Written by: Techub News Compilation
Recently, the four tech giants Google, Amazon, Meta, and Microsoft released record-setting quarterly earnings reports one after another within 90 minutes, sparking intense discussion in the market about whether the AI boom represents real growth or a massive bubble. Ejaaz and Josh, hosts of the Limitless Podcast, conducted an in-depth analysis of this situation. This article summarizes key data from their dialogue, the strategic positioning of the tech giants, and core judgments about the future of the AI industry.
1. Record-Breaking Earnings: Bubble or Golden Era?
“Are we in an AI bubble?” This is the central question in the current market. Ejaaz points out at the outset that the “AI bubble” and “overspending” previously criticized by skeptics have been “completely disproven” by the latest earnings report data. Taking Google as an example, its earnings per share exceeded expectations by 94%, which is no mistake. Its cloud business revenue increased by approximately 50% year-on-year, with a backlog of computing orders valued at nearly $500 billion. Even Meta, previously thought to be lagging in the AI race, achieved massive revenue growth in its advertising business due to AI.
However, Josh presented a more nuanced viewpoint: “Are we in a bubble? The answer is likely a clear ‘yes’. But the key is how big this bubble is? It might be much smaller than many imagine. And how much runway is left? I believe it is much longer than people expect.” The focus of the divergence lies in the definition of a “bubble” and the interpretation of current data.
The core contradiction in the earnings reports is: on one hand, the tech giants are increasingly investing unprecedented capital expenditures (CapEx). Google's guidance for capital expenditures for the next fiscal year reaches $180 billion to $190 billion. Josh described this as a “foolish, crazy number.” In comparison, Google's revenue in a single quarter ($110 billion) has already exceeded Coca-Cola’s annual revenue. Even more astonishing, Google CEO Sundar Pichai stated during the earnings call that capital expenditures in 2027 will be significantly higher than this year’s (which is already at a record level). This indicates that investment intensity will only increase in the coming years.
On the other hand, these investments are generating substantial, unexpected profits. Ejaaz emphasized that it is essential to look at the expansion of profit margins. The average profit margin for Google and Amazon’s cloud businesses has expanded by approximately 50%. This means that providing AI chips and cloud services has enabled them to achieve higher marginal profits. Josh added that pricing elasticity is also very high; even with price increases, customers still pay.
Ejaaz summarized a key observation: the overall revenue growth of these companies is significant but not “amazing.” What is truly “amazing” is the substantial expansion of profit margins. This addresses the core question about AI demand: are companies willing to pay for AI infrastructure? The earnings reports show that demand not only exists but is so strong that even a hyper-scale cloud provider like Google has been “recently constrained by computing power,” and its cloud revenue could have been higher had it met demand. Ejaaz noted that the backlog of $462 billion in orders is secured orders, not forecasts, which strongly proves that market demand is real and urgent.
2. Funding Flow: From Tech Giants to Core Infrastructure
Where will the massive capital expenditures flow? Josh pointed out that this is a process of capital “trickling down.” When companies like Amazon and Google use cash to build data centers, money flows to both upstream and downstream of the industry chain: they need shells, energy, chips.
A vivid example is the energy company Bloom Energy, which saw its stock price rise by 1400% within a year and increased by 24% just yesterday due to strong earnings. Notable investor Leon Cooperman reported in his latest 13F filing that his holdings in Bloom Energy grew from $875 million to $2.7 billion. This illustrates how capital expenditures benefit infrastructure suppliers.
Another example is the storage company SanDisk. Josh mentioned that if one had invested in SanDisk two years ago, the return would have been as high as 30 times. The reason is that all four tech giants’ earnings reports mentioned that memory is a severe supply bottleneck, and costs are rising, while companies are absorbing these costs. Meta or Amazon increased their capital expenditures by $25 billion just due to rising commodity costs.
