NVIDIA's Record Profits: Is This an AI Revolution or the Peak of a Bubble?

If you find this helpful, please click the ♡ first. You can save it to read later.

Conclusion:

NVIDIA's record profits and 10 trillion yen in abundant cash reserves, which support the generative AI boom, are causing tectonic shifts in stock prices from semiconductors to cloud and software..
The stock market is wavering between high-tech overheating and macro risks, funds are circulating into competitive semiconductors and the resource sector, and major cloud companies are in an adjustment phase due to increased investment burdens.
Investors should assess the risk of overvaluation in AI-related stocks and balance defense and offense through sector diversification and cash preservation.

Outline:

Chapter 1: NVIDIA's Record Profits
• Sales and profits hit record highs
• Stock price at all-time high due to generative AI demand
• Market valuation transformed by breaking the $5 trillion market cap

Chapter 2: Data Centers Driven by AI
• Data center sales up 75% year-on-year
• H100/H200 GPUs drive growth
• Supply gap with competitors AMD, Intel, and TSMC widens

Chapter 3: Utilizing 10 Trillion Yen in Cash
• Free cash flow surges
• Strengthening share buybacks and dividends
• Investing funds in R&D and new factory construction

Chapter 4: Pre-empting Demand through Customer Investment
• Large-scale investments in customer companies like OpenAI
• Securing demand through a capital circulation model
• Countermeasures for governance and reputational risks

Chapter 5: Stock Price Impact by Industry
• Semiconductor stocks rise in sympathy
• Major cloud companies (Microsoft, Amazon, etc.) face increased investment burdens
• Software stocks re-evaluated due to generative AI
• Funds circulate into resource and industrial stocks
• Action plan for investors

⸻

Chapter 1: NVIDIA's Record Profits

NVIDIA's latest financial results confirmed with numbers that investment in generative AI is not a one-time event but is continuing as an 'upgrade to industrial infrastructure'.. In the fourth quarter, sales grew to $68.1 billion, a significant increase compared to the same period last year. In terms of profit, net income reached the $43 billion range, demonstrating that its earning power has moved up a level. Looking at the breakdown, the center of growth is data centers, with quarterly sales of $62.3 billion, accounting for the majority of the total. The demand for inference processing, which is being adopted by companies in addition to generative AI training, has become substantial, and the texture of the financial results shows that GPUs have changed from 'parts for research use' to 'components for corporate core investment'.

The background to the stock price sticking to all-time high levels is not just expectation. NVIDIA maintained a high gross margin and indicated around $78 billion as its sales forecast for the next quarter. What is important here is that, contrary to the conventional concern that performance would decelerate rapidly if customer-side investment slowed, the company itself sees high certainty in demand. Buyers have expanded to include not only hyperscalers like Microsoft, Amazon, Google, and Meta, but also AI development companies like OpenAI and large companies promoting in-house AI adoption. As a result, the competitive advantage, which combines GPU supply capacity, software infrastructure, and the speed of product updates, is structured in a way that is unlikely to collapse in the short-term trend.

Symbolizing the change in market evaluation is the market capitalization. Against the backdrop of rising stock prices, NVIDIA reached the $5 trillion market cap level, changing the very way the global stock market views things. What investors have begun to emphasize is not companies that temporarily rise in the business cycle, but whether they are 'companies where their share is easily fixed' in a phase where the center of corporate IT investment is shifting to AI. Furthermore, NVIDIA is showing abundant cash and room for capital policy. Reports state that cash on hand is in the 10 trillion yen range, and moves to expand demand through investments in customers are also being discussed. This is an unusual offensive method for a semiconductor company, where the seller subsidizes the buyer's funding constraints to accelerate adoption.

There are two points that should be grasped from a businessman's perspective. First, generative AI is becoming a foundational investment that simultaneously changes customer touchpoints, business processes, and R&D, rather than just a tool for cost reduction. Second, as long as the bottleneck of that investment is in computing resources and NVIDIA holds the center of that, stock price valuation may move with logic closer to 'infrastructure stocks' rather than 'cyclical stocks'.Reading record profits not as a goal, but as a meter reflecting the speed at which corporate investment is shifting to AI is a realistic way to grasp it.

⸻

Chapter 2: Data Centers Driven by AI

NVIDIA's growth engine is no longer gaming GPUs, but data centers. Data center sales for the most recent quarter reached $62.3 billion, a growth rate of 75% year-on-year. With data centers accounting for the majority of the company's $68.1 billion in total sales, it shows that the spread of generative AI has moved from a few research institutes to the core of corporate IT.

