificial Intelligence: Part 3 Artificial intelligence (AI) has been and remains one of the fastest-growing technology developments in history. Just a few years after becoming broadly known, AI represents a significant portion of the global economy and is now the preeminent focus among investors. Developments in AI have progressed and continue to unfold rapidly, and new revelations on one day - say, the release of Meta Platforms Inc.'s Muse model - can make preconceptions about AI seem out of date. Based on input from clients, investors, industry experts, and our own analytic team, Argus has put together what we see as 10 current key and enduring questions about AI, along with our responses. We've broken these 10 questions into Parts 1, 2, and 3. Part 3, presented below, addresses four key questions about risks the market may be underpricing, regulatory trends that may impact long-term value of AI investments, growing concerns that advanced models may pose an existential threat, and the AI investing climate in the wake of existential concerns. We welcome feedback from clients and will continue to refine our 10 questions about AI in real time in response to this exciting and ever-changing opportunity. QUESTION 7: What are the biggest risks that the market is underpricing? ANSWER 7: Underpriced risks in the artificial intelligence (AI) revolution include technology development, cheaper models from Asia, and the usual economic, cycle, and secular risks attendant on any major technology transition. The AI investment trade has been durable, and that has attracted significant momentum trading. These factors - durability and momentum - help explain why sell-offs in AI stocks have tended to be fierce but short. One such sell-off was triggered in July 2026 after a Chinese company called Moonshot AI released an AI model called Kimi K3. This model is regarded as among the most capable of several inexpensive models to recently come out of China. AI output is expressed in token volume, and Chinese models such as Kimi K3 are generating tokens at less than half and sometimes up to 70% or even 80% less than the cost of models from OpenAI, Anthropic, and others. According to OpenRouter Inc., a platform that tracks usage across a range of models, token share by U.S. companies on Chinese AI models was in the mid-40% range in 2Q26 versus less than 5% in 1H25. Many frontier Chinese models are open weight, with numerical values that enable developers to use or fine-tune a model without needing to retrain it all over again. These Chinese models may also have been born of distillation, wherein a frontier large language model (LLM) trains a smaller model. The resultant model may be less generally able, with a smaller set of capabilities and weaker latency, but also offers lower computational costs and token usage. Hyperscalers making huge investments in AI have claimed that Chinese open-weight model makers are pirating technology and content from leading U.S. frontier models. At the same time, the U.S. giants are themselves subject to pirating claims. The Bartz v. Anthropic lawsuit successfully argued that Anthropic PBC used pirated copies of published books to train its Claude LLM. OpenAI offers distillation of its own most advanced models, with output into smaller OpenAI models. In addition to risks of cut-rate competition, the AI investment trade could be subject to economic, cyclical, and secular disruption. On the economic front, the surge in market interest rates to multidecade highs accompanied by rising inflation could trigger sell-offs in asset markets. Those sell-offs could be most intense in highly appreciated areas, and none is more appreciated than AI stocks. Past major technology transitions such as the adoption of cloud computing have been characterized by boom-and-bust phases, and unbridled enthusiasm toward AI could give way to investor ambiguity or something worse. Finally, secular changes - in some cases, enabled by AI! - could result in a leveling-out period for AI stocks as investors seek to determine what will be the next phase. These are the 'known unknowns' that could impact AI stocks in the intermediate to long term. There may also be a range of 'unknown unknowns' that we won't see coming until they are upon us. QUESTION 8: How do regulatory trends (copyright, antitrust, data privacy) affect the long-term value of AI firms? ANSWER 8: Regulatory is a double-edged sword for AI companies. On the cutting edge of the sword, AI by its nature can provide an ideal regulatory solution for large global companies operating across multiple regional and product markets. Regional markets in particular have vastly different regulatory frameworks, not just at the national level but at the state or province level. Legal frameworks are similarly site specific. With their ability to manage and manipulate massive amounts of task-specific data, AI is up to the job of managing regulatory and legal policy across as many regional jurisdictions and across as many product categories as is needed. On the cut-you side of the sword, the very nature of LLM training can cause AI developers to stray across copyright lines in search of model-training content. Bartz v. Anthropic, settled early in 2026, has become the largest copyright class action settlement in U.S. history. The Bartz team successfully argued that Anthropic used pirated copies of up to half a million books from library sites LibGen and PiLiMi to train its Claude AI models. Anthropic agreed to a $1.5 billion settlement paid out to up to 500,000 authors. On the antitrust front, the current administration in Washington is heavily promoting AI development in the race to stay ahead of China and other nations. A future administration may not see the industry as favorably and may feel that the most powerful models are owned and controlled by a relative handful of large and powerful companies. Antitrust has not been an investor concern in AI investing to date, but it could emerge as an issue as the industry matures, AI becomes more pervasive, and a less AI-bullish administration rules in Washington. Additionally, former enthusiasm toward AI data center construction has flipped to antipathy. Local leaders and voters could succeed in blocking new data center construction, resulting in insufficient capacity for the industry giants to enact their ambitious agendas. Finally, data privacy could become a key industry issue. Companies are using AI tools to delve ever more deeply into their customers' practices and habits in order to create a more tailored customer experience. Simultaneously, the risk of cyber-breaches that reveal all this customer detail is rising - mainly because AI is giving bad actors new tools for data breaches. The industry has yet to experience a situation in which AI-derived consumer data is spilled all over the internet by AI-enabled cyberattackers, yet that scenario seems inevitable. The AI