At a time when everyone is talking about “AI,” many see it as an opportunity to jump into. However, AI is not only technologically complex but also evolving at a startling pace in terms of investment, innovation, and practical applications.
For investors, the stock market may be the best tool to gauge direction, as stock prices reflect in real time which companies are winning or losing ground to competitors. Yet, a key question remains: where should investors begin to understand the AI industry?
Morgan Stanley Investment Management (MSIM) sought to answer this by summarizing “10 Essential Truths About AI” from their report Artificial Intelligence: Ten Investment Truths, outlining what every investor should know to grasp how the AI revolution is transforming the economy and investment landscape.
MSIM views the AI revolution as different from past technologies because it is not propelled by a single factor but by the concurrent development of four key elements:
Since 2017, AI infrastructure investment has accumulated over $2.3 trillion. In 2025 alone, AI model token usage increased more than tenfold, reflecting demand accelerating faster than many anticipated.
If this trend continues, AI systems in 2028 will be approximately 250 times more powerful than today—a growth rate rarely seen in any technology's history.
Investor perspective
Investors should focus not only on companies developing AI models but on the entire ecosystem. As algorithms, hardware, talent, and funding grow together, companies across the value chain may benefit simultaneously.
MSIM highlights that while many still see chips as the main bottleneck in AI, industry constraints have rapidly changed and no longer stem solely from chip production.
The AI industry resembles a vast production chain with interconnected components. If any part cannot scale fast enough, it becomes the new bottleneck. Recent examples include:
In other words, solving one bottleneck often leads to another emerging elsewhere in the supply chain.
Investor perspective
Investors should shift from seeking the “best” companies to those that are “indispensable.” Because AI bottlenecks shift over the cycle, investment opportunities depend on which supply chain segment is the scarcest resource at a given time.
A major industry shift is in software revenue models, moving from user count or licenses to measuring value by the number of tokens used in processing.
When users send queries to ChatGPT, Claude, Gemini, or other AI, it is not a single program activation but massive computer resource usage behind generating responses.
More queries or complex tasks mean more tokens consumed and greater compute resource use. Customers are effectively purchasing compute power represented by tokens, not just software.
Investor perspective
The shift from a software license-driven economy to a token-driven one marks a fundamental change in technology. Companies that can produce tokens at the lowest cost, maintain efficient infrastructure, and convert token usage into steady revenue will hold competitive advantages.
Thus, future software industry growth will stem not from license sales but token consumption, which will become the key economic unit in the AI era and a vital long-term metric for investors.
The biggest upcoming AI change may not just be smarter models but a role shift from AI waiting for human commands to systems that plan, decide, and act independently—so-called Agentic AI.
This transition will reshape work, as AI evolves from a productivity tool to digital labor capable of managing entire processes autonomously.
MSIM also emphasizes Multi-Agent systems, where multiple specialized AIs coordinate and communicate with minimal human intervention.
Investor perspective
Future competition will hinge not only on the best answer-providing AI but on who builds reliable Agents that can replace human tasks, integrate with organizational systems, and collaborate with other Agents, creating sustainable competitive advantages and revenue.
MSIM sees the shift to Agentic AI not as a mere upgrade but a structural revolution in work, potentially one of the AI economy's most powerful drivers.
A common misconception is that AI writing code will harm software companies by lowering development costs and easing new entrants.
MSIM counters this, asserting AI changes the nature of competitive advantage from coding skills to mastery of three key factors, the 3Ds:
MSIM also notes that initially, much capital flows into infrastructure companies, but as infrastructure matures, competition moves to applications. Companies embedding AI in customers’ workflows will generate sustainable recurring revenue.
Investor perspective
In the AI era, investors should not focus solely on companies with the best AI models but on those with hard-to-replicate assets: high-quality data, domain expertise, and strong customer bases, enabling them to convert AI into recurring revenue and maintain long-term edges.
MSIM views recent AI advances as only the beginning, mainly operating in digital realms like coding, data analysis, or image creation.