The main players driving this demand are the H100 and H200. When large-scale inference processing begins to run in actual operations, not just training for model development, the design of the network and the entire server, not just the number of GPUs, becomes the key to success. Huge cloud companies like Microsoft, Amazon, Google, and Meta are building up investments with the feeling of building 'AI factories' rather than just buying GPUs. Therefore, once a procurement decision is made, it does not end in a few quarters, and expansion and updates are likely to continue. The fact that the company's forecast for the next quarter was presented at a high level of around $78 billion in sales is a reflection of the fact that adoption is progressing in a line, not as a point.

On the other hand, there is also a strong aspect of 'wanting to sell but being unable to'. Cutting-edge GPUs depend on TSMC's advanced processes and advanced packaging capabilities, and supply constraints for HBM are also piling up. While SK hynix and Samsung Electronics are rushing to increase HBM production, the structure where demand expands first is unlikely to change. Until supply catches up, NVIDIA is likely to maintain pricing power, which in turn leads to the maintenance of profit margins.

Competitor AMD is trying to increase its presence with the MI300X series and others, but in the cloud field, the accumulation of software assets and operational know-how is effective. Intel is also following, but what customers want is not just a single chip, but certainty that includes adoption and operation.The difference including supply capacity and ecosystem is likely to fix the power map in the data center market for the time being. The implication that businessmen should get from this is the fact that the main battlefield for AI investment is in computing resources, not apps, and that hegemony in data centers is shifting to the center of corporate value.

⸻

Chapter 3: Utilizing 10 Trillion Yen in Cash

NVIDIA's strength lies not just in the size of its profits, but in the speed at which cash is accumulating. In its most recent full fiscal year, NVIDIA generated $102.7 billion in operating cash flow and $96.6 billion in free cash flow, meaning the profits earned are not just accounting figures but have materialized as capital for investment and shareholder returns. Furthermore, at the end of the period, cash, cash equivalents, and marketable securities reached $62.6 billion, with the Nikkei reporting that cash on hand has swelled to the 10 trillion yen scale. With such deep liquidity, decision-making is unlikely to slow down even if external shocks like economic recessions or regulations occur.

How this cash is used will be the next focus for stock valuation. First, regarding shareholder returns, the company allocated $41.1 billion to share buybacks and dividends for the full year, and as of the end of the period, a balance of $58.5 billion remained in the share buyback authorization. The design is to keep dividends at a symbolic level while primarily returning capital flexibly through buybacks. Quarterly dividends remain at $0.01 per share, keeping the return stance consistent while limiting it to a "minimum fixed cost that does not interfere with growth investment." This approach differs from the high-dividend philosophy of mature companies, focusing instead on allocations that bet on the next growth phase while maintaining capital efficiency.

On the other hand, the core of investment is research and development and securing supply capacity. Demand for generative AI has entered a phase where victory is determined not just by performance, but by supply capability. While NVIDIA indicates the introduction of next-generation platforms, manufacturing depends on TSMC's advanced processes and packaging capabilities, and memory depends on HBM supply from companies like Samsung Electronics and SK hynix. In other words, as long as strong demand continues, upfront investments and long-term contracts aimed at resolving bottlenecks become competitiveness itself. Having abundant cash acts not only as pricing power but also as the ability to secure resources across the entire supply chain.

Furthermore, the "investments in customers" reported by the Nikkei suggest the possibility of taking capital allocation a step further. In situations where AI adopters face capital constraints, if the supplier provides funding to pull demand forward, GPU utilization accelerates, and software and operations become easier to lock in. This is a move that works not only for short-term sales but also for medium-term customer lock-in. However, because investments increase issues regarding return uncertainty and governance, investors will strictly scrutinize "how much, where, and under what conditions." Ultimately, NVIDIA's cash is not being used for defense, but is beginning to be used as a weapon to capture demand.

—

Chapter 4: Pre-empting Demand Through Customer Investment

NVIDIA is taking a step beyond being an "AI supplier" by investing in customers and peripheral companies. The goal is not just simple financial returns. If customers obtain funds to expand data centers and that money goes toward purchasing NVIDIA GPUs, demand can be created through capital design rather than economic waves. It is a way of fighting that increases the probability of demand occurring, rather than just selling products.