Ejaaz and Josh both believe that this inverted phenomenon of funds flowing “down” from giant tech companies into core infrastructure is a change in the market that has not been adequately priced in. Traditionally, these tech giants absorbed a lot of free cash flows; now, they are spending it, and this money is creating new growth points in other areas.
One important rebuttal to the “bubble theory” is the financial condition of companies. Ejaaz stressed: “By definition, an AI bubble means these companies have high balance sheet leverage. However, the companies we are discussing today—Google, Amazon, Meta, and Microsoft—are not leveraged.” He pointed out that although Amazon has committed to using 94% of its free cash flow for AI infrastructure investments, which is a very aggressive strategy, they are not in a position of debt financing at present. As long as these investments can continue to generate returns above costs, this spending is sustainable.
3. More than Infrastructure: How AI Empowers Traditional Businesses
In addition to cloud infrastructure, AI is also directly transforming and enhancing tech companies' traditional core businesses, with Meta as a typical case.
For a long time, Meta has competed with Google in the advertising field, with Google holding a dominant position in search advertising due to its “internet portal” status. However, in the past quarter, the situation changed. Meta began integrating AI, especially new technologies like Manus (even though this deal was recently vetoed by China), achieving a higher AI search conversion rate. Its earnings per share exceeded expectations by 53%, and revenue grew by 33% year-on-year. Ejaaz pointed out that search engine optimization (SEO) is a massive business worth billions, and Meta is pioneering its AI version.
This case is similar to Google: previously, there were concerns that AI might erode Google’s search revenues, but its search business actually grew by 20%. This indicates that AI makes money not only through infrastructure (cloud) but also by penetrating every product category and making them more profitable.
Josh pointed out an interesting market phenomenon: despite strong earnings data from Meta, its stock price fell by 9% on the same day. He believes this may be due to the volatility of advertising businesses compared to computing power. Cloud businesses have long-term contracts and repeat customers, while advertising businesses could change in the next quarter due to competitors (like Google) launching better AI search optimization systems. However, Ejaaz believes that the market's fear of Meta may be exaggerated; people still view it as a social media company rather than as an advertising business giant.
Josh reminded the audience to reevaluate these companies: we interact with their consumer layer daily (like the Amazon shopping app, Instagram), but their underlying enterprise-level businesses (like AWS, advertising platforms) that support much larger scales are the real growth engines.
4. Competitive Landscape: Microsoft’s Challenges and Future Variables
Microsoft’s earnings performance is also strong, but its stock price performed as the second-largest drop among the four giants. Its cloud business Azure grew by 40%, with lucrative profits. The issue lies in comparison and expectations.
Ejaaz analyzed that AWS and Google Cloud’s performance “crushed” Azure. Although Microsoft's percentage growth is decent, investors may prefer to choose Amazon and Google when competing with the other two giants. Additionally, Microsoft raised its capital expenditure guidance for the calendar year 2026 to $190 billion (25% higher than previously estimated), which exceeds Greece's GDP and the total annual capital expenditures of Boeing, Lockheed Martin, and General Motors. Such a massive expenditure scale has left some investors confused and worried.
Ejaaz also pointed out a potential weakness for Microsoft regarding AI products: the lack of a clear AI leader. He mentioned that Mustafa Suleyman, the head of Microsoft's AI team, seems not to have delivered the products expected by the market. Although Microsoft is an early investor and important partner of OpenAI, with a first-mover advantage, the adoption rate of its Copilot product among enterprise customers is not ideal. Customers are instead directly requesting to use Claude or ChatGPT. Microsoft’s response was to launch “Microsoft Features” powered by Claude and ChatGPT. However, with Microsoft's exclusive cloud service agreement with OpenAI breaking this week, Microsoft will face challenges in the future when the relevant intellectual property expires in a few years. Ejaaz believes the market may be pricing in this “bearish” scenario, opting for the safer choices of Google or Amazon instead.
Josh summed up that as long as the incremental value discovered by users in AI tools and services exceeds the cost, this cycle can continue. Problems will arise once this chain stops when models become “dull” or use cases run out. “But we are not there yet,” Josh stated, “so I feel optimistic.”