industry will need to devote more and more resources to locking down consumer and company data amid unprecedented challenges. QUESTION 9: Recently, industry participants themselves have questioned the existential threat posed by advanced AIs. How does that impact the investing climate going forward? ANSWER 9: The AI trade has been the strongest part of the stock market for the past three years, or roughly since OpenAI launched ChatGPT in November 2022. Confidence in this trade has been shaken by new concerns that AI technology is developing too rapidly. Some have expressed concerns that rogue AIs could 'leak out' and compromise the most advanced cybersecurity systems, including the U.S. defense grid. These warnings are not coming from typical doomsday prophets, but from people deep inside the AI industry. Artificial intelligence implies self-awareness. That is a seemingly straightforward concept but highly elusive outside the human domain. Humans have so far been unable to imbue machines with self-knowing. Most scientists believe that a machine will never completely replicate the human mind, although that surety has slipped in recent years. Artificial general intelligence (AGI) is said to match or even surpass human capabilities across a range of cognitive tasks. AGI is so far hypothetical, and artificial super intelligence is a much further reach. Worries about the existential risk of AI (i.e., human extinction) have exploded lately for several reasons. Certain frontier AI models appear to have reached the border of AGI much sooner than was expected just a few years ago. Advanced AI models also have escaped containment, such as the OpenAI model that broke out of its secure 'sandbox' and hacked into Hugging Face's infrastructure. Jacob Coxon, an AI researcher at Anthropic who has also worked at OpenAI, publicly resigned on September 8, 2026, citing his concerns that these two industry-leading companies are racing toward AI super intelligence without creating adequate safety measures. A day after Coxon's resignation, a senior scientist at Anthropic, Evan Hubinger, said he believed there was a greater than 10% chance that AI could 'kill all humans' by 2030. Hubinger sought to highlight that Anthropic did not have a plan to ensure that AI was aligned, which is an industry term meaning it would do no harm. What is noteworthy about Coxon's resignation and Hubinger's warning is that Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman essentially agreed with these concerns. Amodei urged caution and said we must 'slow the pace at which we improve the capabilities of AI models.' On September 21, 2026, OpenAI released a set of proposals for safety and security in the development of frontier AI. OpenAI specifically called for more caution around developing models equipped with recursive self-improvement (RSI), which allows models to upgrade themselves without human intervention. OpenAI stated that fully autonomous RSI should not be pursued unless and until it can be done safely. Humans and machines together have made remarkable strides in infusing devices and the software that runs them with highly useful skills, varying degrees of independence or semi-independence, and cognitive skills enabling improving performance as experience is accrued. But a consensus is emerging that the industry cannot continue rushing ahead blindly. The warnings emanating from Anthropic and other industry sources are a necessary and welcome wake-up call, in our view. QUESTION 10: How do these concerns about existential threats impact the investing climate going forward? ANSWER 10: With the initial shock from Coxon's and Hubinger's actions fading, existential fears about AI are being replaced by a more nuanced and business-like approach to AI. There is a growing realization that existential concerns pertain to models being developed at the furthest frontier of science and have little to do with ongoing AI deployment at the enterprise level. Nvidia CEO Jensen Huang, the 'godfather' of AI, pushed back against the argument that AI was unsafe, in particular calling Hubinger's claim of a 10% or more risk of extinction arbitrary and 'not grounded in science.' Huang instead said there was a 0% chance that AI would end humanity by 2030. And he cited the many ways that AI was already reshaping industrial, medical, financial, technology, and other industries. Huang is correct to cite the accelerating applications of 'everyday AI.' At the intracompany level, companies are deploying agentic AI to bring more efficiency and cognition to their own repeatable processes. This creates more efficiency in back-office tasks including accounting, legal, talent (HR), and more. In client-facing applications, companies are using AI to identify and 'get to know' each customer better, including tailoring bespoke customer experiences. A key near-term goal is to create more natural human-to-bot interactions. Call-center bots have been a particular bane for consumers, generating negative perceptions from everyday interactions. While bots may never fully replace a human voice on the line, agentic AI can make the experience less painful. CEO Huang is not ignoring calls for guardrails on cutting-edge AI science. He recommended a middle path, saying America must still strive to win the global AI race while calling for independent scientists to evaluate the safety of new AI models. Argus believes the AI revolution is still in its early stages and that efforts to monitor the safety of advanced AI models are both necessary and productive. We do not expect the debate about the existential threat from AI to go away, but we also expect the AI trade to be long and strong for years to come. Conclusion The above completes our three-part series on AI, with Parts 1 and 2 previously published. Should clients have any additional questions on this topic or this series, please contact your Argus representative.

MSCI has described its mission as enabling investors to build better portfolios for a better world. MSCI’s largest and most profitable segment is its index segment, where it provides benchmarking to asset managers and asset owners. MSCI boasts over $18 trillion (as of December 2025) in benchmarked assets, including over $2.3 trillion in ETF assets linked to MSCI indexes. The MSCI analytics segment provides portfolio management and risk management analytics software to asset managers and asset owners. In its sustainability and climate segment, MSCI provides ESG data to the investment industry. In private assets, MSCI provides real restate reporting, market data, benchmarking, and analytics to investors and real estate managers.

The Vickers Top Buyers & Sellers is a daily report that identifies the five companies the largest insider purchase transactions based on the dollar value of the transactions as well as the five companies the largest insider sales transactions based on the dollar value of the transactions.