A symbolic example is the movement surrounding OpenAI, where reports indicate that NVIDIA is considering an investment in a large round, and while the scale is being adjusted, the possibility of investment proceeding is suggested. Furthermore, there is speculation that capital-rich players like Amazon and SoftBank are involved in the same round, and the more capital that is gathered, the more likely demand for computing resources will be pulled forward. What is important here is that the "investment fatigue" that often occurs when cloud investment approaches its limit can be offset by the customer's own fundraising. By the supplier supporting customer growth as a shareholder, GPU utilization rises earlier, and operations become easier to lock in. This is not just sales promotion, but an extension of platform dominance.

This structure is strengthened by NVIDIA's investment function. By engaging broadly with AI startups through frameworks like NVentures, they can create a state where "demand for GPUs increases regardless of who wins," rather than trying to pick a winner. Customer cash flow improves, adoption speed increases, and the ecosystem thickens. As a result, the market is becoming more likely to evaluate NVIDIA not as a semiconductor company, but as a "core company of AI infrastructure."

However, this approach also increases risks. First is governance. As investments in major customers progress, the market will strictly scrutinize whether transaction terms are fair, whether information is managed appropriately, and whether there are conflicts of interest. Second is reputational risk, where capital involvement in a specific AI company could cast a shadow over relationships with other customers and competitors. Third is regulation and oversight, as massive investments are likely to invite antitrust and disclosure issues. That is precisely why the focus will be on not just the scale of investment, but the handling of voting rights, conditions for staged investment, and the transparency of the use of funds.

Ultimately, customer investment is not a panacea. However, in a phase where supply constraints remain, the very idea of locking in demand through a capital circulation model pushes NVIDIA's competitiveness to another dimension. The point that business people should look at is that the subject of AI investment is shifting from "tech company enthusiasm" to "capital design." Investment creates demand, and demand calls for further investment. Companies that have entered this cycle will remain at the center of the market.

—

Chapter 5: Stock Price Impact by Industry

The essence of the AI market is not NVIDIA's strong earnings themselves, but which industries capital is gathering in and where it is exiting from. The market is not monolithic, and even within tech, there is a split between winners and losers.

First, semiconductors are likely to rise in tandem. The reason is simple: the "site" of AI investment is the data center, and it is not just GPUs that are needed. In addition to NVIDIA, capital will ripple out to TSMC, which handles manufacturing; ASML, which provides equipment; and peripheral companies that support materials and processes. In Japan, stocks that trigger associations, like Tokyo Electron, are also likely to move. What is important here is that as long as AI investment continues, semiconductors will be a battle of equipment and supply capacity, and the certainty of orders and production increases is more likely to push up stock prices than short-term performance. Competitors like AMD and Intel are also being scouted, but their evaluation depends more on supply outlooks and the reality of implementation than on "whether they can catch up."

Next, cloud giants are difficult to evaluate. While Microsoft, Amazon, Alphabet, and Meta are accelerating AI investment, spending tends to come first. What the market dislikes is not losses, but the time when investment increases but profits are hard to see. Especially in a phase where infrastructure expansion continues, stock prices are more likely to be shaken by the weight of capital expenditure than to rise on expectations. Even for companies like Oracle that have expectations of capturing AI demand, if fundraising and investment burdens are on the radar, evaluations are unlikely to grow.

Software is at the center of re-evaluation, but it is prone to swinging both up and down. As generative AI spreads, there is a risk that some functions of existing SaaS will be replaced. When signs of slowing growth appear, as with Workday or Snowflake, they are easily sold off, and Adobe, Salesforce, ServiceNow, and Palantir are also experiencing both the expectation of "growing with AI" and the concern of "being eroded by AI." The turning point here is whether they can go beyond just adding AI features and dig into the customer's actual business. Simply adding generative functions is unlikely to be highly valued.

And resources and industry are likely to become a receptacle for capital. In phases where tech anxiety intensifies, capital tends to circulate into energy, materials, heavy industry, and infrastructure, and their relative superiority is also evident in indices. Overseas, resources like Exxon Mobil and Chevron, and construction machinery like Caterpillar are easily chosen. In Japan, INPEX, heavy electrical equipment, and infrastructure-related stocks are also likely to be relatively strong. As AI investment spreads to the renewal of social infrastructure, associations extend to power, cooling, equipment, and resources.

The practice for investors is simple. First, view semiconductors not as points, but as a supply chain. Second, for cloud, track the speed of investment recovery rather than growth rates. Third, for software, choose "defensible revenue models" based on the premise of replacement risk. Fourth, include resources and industry as a buffer in your portfolio to reduce volatility during overheating phases.

—

Please also check out our other NVIDIA-related articles.

If you found this helpful, a ♡ encourages me. It increases the priority of sequels.
💬 What do you think?
👉 Let me know your opinion in the comments!