The main investment theme for the next phase of AI is one sentence: whoever can produce tokens on a large scale cheaply, like water, electricity, and coal, will make big money.

In the past three years, everyone competed to train the "strongest model"; in the next three to five years, what determines profit is "who can embed intelligent, cheap, stable, and large-scale AI into every workflow."
Inference demand will be 100, 1,000, or even 10,000 times the current training demand. Tokens are about to become a fundamental resource that companies need to buy daily.
It's not that models are unimportant, but the marginal returns from training have already decreased significantly. Capital markets now focus on real revenue and profit.
How can ordinary people participate?
1. From a stock investment perspective (most direct)
· Focus on the three types of companies above:
Entry points selling tokens: $CRM, $ADBE, $MSFT, $DUOL , and other software stocks with real users and data.
Cloud giants creating tokens: $GOOG, $AMZN, $MSFT (the most stable).
Inference cost-reduction chips: $AVGO, $NVDA (plus related chips, data centers, power stocks).
· If you don't want to pick individual stocks, buy related ETFs: such as AI-themed, cloud computing, semiconductor, and data center ETFs, which are easier for long-term dollar-cost averaging.
· Remember: don’t go all-in, diversify, and pay attention to CapEx (capital expenditure) and AI revenue growth in earnings reports.
2. From a personal ability perspective (most important)
The biggest opportunity for ordinary people isn't stock trading but transforming into "people who can use AI."
· Use tools like Cursor, Claude, Copilot, Manus daily, integrating AI into your workflows (coding, spreadsheets, copywriting, customer service, data analysis).
· Learn prompt engineering, agent building, and cost control. Those who can do more with fewer tokens are more valuable in the workplace.
· Many jobs will be replaced by agents in the future, but "people who can command agents" will be even more in demand.
3. Daily focus points
· Watch if token prices continue to drop significantly (the cheaper, the more used).
· Keep an eye on AI capital expenditure and inference revenue in cloud giants’ earnings reports.
· Notice which companies truly embed AI into their internal operations rather than just playing with concepts.
· Pay attention to news about power, data centers, liquid cooling, and other "behind-the-scenes" infrastructure.
AI is moving from an era of "scarce capability" to one of "cost discipline." The best way for ordinary people to participate is to invest in related stocks while actively integrating AI into their work and life.
Whoever turns AI into their productivity first will stand in the next wave of dividends. The story isn't over; it has just moved from the lab into factories.