20H·

Are we in an AI bubble?

Many new investors on Getquin are expressing the fear that we could find ourselves in a bubble like in the 2000s (dotcom), which could lead to a crash. 2025 the same only with the AI hype. I would look at this in a differentiated way, with CISO and other dotcom companies there was a rapid price increase without sales increasing due to the Internet, but rather the hope for it, so there was a lack of real orders. NVIDIA and other companies are generating real sales through AI (even if some of them are still unprofitable), but the direction is different.

$NVDA (+0,32 %)
$MSFT (+1,74 %)
$CSCO (-1,67 %)

10 Commentaires

Chris@Multibagger

18H

As long as Bild doesn't call on Germans to close their savings accounts and buy shares instead, there will be no bubble.😉😎😂

•

All-in-or-nothing@All-in-or-nothing

16H

@Multibagger Classic, the picture indicator, who doesn't know it😁👍🏻.

••

Mo@BigMo

17H

If one of the Mag7s shows a clear earnings miss or poor guidance, we should be worried.
Due to the high index concentration of Mag7s, there will then probably be a cascade of sell-offs.

•

Niels@Niels19

16H

Hans Bernecker thinks it will soon rumble because of the bond market and the escalating global debt. Does he recommend a cash ratio of 30% as he thinks there could be a 10-15% setback?

https://www.finanzen100.de/video/erst-bond-crash-dann-aktien-rekord-hans-a-bernecker-im-rahmen-von-bernecker-tv_H1850360271_45922/

•

Cekay@CMustermann

12H

@Niels19 Oh come on, don't let such a peasant take the piss out of you. Of course the guy sells 6 different "stock market letters" for only €10-30 per month.

The following ALWAYS applies: If someone has the knowledge to beat the market by the average market return, then he uses it himself to get rich secretly and quietly. The result is, for example, the Medallion Fund, a closed fund that outperforms every year but you can't even get in.

If someone needs to sell information to people, then you are the product. Especially with crash professionals, money is made with panic and doubt. Don't let yourself be taken for a ride.

••

Cekay@CMustermann

12H

I heard an interesting perspective on this in the last Marktgeflüster episode - and that is that we can't take the pure earnings of hardware manufacturers and think everything is fine with the AI boom, they are deceptive.

In the end, the AI software providers demand high prices, which are then driven even higher by OpenAI and the like, which are then driven even higher by data center providers, and in the end NVidia adds another x-fold. The margins are therefore completely exaggerated and nobody earns money - except Nvidia. This cannot be sustainable and will have to be corrected sooner or later, the only question is when.

•

Tenbagger2024@Tenbagger2024

10H

The AI sector is very volatile. Perhaps also nervous.
DeepSeek has shown how quickly AI shares react to news in the short term.
New CPUs are being developed. From energy-saving to new materials.
Anyone who misses the boat here is left behind, see Intel. Or Nokia back then.
So I don't necessarily see an AI bubble. But some companies may be the losers.

•

LongtermValueInvestor@LongtermValueInvestor

Do this with $PLTR 😂😂

•

Nom@user28461

14H

The only question is whether this turnover is sustainable. If many Nvidia customers do not see any (immediate) economic benefit from the AI investment, they will abandon AI again.

AI must also be integrated into internal processes and not just be a chatbot that tells you where to find which form that then needs to be filled out and read/processed by an employee.

Show me an AI that does the following independently via chat and without further paperwork:
- Change my home contents insurance and add policy XY, and since competitor X gives a discount for it I want you to give me the lower price
- Increase loan Y by 50% and increase the term by 25%
- I send you my car via Autopilot and you repair it, then you send it back to me at a different address
- Tell me if medical research on patients with a certain diagnosis is possible in Bavaria (enough cases available) or if I have to ask other federal states and which ones
- List me the current state of research and open research topics on antihypertensive drugs
- show me the regional vending machines that still have a certain type of cheese available, and how many of them
- Book me the cheapest and at the same time best (according to the preferences you know about me) vacation in Sardinia
- My plane is delayed, tell me the alternative without having to talk to the airline. Book my seat, just tell me the gate and time
...

There's still a huge amount of wood to be chopped, but I think it will be like this or something similar for the next few decades.

••

Nom@user28461

14H

@user28461 Here still extended by ChatGPT:

Great - great topic, and you've already provided strong examples. Here are 20 more concrete everyday scenes in which an AI assistant could act completely end-to-end (i.e. not just give advice, but make decisions, execute transactions and provide feedback) - each with a brief description of exactly what the AI does and the typical hurdles involved.

20 everyday cases in which AI can replace human assistants (end-to-end)

1. automatic tax return & refund
AI collects wage and bank data (with permission), fills out tax forms, submits electronically, negotiates with the tax office in case of queries and initiates credit notes.
Hurdles: Proof of identity, legal authorization, audit logs.

2. complete rental contract management
AI negotiates rental conditions (incl. credit check), draws up legally compliant contracts, organizes key handover and books final cleaning/inspection.
Hurdles: notarial/legal requirements, liability issues.

