Is the Future of the World Tied to Chips? Data Centers, the Bubble and Jobs

In the Teke Tek Bilim studio, Fatih Altayli and Aysegul Ildeniz debated the direction of the world at length. At the opening, Ildeniz gave young viewers a one-sentence compass: master a craft to sense technology, yet no expertise pays off without getting along with people. In her view, memorized knowledge will not decide the future; the instinct for reaching the right source and the right person will. Empathy, persistent questioning, and checking whether the numbers match the story form the three legs of that compass. To me this was the most genuine part of the program, because the career advice came from field observation rather than cliches.

Why the rise of the chip cannot be stopped

Agent-style AI workloads give the current answer to that question. A single agent browses dozens of sites in one run, translates the results, and shuttles between memory, storage and network gear, while the central processor coordinates the whole crowd. Where four graphics processors to one central processor was once assumed, today a one-to-one ratio is discussed. This picture explains why classic producers such as Intel are valued again. The rebound of the company on the stock market and the partnership of the American administration were told in this frame. The technical lesson is plain: speed alone is not enough, orchestration is everything.

The US administration taking a 10 percent share of Intel capital was one of the most concrete topics of the program. Reuters records report the deal at 8.9 billion dollars, announced with an emphasis on domestic production. Ildeniz counted the move among the rare accurate steps of Washington. The 411-billion-dollar revenue claim voiced for Nvidia in the same segment deserves caution; the official company statement shows fiscal 2026 revenue at 215.9 billion dollars. The name Nvidia appears twice in this article: first with the record table, then with this correction. Figures invite exaggeration, filings stay on record.

The price of admission to the chip business was on the table too. The design side alone costs billions of dollars; the manufacturing side costs multiples of that. On the Indian side, the Dholera push of the Tata group signals a domestic production claim; the name Tata appears in this article in both factory and ecosystem contexts. On the Taiwanese producer front, the Arizona investments of TSMC are expected to reach 165 billion dollars in total. Ildeniz sees a chance for Türkiye on the design side rather than manufacturing, citing the quality of Turkish engineers working in Europe and America as proof. The message is clear: expensive but not impossible.

The picture on the China front is hazier. The topic passed quickly in the program, but the implication was plain: restrictions and supply-chain searches are hardening the race. Western firms looking for alternatives across Taiwan, Singapore and the Far East line reflects that unease. Ildeniz stressed without detail that the Chinese side is drawing its own path. Careful reading is required here, because anyone speaking about a closed box is partly guessing. Even so the direction is clear: nobody wants to stay dependent on a single hub.

Data centers and the energy wall

The scale of data center spending is staggering. The program named an 800-billion-dollar infrastructure investment for Google in a single year and a one-trillion-dollar threshold for the next. Independent trackers show the four large cloud firms hovering around the 700-billion-dollar band for 2026; the direction is right, the magnitude debatable. The Brookings calculation widens the frame further: as much as 10.3 trillion dollars could flow into AI infrastructure over 2025-2032. The name Brookings appears in this article with that long-range projection. The Sequoia question echoed via Yahoo still stands: where is the annual revenue that pays for all this. The name Yahoo appears in this article in the context of the 600-billion-dollar question.

The payback arithmetic was the boldest part of the program. A duty to generate 1.4 trillion dollars of revenue every year by 2030 was pronounced. The basis of that math is user scale: 2 billion people use these tools today, assumed to reach 8 billion in a decade. The Anthropic example fuels this thesis; Reuters records show the annualized revenue of the company had climbed to 3 billion dollars by the spring of 2025, with a band around 9 billion discussed for year end. The figures impress, yet the question stands: whom will a technology that has not seeped everywhere bill for this.

The energy wall is as critical as the spending. Where renewables cannot keep up with this pace, eyes turn to small modular reactors planned next to the data centers. Reuters records present the Tennessee site of the Google and Kairos partnership as an example of that plan. Similar declarations of intent exist on the Amazon side. Ildeniz, a former head of a smart energy company, recalled that the American grid carries a century-old load. A 20-trillion-dollar magnitude pronounced in the program for grid renewal must be filed as a claim, not as an invoice.