5. Industry Dynamics: From AI Agents to Music Generation
In addition to earnings reports, the conversation also touched on several important AI industry dynamics that further underscore technological evolution and innovation in business models.
1. Anthropic's “Project Deal” Experiment: The Anthropic team conducted a week-long experiment to create an internal classifieds market operated solely by AI agents (Claude). Humans could not post items, write descriptions, or haggle; everything was done by the AI agent. The experiment ultimately resulted in a transaction volume of $4,000, and participant feedback was positive. Josh pointed out that such “agents” workflows will consume vast amounts of tokens, thereby promoting the growth of underlying computing power demand from another angle.
2. Cursor Releases AI Agent Harness: Cursor released its API for the AI Agent Harness this week. Ejaaz admitted that he had previously underestimated Cursor, thinking it was just an “AI wrapper” with no moat. However, it turns out that this “harness,” which can finely tune for specific use cases, set up development environments, and integrate APIs, has values equivalent to the model itself. Sam Altman also expressed a similar view this week. By opening and charging for its harness as an API, Cursor has built its own commercial moat. Notably, Elon Musk’s xAI has obtained options to acquire Cursor.
3. Eleven Labs' Music Generation Platform and Creator Economy: Leading company in audio AI, Eleven Labs launched a music generation platform and announced that it has paid creators $11 million through its “sound library.” Ejaaz believes this platform is significant as it directly addresses artists' concerns about AI stealing IP without compensation. The platform offers a transparent and traceable way for sound IP owners to profit from the AI-generated music created from their sounds, even if the music is not composed by them. This opens up new possibilities for the creator economy.
4. Claude Adds Creative Software Connectors: Claude released a new series of connectors linking creative software like Blender, Adobe Creative Cloud, Ableton, Canva, and SketchUp. After testing, Josh found that while powerful, the current processing speed is slow, and there is a risk of errors in complex tasks (such as 3D renderings containing thousands of parts), but it is still a noteworthy technological advancement.
5. Amazon's “AI Podcast” Feature (A Counter example?): Amazon tested a new feature on its e-commerce platform: generating a podcast discussing a product hosted by two AI presenters on the product page, with real-time user question support. Both Ejaaz and Josh found this perplexing and even somewhat “off-putting,” considering it potentially “the worst AI product ever.” It more so demonstrates that the cost of generating AI content has decreased to such an extent that Amazon can try it “for free,” but not everything needs to be AI-generated or turned into a podcast.
6. Conclusion: The Bubble Has Not Yet Arrived, But Key Indicators Need Attention
Returning to the initial question: Are we in an AI bubble? Ejaaz's conclusion is: not yet. He believes that a bubble will only arise when these companies start accumulating large debts and their price-to-earnings ratios (P/E Ratio) soar. In the future, when companies like SpaceX and OpenAI go public and capital is drawn away from the current giants, the situation may change, but we are not at that stage yet.
Josh provided support from a valuation perspective: currently, the pricing of these core businesses is not high. Meta's P/E ratio is 16 times, Microsoft’s is 25 times. In contrast, at the peak of the internet bubble, Microsoft's P/E ratio reached 73 times, Cisco exceeded 200 times, and Yahoo even surpassed 800 times. Currently, these companies are not over-leveraged and are not spending beyond their means; they are merely reinvesting the revenue earned from customers to expand their businesses.
Ultimately, this discussion reveals a clear picture: profits related to AI are growing significantly, primarily from infrastructure and margin expansions, especially in cloud businesses. As long as every dollar invested in computing power remains valuable to customers, and products like Claude and ChatGPT are deeply embedded in high-value enterprises that generate billions of dollars in revenue each year while discovering profound value, this AI-driven capital feast and growth story will still have room to continue. The direction is currently positive, but market participants need to closely monitor changes in key indicators such as capital expenditure, debt levels, P/E ratios, and value creation for end users.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