3. automatic complaints and returns processing
User photographs defective product → AI initiates return slip, arranges collection, monitors refund/exchange and arranges for replacement.
Hurdles: Logistics APIs, retailer guidelines.

4. end-to-end construction project coordinator
User wants extension: AI organizes permit check, obtains cost estimates, commissions tradesmen, monitors construction progress, pays after acceptance.
Hurdles: local building permits, liability, quality control.

5. personal investment portfolio manager with execution
AI optimizes asset allocation, trades automatically with brokers, realizes rebalancing, controls tax minimization.
Hurdles: regulatory approval (MiFID/AFM/etc.), trust issues.

6. automatic patient appointment & therapy manager
AI evaluates symptoms + history, books specialist/surgery appointment, orders prescriptions, coordinates follow-up care.
Hurdles: Data protection (medical data), medical responsibility.

7. final warranty repair processing for the car
AI diagnoses via telemetry, books workshop with collection, arranges spare parts, checks repair, delivers car to desired address.
Hurdles: physical repairs, insurance coordination, liability.

8. automatic labor law changes (HR)
AI manages contract changes: Salary adjustment, approval workflow, tax class change, new company car, all documented and executed.
Hurdles: Works council, compliance.

9. real-time energy optimization for single-family home
AI negotiates energy contract, controls PV battery/heating, sells surplus, organizes maintenance.
Hurdles: Smart home standards, grid restrictions.

10. independent submission and pursuit of insurance claims
After damage: AI clarifies coverage, submits claims, organizes adjuster, manages payout or repair shop.
Hurdles: Fraud prevention, verification processes.

11. automated global visa and travel paperwork
AI checks entry requirements, applies for visa/ETA, books required tests/vaccination certificates, updates travel guidance.
Hurdles: government interfaces, rapid legislative changes.

12. house sale incl. notary appointment & relocation coordination
AI organizes exposé, negotiates with buyer, coordinates notary, tax consultant, arranges removal company and handover.
Hurdles: notarial formalities, legal liability.

13. automatic compliance reporting for small businesses
AI creates VAT, accounting and reporting reports, submits, pays duties due.
Hurdles: Audit trails, client authorization.

14. end-to-end school/childcare place management
AI finds suitable daycare/school, submits applications, manages waiting lists, organizes settling-in appointments.
Hurdles: public administrative processes.

15. fully automated ordering system for small companies
AI monitors stock, negotiates supplier prices, triggers orders, organizes delivery and quality control.
Hurdles: Supplier integration, contract rights.

16. automatic legal dispute resolution (mediation)
AI collects facts, proposes settlement offers, initiates payments/obligations if agreed.
Hurdles: Legal validity, enforceability.

17. dynamic parking space reservation & autonomous parking
AI reserves spaces, coordinates autonomous parking, pick-up and billing.
Hurdles: Autonomy capability, operator integration.

18. complete GP follow-up care for chronic patients
AI monitors telemetry, adjusts medication dose (in consultation with doctor), orders follow-up tests, contacts specialists.
Hurdles: medical responsibility, regulatory approvals.

19. automated training & onboarding program for new employees
AI creates personalized learning paths, books training, documents completions, creates work access.
Hurdles: Company systems, data protection.

20 Automatic recall in the event of product safety problems
AI identifies affected customers, organizes collection/exchange, coordinates notifications to authorities.
Hurdles: legal reporting obligations, logistics.

---

Why many of these scenes are not yet running everywhere - and what is needed to make sales sustainable

In a nutshell: many scenarios are technically feasible - but sustainable sales depend on several non-technical factors:

Real economic advantage (ROI): Companies need clearly measurable savings (labor costs, error reduction, time-to-value). If an AI investment does not improve margins, it will be canceled - exactly your point to Nvidia customers: Hardware alone is not enough if there is no integration and measurable benefits.

Deep integration: Autonomous processes must be integrated into ERP, CRM, banking, insurance, government APIs. A pure chatbot is not enough.

Legal & liability: Financial, health and legal actions require approvals, audit trails and often human oversight. Many tasks remain human until the regulation is clear.

Data access & trust level: Fully autonomous actions require secure identity and authorization flows (e.g. eID). Customers need to gain trust, especially for money/health.

Business model & price communication: Providers must be transparent about who receives what benefit (end customer, company, platform). Added value must be shared (e.g. through success fees).

User experience & recovery: If the AI makes mistakes, reversals must be simple - otherwise the customer will quickly lose trust.

Network and platform effects: Ecosystems (e.g. insurers + garages) make offers stronger; this generates recurring revenue if the platform has enough partners.

---

Conclusion / Recommendation

There is huge potential - especially where there are legal, logistical and system interfaces. But sustainable sales will only be generated if the AI:

1. delivers measurable ROI,

2. is deeply (not selectively) integrated into processes,

3. fulfills legal & security requirements, and

4. has clear recovery/audit processes.

••