Today’s three AI stories may seem unrelated at first glance.
One covers financial reports.
Another discusses gas turbines.
The last one is about an online gig platform that’s been around since 2005.
But when viewed together, a clear shift emerges:
AI is repricing the real foundational assets that keep the models running.
How money is counted.
Where the electricity comes from.
And which tasks still require human labor.
The AI boom isn’t just about launching stronger models anymore.
It’s transforming the entire economic structure.
First Story｜Tech Giants Report Over $160 Billion in Quarterly Revaluation Gains from AI Investments
Financial Times compiled a striking figure today.
Tech giants like Alphabet, Amazon, Microsoft, and Nvidia have collectively gained over:
$160 billion
in just one recent quarter from the rising valuation of the AI companies they hold stakes in.
First, it’s important to clarify:
This does not mean these companies received $160 billion in extra cash from selling AI products.
Much of it is:
paper gains from investment asset revaluations.
For example, if a tech company invested in Anthropic years ago, and Anthropic later raised funds at a higher valuation,
the value of the shares they hold adjusts upward accordingly.
This increase appears as investment gains in the company’s financial statements.
Why is this huge jump worth noting now?
Because tech giants are investing heavily in each other’s AI companies.
Amazon has invested in Anthropic.
Microsoft has long invested in OpenAI.
Google also holds shares in Anthropic.
Other companies similarly own stakes in large AI, infrastructure, and related businesses.
Whenever these AI companies raise funds at significantly higher valuations,
the original investors’ asset values jump accordingly.
For example, Amazon’s filings with the U.S. SEC show:
In Q2 2026 alone, the upward valuation adjustment of Anthropic’s shares from a new round of financing was about $50.5 billion.
This figure does not imply:
Amazon received $50.5 billion in cash from Anthropic.
Rather, it means:
The market's higher valuation for Anthropic led Amazon’s existing investment to be revalued upward.
Is there a problem with this?
Not inherently.
Holding rapidly appreciating companies naturally increases investment value.
The key takeaway is:
The scale of cross-investments among AI companies is now large enough to noticeably impact tech giants’ financial reports.
So when you see:
Big jumps in net income.
Huge other income increases.
Significant investment gains.
Don’t immediately interpret it as:
“AI businesses suddenly made that much cash.”
You’ll want to ask:
Is this real revenue from customers?
Or is it just paper gains from revalued holdings in other AI companies?
Both are legitimate financial outcomes,
but their business implications are very different.
Second Story｜SpaceX Is Manufacturing Its Own Gas Turbine Blades
The second news story demonstrates just how deep AI’s reach has become.
Over the weekend, Elon Musk confirmed:
SpaceX is building a new Blades and Vanes Foundry in Bastrop, Texas.
Meaning:
A foundry for casting turbine blades and guide vanes.
At first glance, this sounds odd.
Isn’t SpaceX a rocket company?
Why are they suddenly making high-temperature turbine parts used in natural gas power generation?
The answer is:
Power.
Now AI Data Centers Struggle Not Just to Get GPUs, but the Electricity to Run Them
Large AI data centers demand enormous amounts of electricity.
Even if:
Land is secured.
Data centers are built.
GPUs have arrived.
If power can’t be connected in time,
all those GPUs remain just expensive machines.
Musk stated that Tesla and SpaceX are heavily investing in solar capabilities, but natural gas power will still be needed to support and supplement supply in the coming years.
The problem is:
You can’t just order a gas turbine off the shelf either.
One of the toughest parts to produce is the blades inside the turbine, which endure extreme heat and huge mechanical stresses for long periods.
Even their current job listings explicitly state:
"Power generation poses one of the key challenges that could slow the worldwide adoption of AI."
In simpler terms:
Power generation itself might slow AI’s global expansion.
Why Make the Blades In-House?
These aren’t ordinary metal sheets.
Gas turbines operate in very high-temperature environments.
The blades must be:
Heat resistant.
High strength.
Precisely cooled.
Reliably operational over long periods.
Manufacturing involves special alloys, precision casting, internal cooling channels, and stringent quality standards.
Musk explains:
Handling the supply of blades and guide vanes internally can put more gas turbines into operation sooner.
Some reports estimate this could bring new power capacity online about 18 months earlier.
The critical point is not just:
SpaceX building another factory.
But that:
AI companies are now chasing upstream in energy equipment manufacturing to secure the electricity they need.
