The most revealing line in this story isn't a stock pick. It's the premise underneath it: the best way to invest in the future of artificial intelligence is to buy the companies that already run the present.
That is a strange thing to say about a supposedly disruptive technology. Disruption is meant to scramble hierarchies. Yet when a famous market voice is asked where the AI opportunity lies, the answer is Meta, Microsoft, and a supporting cast of chipmakers and security vendors. The market has looked at AI and concluded that it will mostly deepen existing positions of power. Investors may be right, and that should worry us more than any bubble would.
The fair counterargument is that this is the sober choice. Meta and Microsoft fund their ambitions from enormous existing revenue, not from speculative start-up money, which is arguably healthier than the dot-com era. Cramer himself concedes the risks of high valuations and overenthusiasm. I accept that. But a sturdier balance sheet doesn't make the arrangement fairer. It makes it more durable. A boom run by firms that already control distribution, cloud infrastructure, and troves of user data is not a boom that spreads opportunity. It is one in which the gains are likely to be captured by those best placed to capture them.
Consider what is actually being sold. Meta pitches Muse as a tool that completes tasks rather than merely chatting, part of an effort to expand beyond advertising. Microsoft is credited for confidence in Copilot and Azure. "Completing tasks" is a polite phrase. In an enterprise setting, a task is something a person is currently paid to do. The ThoughtSpot chief executive says adoption of AI agents is still early and everyday uses aren't firmly established. That is a candid admission, and it sits awkwardly beside the valuations Cramer says investors are paying. The market is pricing in a future that the people building it say hasn't arrived. Someone will eventually be asked to make that future real, and the likeliest answer is that workers will be told to be more productive, or be replaced, so the multiple can be justified.
Notice who is missing from this conversation. Workers whose jobs are the raw material of the 'task completion' pitch have no seat in it. Users whose data trains and feeds these systems are present only as a risk factor. IDC's Meredith Whalen flags user trust and data privacy as significant challenges, and the phrase deserves more attention than it gets. Trust and privacy are not technical bugs. They are the price people are being asked to pay, often without choosing to, for a product whose upside flows to shareholders. When the cost is diffuse and the profit is concentrated, markets will not correct the imbalance on their own. That is what regulation and public standards are for.
Then there is the investment advice itself. Stock-picking television frames AI as a menu of individual decisions, as if households were choosing among tickets. But most Americans' exposure comes through retirement accounts and index funds, where these giant firms already carry outsized weight. Meanwhile, direct stock ownership remains heavily skewed toward the wealthiest households, with the top tenth holding the large majority of equities. So the AI bet is double-edged. The affluent own the upside directly. Everyone else is quietly pulled into the concentration risk, and if the euphoria Cramer warns of turns out to be real, pensions and savings absorb the fall, while the asset owners who got in early have already taken their gains.
None of this means AI is useless or that Copilot and Muse won't find real uses. It means the interesting question is not which ticker to buy. It is who gets to set the terms: who owns the models, who profits from the labor they displace or augment, who bears the privacy costs, and whether the public gets any claim on a technology built on public data and, increasingly, on public energy and infrastructure. A society that treats those as afterthoughts to a stock tip has already answered the question of whom the technology is for.
How it may affect me
For ordinary people, the practical stakes run through three channels. First, work: if AI agents are sold on completing tasks, many office and service workers could face pressure to produce more, accept less leverage, or compete with software, while the gains go to shareholders unless workers have bargaining power, notice requirements, or a share in the productivity. Second, privacy and trust: the data people feed into these tools may become the product, and without clear rules the burden of protecting themselves falls on users. Third, savings: retirement accounts heavily invested in a few giant tech firms mean that a correction in AI-driven valuations could hit pensions and 401(k)s, while the wealthiest owners, who hold most stock directly, have the most capacity to profit from the run-up and absorb a downturn. Over time, if AI mostly strengthens incumbents, it may reduce competition, raise dependence on a few platforms, and make the public's choices about the technology narrower, not wider.
