Is it time to stop betting on the next breakthrough and start betting on mass scale?
Who will ultimately win the AI race? Which chip producer? Which model creator? With several AI mega-IPOs expected this year, including OpenAI and Anthropic, these questions are more pressing for investors than ever.
Investors can win most of the battle by simply understanding the two roads to an AI-fueled economy: one is through technological supremacy, the other through mass deployment. Whichever road the technology takes will determine the winners on all levels.
Every new technology eventually reaches a tipping point where investors must ask a fundamental question: Is it time to stop betting on the next breakthrough and start betting on mass scale?
At this point, customer adoption curves are the best indicator of whether the technology and the market are ready for mass rollout. The situation with AI is trickier. While there is endless demand even for its current capabilities, AI has a fundamental problem. AI models "hallucinate," meaning they routinely fabricate information, without any sign that the answers could be made up. This introduces unreliability into every single AI operation.
As a result, AI today is a supportive tool, but we're far from the promised computer autonomy and meaningful agentic AI. Who would place an AI agent in charge of answering customer inquiries? Coding a firewall? Legal advice? Holiday shopping with your credit card? Hallucinations preclude AI autonomy and automation in any high-stakes environment. Humans must always be in the loop.
This is why most American companies today still prioritize resources - money, talent, high-end chips, electricity - in training frontier models over scaling the usage of their existing ones. This focus on technological supremacy is most evident in the companies expected to go public this year: Anthropic, OpenAI, and xAI (as part of SpaceX (SPCX)).
Companies need artificial general intelligence if they are ever to see returns that justify some of the exorbitant P/E ratios in today's valuations.
These three are by far the most ambitious AI players - and the highest-stakes bets for investors. The group can best be described as AGI-first. AGI is short for artificial general intelligence, the holy grail in the AI world. AGI is the point at which AI models reach human-level performance across most intellectual tasks. Today's AI models can excel at single tasks, such as translation or image generation, but they cannot reason like a human. AGI can. It can distinguish between things it knows to be true and those it doesn't. It can also comprehend varying degrees of certainty.
To achieve true autonomy, there is no way around AGI. Companies need AGI if they are ever to see returns that justify some of the exorbitant P/E ratios in today's valuations. The most critical question for AI investors is whether AGI can be delivered at all, and by when. The public discourse is bursting with timelines, but frankly, those are all just shots in the dark, as AGI will require another breakthrough moment like the launch of ChatGPT. The timeline cannot be projected simply based on successive improvements.
Luckily for investors, the U.S. market offers diverse approaches despite its overall focus on cutting-edge algorithms. Investors who conclude that the AI race will be won by rapid, massive deployment rather than the perfect AI model don't have to buy shares of Alibaba $(BABA)$( HK:9988) or Baidu $(BIDU)$( HK:9888). Some U.S. tech giants are clearly building strategies around large AI rollouts, as imperfect as the current models might be.
For example, Amazon.com (AMZN) monetizes different models through its cloud arm AWS and uses its own AI capabilities for its shopping assistant Rufus. Meta Platforms' (META) Llama AI fine-tunes ad targeting. Apple $(AAPL)$, rather than competing with its own cutting-edge general-purpose AI, seeks to monetize its gatekeeper function once usage explodes.
Alphabet will be a winner in a world where traditional AI is scaled. Microsoft is the textbook example of how to leverage parallel bets.
There's a third group of AI players, led by the two most powerful companies in the AI race: Alphabet $(GOOG)$( GOOGL) and Microsoft $(MSFT)$. They are both running powerful hybrid strategies. Alphabet's DeepMind ecosystem is just as advanced and ambitious as the algorithms of Anthropic, OpenAI and xAI, but with far less risk should AGI not be soon achieved.
Alphabet will be a winner in a world where traditional AI is scaled. It is a cloud powerhouse and dominates search, mobile operating systems and numerous other applications in which Gemini can be integrated easily. Microsoft is the textbook example of how to leverage parallel bets. In the AI age, Copilot and Azure provide a base for corporate adoption, plus Microsoft holds a significant stake in OpenAI and continues to work on its own frontier models.
Diversification in the tech sector is not about different geographies, sizes and business models. Whether by focusing on such hybrid players or by building a portfolio balancing pioneers and scalers, savvy tech investors always account for multiple likely scenarios.
Comments