From Engineers to Metal Blades — The New Bottleneck for AI Infrastructure
This is where AI infrastructure becomes truly fascinating.
Initially, competition centered on:
Who has the best model.
Then:
Who has the most GPUs.
Then:
Who has the largest data centers.
Now it’s about:
Who has the power?
When will the gas turbines be delivered?
Can the grid hook up the electricity in time?
Are cooling systems adequate?
Are transformers available?
Even:
Can turbine blades be manufactured fast enough?
The real limit on AI growth might not be the newest, coolest model.
It could be a humble industrial component few paid attention to before.
Third Story｜Amazon to Shut Down Mechanical Turk After 21 Years
The third story brings us from machines back to people.
Amazon has officially announced that:
Mechanical Turk will permanently close on September 30, 2026.
If you’ve done online work early on, you may have heard of MTurk.
It launched in 2005.
The concept is simple:
Some tasks computers can’t do.
Break those into many tiny tasks and have humans complete them online.
These were called:
Human Intelligence Tasks, or HITs.
What Did Humans Do for Computers Before?
Tasks like:
Identifying what’s in an image.
Checking if two data points are duplicates.
Filling out surveys.
Transcribing audio.
Classifying content.
Data validation.
Conducting research.
Helping annotate training data for machine learning.
Businesses wouldn’t need to hire large formal teams.
They could split thousands or tens of thousands of small tasks for workers around the world to complete individually.
In a way,
Mechanical Turk was an early form of:
A human API.
Where the computer called a human when it couldn’t do something.
Why Is It Closing Now?
Here, it’s important to separate facts from assumptions.
Amazon has confirmed:
Mechanical Turk will close permanently on September 30.
Workflows using Mechanical Turk Workers in Amazon SageMaker Ground Truth and Amazon Augmented AI must also migrate to other workforce providers.
However,
Amazon has not explicitly cited "generative AI" as the reason for shutting it down.
So it’s too simplistic to say:
ChatGPT defeated Mechanical Turk, so Amazon closed it.
The real market change is more complex.
On one hand,
Many simple tasks previously done by humans — classification, summarization, data verification, content organization — can now be automated by AI.
On the other hand,
AI companies still require large amounts of human labor,
but people’s roles have shifted from:
“Doing simple tasks for machines”
to
“Making more complex judgments for AI.”
For example:
Model evaluation.
Expert data contributions.
Red team testing.
High-quality annotation.
Answer comparison.
Safety reviews.
Domain expert validation.
Ironically, MTurk Work Has Also Been Infiltrated by AI
Researchers have found since generative AI became widespread:
Some tasks originally intended for real humans on MTurk have been completed by workers using large language models.
This creates a curious loop:
Businesses think they’re paying for:
Human answers.
But workers may be outsourcing those tasks to:
AI models.
This blurs the line between buying human judgment and model output.
Amazon’s official move to shut down MTurk signals the end of an era.
Not because:
Humans don’t need to work anymore.
But because:
The lowest-cost, most repetitive microtasks are being repriced.
Why Look at These Three Stories Together?
The first shows:
AI companies’ valuations have grown enough to noticeably affect tech giants’ financial statements.
The second shows:
AI’s power demands have grown so large that companies are moving upstream in the energy supply chain to secure turbine parts.
The third shows:
The global microtask market, which used humans to augment computers, is winding down.
They seem different on the surface.
But at their core, they reveal the same truth:
As AI capabilities increase, the economy is redefining what is valuable.
Labor-intensive microtasks once highly prized may now decline in value.
Industrial capabilities like turbine blade foundry, once irrelevant to tech, now may determine whether AI data centers can operate.
AI company equity that used to be just a startup portfolio item can now increase tech giants’ financial statements by hundreds of billions per quarter.
What Should the General Public Take Away?
Don’t just chase:
What’s the next new model called?
Pay more attention to what is becoming more or less expensive behind the scenes.
Which tasks are losing value?
Which expert judgments are becoming more valuable?
Which traditional industries suddenly become bottlenecks for AI?
Which financial gains represent real operational income?
Which are merely asset revaluations?
Because a truly major technological revolution never only changes the products.
It rearranges:
Money.
Equipment.
Energy.
Companies.
And human labor.
Today’s three news items reveal this process with increasing clarity.
Today, take a step forward with AI.
Learn an AI skill every day.
Save a bit of time daily.
Improve your abilities little by little.
SasaDaily grows with you.
