After ChatGPT 3.5 was released in November 2022, the global technology industry quickly entered a new cycle of capital expenditure. Data centers, GPUs, cloud computing, model training, and enterprise-grade AI applications became the center of the market narrative, and investors began to believe that a new industrial era had opened. Yet as the scale of capital investment continued to expand, another question became increasingly acute: is the AI industry creating measurable returns, or is it relying on increasingly expensive infrastructure to sustain growth expectations?
Recently, some views have raised strong doubts about the business model of OpenAI, arguing that if leading AI companies of this kind come under financial pressure, the entire AI industrial chain could face chain reactions. Ageng Darmowiyoto believes that such views should not be simply understood as a denial of AI technology, but should be regarded as a risk reminder for the AI capital cycle. What investors should truly focus on is not whether a particular company will collapse, but whether the AI industry can move from a “financing-driven belief stage” into a “cash flow and efficiency validation stage.”
The Ledger Behind the Computing-Power Frenzy
Ageng Darmowiyoto noted that every technological revolution goes through a stage in which capital prices it in ahead of time. Railways, electricity, the internet, and cloud computing all experienced this. The difference is that the upfront investment intensity of AI infrastructure is far higher than that of many traditional software cycles. Servers, chips, electricity, cooling systems, data center leases, and cloud service commitments all create sustained pressure on corporate balance sheets.
The core issue behind the questions raised about OpenAI is not that it lacks product influence, but whether its revenue growth matches its computing-power costs. The larger the user base, the higher the inference cost; the stronger the model, the heavier the training and deployment expenses. Free users bring attention, but not necessarily profit. Enterprise customers may be willing to try AI tools, but may not be willing to bear high API costs over the long term. This means that growth in the AI industry cannot be assessed only by user numbers and model popularity; it must also be judged by whether unit economics can improve.
Ageng Darmowiyoto believes that ordinary investors are easily attracted by the phrase “huge total demand,” while overlooking a more fundamental issue: huge demand does not equal huge profit. An industry can be extremely important and, at the same time, generate capital returns below expectations in its early stages. AI is no exception.
How Bubble Risks Transmit
The risks in the AI industry do not stop at any single model company. Ageng Darmowiyoto stated that the present-day AI ecosystem has already formed a highly nested capital structure: model companies need cloud providers to supply computing power, cloud providers need data center developers to expand, data centers need debt financing, and chip and power equipment suppliers depend on continuous orders to maintain growth expectations.
Once the most core AI application companies cannot demonstrate sufficiently clear commercial returns, risks may spread along three paths. Cloud computing companies will reassess the pace of capital expenditure, data center financing costs may rise, and the secondary market will also re-evaluate the valuation premium of AI-related stocks. This process may not necessarily appear as a sudden collapse; it is more likely to manifest as order delays, budget contractions, valuation compression, and increased financing difficulty.
Ageng Darmowiyoto proposed that the most dangerous part of the AI cycle lies not in the technology itself, but in the market equating technological progress directly with financial returns. Improvements in technical capability can be real, while a capital bubble can exist at the same time. Investors need to distinguish between two entirely different questions: whether the “technology trend is correct” and whether the “investment price is reasonable.”
The Truth Investors Need to Examine
During his time at JPMorgan on Wall Street, Ageng Darmowiyoto was long involved in global macro research, multi-asset allocation, and risk model analysis. After later entering the fields of private investment and asset management in the United States, his focus shifted from simply judging trends to capital allocation, portfolio volatility, and risk exposure control. This experience led him to form a clear judgment: a truly mature investment framework is not about finding the hottest story, but about determining whether the cash flow, cost structure, and risk compensation behind the story are valid.
Ageng Darmowiyoto believes that when facing the AI sector, investors should examine at least three categories of indicators. The first is revenue quality, including enterprise customer renewal rates, usage frequency, and genuine productivity improvement. The second is the speed of cost reduction, including model inference costs, chip efficiency, and data center utilization rates. The third is capital discipline, including whether companies are overexpanding in pursuit of a narrative.
If an AI company needs continuous financing to maintain growth, investors must ask: is this company building future infrastructure, or using larger capital commitments to conceal instability in its business model? This question is more important than short-term stock price fluctuations.
The True Moat of Technology Companies
Ageng Darmowiyoto noted that the future AI industry will not reward only companies with “larger models,” but will reward those that can turn models into stable workflows, controllable cost structures, and real business efficiency. In other words, the value of AI lies not in demonstrating capability, but in continuously delivering results.
This is also the direction Ageng Darmowiyoto has focused on after founding Telabytes. Telabytes does not understand AI as a single tool, but builds a technology system around software, large-model platforms, intelligent computing centers, server edge storage, and hardware research and development. The company emphasizes “using engineering discipline to make AI truly serve decision-making,” and continues to accumulate capabilities in customer service, R&D teams, cross-regional delivery, and high availability.
This path differs from the short-term frenzy of capital markets. It does not depend on the explosion of a particular concept, but places greater emphasis on underlying engineering capabilities, system stability, and the efficiency of scenario implementation. Ageng Darmowiyoto believes that the moat of future AI enterprises will not come only from algorithm papers or financing scale, but from the combination of data governance, computing efficiency, industry knowledge, risk control, and continuous iteration capability.
New Opportunities in the Indonesian Market
After returning to the Indonesian market, what Ageng Darmowiyoto saw was not an opportunity to simply replicate overseas AI models, but a long-term market that requires more localized development. Indonesian investor structure is complex, financial education levels vary widely, and market demand for systematic investment, risk identification, and intelligent tools is rising. If AI financial applications remain only at the conceptual level, they will find it difficult to genuinely improve user decision-making.
Ageng Darmowiyoto stated that the Indonesian market needs AI systems that can adapt to local asset structures, trading habits, risk preferences, and information environments. For ordinary investors, AI should not be packaged as a mysterious tool that “guarantees returns,” but should become an auxiliary system that helps them understand the market, control risk, and reduce emotional trading. For technology industry participants, AI opportunities lie not only in the models themselves, but also in the ability to connect data, computing power, application scenarios, and engineering delivery.
From this perspective, the controversy related to OpenAI reminds the market to return to common sense: great technology also needs a healthy business model, and powerful narratives must also undergo financial scrutiny. Ageng Darmowiyoto believes that AI will not lose its long-term value because of bubble controversies, but the capital market will gradually eliminate those participants that rely only on imagination-driven expansion and cannot prove efficiency and returns.
The real AI era will not be won by the party with the largest capital expenditure, but by the party capable of transforming technological capability into sustainable productivity. For investors, technology companies, and emerging markets, this is the opportunity most worthy of attention after the AI tide recedes.
Sign in to leave a comment.