The largest and most advanced factory in the world is being built; it's called Terafab

1️⃣ The 4 tech giants are spending the equivalent of Singapore's GDP

Now that we have the reports of the 4 super-spending giants in our hands, we need to talk about CapEx. The capital expenditure of Alphabet, Amazon, Meta, and Microsoft reached $301 billion dollars in the first half of 2026 alone, and the updated guidance of the four companies totals ~$732.5 billion dollars for all of 2026. 

To put it into perspective, an amount exceeding the gross domestic product of Singapore is being invested in data centers, servers, and chips. 

The market's question remains: where is the return on this investment?… Well, this earnings season provided answers, quite convincing ones in the case of the cloud business: Google Cloud grew by 82% year-over-year, Amazon Web Services by 37%, and Microsoft Cloud by 26%.

Although Google Cloud has the most impressive figure, almost doubling its business, Amazon was the clear winner since it has the largest share of the pie, and its nominal growth is absurd. We'll talk about Amazon further down.

Meta's case requires another metric. Since it does not participate as a provider in the cloud business (for now), Meta seeks to monetize infrastructure by accelerating the performance of its AI ad optimization suite, thereby raising the average price per ad (CPM) and boosting its total advertising revenue.

Looking at it from that angle, Meta knocked it out of the park. It grew 27.4% due to an increase not only in ad price but also in volume.

💡 What you should know: The unprecedented CapEx is already monetizing with surprise growth in the most profitable businesses of these companies. The craziest part is that obligations (closed sales that have not yet been delivered) continue to grow. So, judging by the direction of the wind, capital spending could be justified.

2️⃣ SpaceX moved up its projection of $1T* in revenue to 2030

*Trillion on the short scale.

Following its historic IPO, SpaceX released its first earnings report. The company beat expectations in revenue and net loss (yes, loss). Although it remains an unprofitable company, it lost less than expected.

SpaceX reports its revenue in 3 segments: Space, Connectivity, and AI. Connectivity (where Starlink is) is the company's financial lifeblood, contributing ~55% of revenue, and is also where SpaceX's short-term catalyst lies by offering Direct-to-Cell services alongside global partners like T-Mobile to connect phones directly to satellites.

During the earnings call, Elon Musk mentioned that Starship (part of the “Space” business) controls ~90% of the mass put into orbit worldwide (the total tonnage sent into space), reducing the cost per kilo launched by more than 80%.

Also, one of the most revealing pieces of data was the expectation of reaching an annual recurring revenue (ARR) of more than $100 billion dollars by December of this year. And the projection of reaching one trillion dollars in revenue is now expected by 2030, a year earlier than previously projected.

💡 What you should know: SpaceX is not being quiet, but many investors are nervous about “over-promising.” All of SpaceX's projections are internal, and at least half of the market knows that Musk is not usually the most punctual. However, the other half of the market doesn't want to miss the opportunity of the rocket taking off without them.

3️⃣ The largest and most advanced factory in the world is being built, and it's called Terafab

Back-to-back SpaceX news? Yes… but this really deserves its own space.

Tesla and SpaceX began construction of Terafab, a factory that wants to take vertical integration of semiconductors to the next level. It has a projected footprint of 10 million square meters and a phased investment of ~$119 billion dollars.

Unlike traditional manufacturers, Terafab wants energy and industrial integration. The idea is to operate with a massive solar generation farm and Tesla Megapack storage to shield the power supply.

They also want a dedicated architecture exclusively manufacturing three families of proprietary chips: the FSD self-driving chip for Tesla, the Optimus computer processor, and semiconductors for Starlink satellites.

And they want to cut production time by integrating advanced lithography and packaging into the same factory, eliminating international shipping.

That said, Terafab does not completely eliminate external dependency; they still require ASML's extreme ultraviolet lithography machines, as well as the global supply of silicon wafers. But it is undoubtedly the most radical attempt to integrate semiconductor manufacturing directly with end-consumer AI hardware, all under one roof, and cutting weeks of production.

💡 What you should know: Terafab will give Tesla and SpaceX direct control over the design and packaging of their key chips, but will plunge both into an intense capital investment cycle. Most supporters value the reduction of geopolitical risks around the supply chain, although the deployment of free cash flow over the next few years could be a negative factor for several investors.

4️⃣ There is nothing stopping this company | AMZN Q2 2026

As we mentioned above, the capital spending race had a clear winner this season: Amazon. 

Amazon's Q2 report just proved to the market that it doesn't care as much about the amount of capital spending, but rather how the bill is justified.

The company reported incredible acceleration in revenue in its higher-margin segments, sending a very clear signal that investment in AI infrastructure is paying off today. 

However, not all are green flags. Negative free cash flow was reported, CapEx guidance rose again, and there are some significant distortions.

In this video, we review Amazon's Q2 earnings report and determine if it can continue to be the winner in the AI infrastructure race 👇

5️⃣ There were fewer jobs available in July, but there was also less unemployment… What?

The latest labor market report in the United States presented somewhat contradictory data: the economy lost 23,000 jobs in July, but the unemployment rate fell from 4.2% to 4.1%. What?… How can there be less unemployment if jobs were lost?

This is explained by how the official unemployment rate is calculated:

Imagine a community of 100 people of working age. If 90 people have jobs and 10 do not but are actively looking, the labor force is 100 people and unemployment is 10% (10 out of 100). Suppose that the following month 2 jobs are lost (leaving 88 people employed), the unemployed would rise to 12.

If during that same period 8 people decide to stop looking for work entirely (they retire or simply stop looking), the number of "active" unemployed drops to 4 (12 - 8). The new official labor force is reduced to 92 people, the same 88 employed + 4 unemployed looking for work. The new unemployment rate drops to 4.35% (4 out of 92).

That's what happened in July: labor participation fell. There were indeed fewer jobs, but there were also fewer people looking for the jobs that still remain.

Markets interpreted this moderation in employment as confirmation that the economy is cooling gradually, reducing pressure on the Federal Reserve to maintain restrictive policies and opening up room for interest rate adjustments.

💡 What you should know: The drop in the unemployment rate was due to workers leaving the active population, confirming a possible cooling of the labor market that gives the Fed room to maneuver.