Going forward, AI will increasingly operate in the physical world, controlling machinery, robots, vehicles, and industrial systems that impact the real economy.
“AI will no longer just analyze the economy but become an economic driver,” MSIM states in the report.
Robotics will be among the sectors most transformed by AI, shifting from repetitive programmed tasks to complex, flexible operations beyond traditional automation's reach.
This will revolutionize entire factories, lowering costs, boosting output, and enabling faster market responsiveness.
Investor perspective
MSIM urges investors to see AI as a General-Purpose Technology penetrating all economic sectors, akin to electricity or the internet.
Initially, winners build AI infrastructure, but later, those integrating AI with physical assets—robots, factories, vehicles, industrial systems—to drive productivity and reduce costs will prevail.
MSIM notes a common investor error is viewing AI as solely a tech company theme or limiting it to chip makers and AI model developers. In reality, AI is a full-stack capital cycle encompassing everything from foundational infrastructure to everyday consumer services.
Building new AI systems involves more than buying GPUs; it requires investment across power plants, transmission lines, data centers, networks, chips, memory, software, and applications.
Investor perspective
The key to point 7 is shifting perspective from AI as a product to AI as an investment cycle. MSIM compares AI to building railroads or the internet, requiring massive layered investments before economic benefits reach end users.
Thus, AI-era winners include not only model and chip developers but companies at every AI stack level—from power producers and infrastructure builders to cloud providers, app creators, and traditional firms using AI to improve operations.
AI has evolved from a commercial technology to a strategic asset critical to national security, comparable to nuclear energy or the space industry in the past.
While competition initially was among companies like OpenAI, Google, or Anthropic, MSIM now sees it as a contest between national AI ecosystems, especially the US and China.
The core competition is not just who has the smartest model but who can build a complete, efficient AI ecosystem—from chips to applications—first.
Moreover, the world may not use a single AI technology set; regions may develop models tailored to their data standards, cloud systems, infrastructure, and regulations. AI competition is thus also about setting global standards and technological influence.
Investor perspective
Investors should view AI through geopolitical as well as technological and financial lenses, as future access to compute, chips, energy, and infrastructure may be shaped by government policies.
Companies within strong AI ecosystems, possessing complete supply chains and adaptability to international policy shifts, will have better long-term competitive advantages.
A major challenge in the AI era is not the technology's capability but its rapid development outpacing laws, policies, and regulatory systems.
Despite increasing AI use, most legal frameworks remain nascent. Many countries lack clear answers on critical issues like liability for AI-caused harm, ownership of AI-generated works, transparency obligations for AI training data, or AI's use in security.
These issues lack international regulatory standards, and divergent national approaches often leave private companies or AI developers making key decisions rather than government agencies.
Investor perspective
Investors should incorporate regulatory risk into investment analysis, not focus solely on AI model potential or revenue growth.
Long-term winners will be companies that build safe, transparent, compliant AI earning trust from governments and customers, outperforming those focused only on technology development.
MSIM notes that with every new technology wave, investors first focus on infrastructure builders, but over time, highest returns come from companies creating new products or services from that infrastructure—often unknown at the technology’s start.
AI is currently in its infrastructure phase, with capital flowing into GPUs, data centers, cloud, networks, and energy. When infrastructure matures, AI costs fall, efficiency rises, and access broadens, enabling widespread innovation and new businesses.
The report warns that today's highest-value AI companies may not yet exist. Like the internet era before Google or Facebook, and the smartphone era before major successes appeared post-platform maturity.
MSIM also believes the top-performing companies over the next 10 to 20 years may be in labs or not yet founded.
Investor perspective
Investors must understand the AI revolution is not finished with chip or model creation but is in an early, long investment cycle spreading across the economy, similar to the internet and smartphone revolutions, where infrastructure precedes innovation and winners.
Source:Morgan Stanley
Follow the Facebook page: Thairath Money at this link - https://www.facebook.com/ThairathMoney