Whether data centers can spread across the globe is therefore as geographic as it is technical. The trio of energy, water and grid is not the same in every country. The topic stayed open in the program, but the hint was plain: even with the investment money ready, the physical backbone is not ready everywhere. That uncertainty is both risk and opportunity window for countries like Türkiye. Whoever solves energy captures the data center.

Bubble arithmetic and giant firms

Company valuations reaching mind-bending sizes formed the philosophical stop of the program. Facing valuations of several trillion dollars, Ildeniz warned of concentration in two hands: few actors setting both price and intent. The power displays of recent years feed that fear; the examples named in the program are debatable for that reason but instructive. My note: the monopoly critique is fair, yet the remedy lies in distributed production and an open ecosystem. Grow the alternative instead of surrendering to fear.

The bubble question was asked directly and answered without evasion. Unless the gap between spending and revenue closes, a correction is unavoidable. The math of the program is summarized above; the added point here is that the claim of AI spending reaching 4.5 trillion dollars so far, with a forecast that it must double, belongs to the same table. These figures must be read as program claims. Even so one fact cannot be denied: the revenue curves of firms like Anthropic are steepening. The line between bubble and growth will be drawn by whether revenue catches the bill.

Jobs, law and ideas

Ildeniz answered whether professions will end with a macro picture. Automation has transformed jobs throughout tech history, she argued; the difference this time is a new kind: AI fused with robots. The metaphor splitting the world into the living and the lifeless is eerie but clarifying. Routine cognitive jobs erode in the first wave, while jobs needing human contact and responsibility hold out. The compass given to the young applies here too: those who see the whole picture, reach the source, and get along with people will endure.

The legal backbone was the least discussed yet most critical topic of the program. AI models consuming shared output without permission inflate the copyright dispute; exploiting vast corpora without paying anyone is not fair. For self-driving cars and robots the liability question is even thornier: who gets the bill at the moment of a crash. Ildeniz stressed that case-law search will accelerate with data processing power, but the conscience element cannot be handed to machines. Altayli recalled that law means conscience. That sentence read like the moral summary of the program.

Whether AI can produce ideas carried the studio into philosophy. The detail that thrilled Ildeniz was a philosopher and a scientist named in the program meeting at the same point: two different schools, one junction. The lesson is that the strategic reading and synthesis capacity of models should not be underestimated. Of course the consciousness debate is separate from the utility debate. But the practical result does not change: those who ask good questions get good ideas.

Bookshelf, automotive and closing

Ildeniz answered the next-big-thing question with a laugh and a confession: someone often wrong on it. Fifteen years went into smart glasses and similar wearable hopes, yet the expected leap never came. Here shows the gap between Silicon Valley fandom and reality. The lesson is modest: what matters is not predicting the future but being ready when it arrives. So the program should be watched as a readiness list rather than a prediction list.

Ildeniz said she wrote her 300-page book to change behavior; the work named in the program as Last Exit to the Future was presented as a self-building guide for the AI age. One thesis stands out: skilled users of the tool pass into a strengthened version of themselves. The brain rot concept in the book touches a live wound: uniform use works a single muscle of the brain while other regions idle. BBC records confirm the phrase was chosen as the 2024 word of the year by the Oxford dictionaries. The name BBC appears in this article with that confirmation. The message is plain: use the tool diversely, do not narrow the mind.

The automotive segment showed software swallowing the car. Where hundreds of small control units once sat, a few central computers now run everything. The Mercedes anecdote told in the program was striking: a car bought a few years ago carried more computing power than the newspaper backbone where the guest worked at the time. The layered rules written for self-driving show the seriousness of the task; a traffic officer hand signal can overrule even a red light. The bold trials on the Tesla side test these layers in the field. Consolidation is done, the safety layer is still under construction.

The closing carried equal opportunity and a call. By the table shared in the program, women use computing tools markedly less than men and lag in access; the gap between old and young adds to it. Ildeniz voiced the fear of a polarized world split between those inside the dataset and those outside it. Altayli set the meteor metaphor aside and proposed the train metaphor: one must board the loudly arriving train, watching from the trackside is not enough. The message for Türkiye is the same: focus, prepare, board the train. That call deserves to be the headline of the program.