Maruthi

[Maruthi @maruthid1]

Very interesting take on AI https://t.co/HOFECqcLaa

[Maruthi @maruthid1]

💯

[Jeff Bezos @JeffBezos]

Thank you. The important part is zeroing out taxes on the bottom half. Best way to put money in someone’s pocket is to not take it out in the first place. Bottom half is only 3% of total tax revenue. But it’s very meaningful to that person. Zero it out.

[Maruthi @maruthid1]

Employees use computers → Meta tracks interaction patterns to train AI → AI learns how “smart people” work → AI automates workflows → layoffs happen.
That’s it

[Maruthi @maruthid1]

@Nithin0dha This is 100% true. I tried two attempts to open zerodha account and could not go through. I lost interest in india investment. Legal process, KYC, banking technology/apps, investment apps are very poor customer experience & complex which will discourage foreign capital

[Maruthi @maruthid1]

💶

[IvanaSPEAR @IvanaSpear]

Why optics are the new memory ❓

Optical Components matter more than ever: we’re moving from human internet → AI internet

For decades, demand for bandwidth was growing predictably.

It was driven by people:
👉 number of users × time spent online × video consumption

That math was finite.

But that model is breaking.

We are now entering a world of AI agents — and each agent behaves like a user (or many users).

They generate, process, and move data constantly.

👉 The result: bandwidth demand is going exponential

That shift was front and center at OFC (Optical Fiber Communication Conference) — the key global event for optical networking.

🔑 Key debates from OFC:

1) Bandwidth is becoming the bottleneck

The biggest shift: how AI systems are connected internally

New “scale-up” networks (inside AI clusters) need ~10x more bandwidth

Industry demand could grow 10x in just a few years

👉 This isn’t incremental growth — it’s a structural reset

2) Optics for Scale-up

AI clusters are rapidly scaling (hundreds → thousands of chips working together)

This requires optical connections in addition to copper depnding on the use case.

Nvidia is pushing this transition forward with its new chip - Feynman

⚠️ Key debate: timing

Massive opportunity… but more likely 2027+

3) “Scale across” is happening now

Not just bigger clusters — but clusters connecting across campuses and regions

Companies like Cisco and Ciena are already seeing this demand

New technologies allow networks to dynamically reconfigure at massive scale

👉 This is the real-time buildout happening today

4) Pluggable optics are still in high demand

New approaches (like co-packaged optics) are gaining some traction but key is to understand the differences in use cases

👉 Translation: evolution, not disruption

5) Supply could become the next constraint

Today: lasers are already tight

Tomorrow: the entire optical supply chain could be stretched

👉 The risk isn’t demand — it’s keeping up with it

🧠 Bottom line:

We’re early in a multi-year infrastructure cycle driven by AI

The internet was built for humans

The next phase is being built for machines talking to machines

And that requires orders of magnitude more bandwidth.

I ll do a detailed thread sharing notes from each company I met with…

Follow @IvanaSpear if you are not already.

$COHR $LITE $CIEN $CSCO $AAOI

[Maruthi @maruthid1]

@ariaradnia Good realization…it’s part of the game

[Maruthi @maruthid1]

@hamids @dkhos Agree! Rivian is getting cash for staying alive for some more time. Who’s next after burning this cash?

[Maruthi @maruthid1]

@jack @AJA_Cortes @blocks Leadership is tough, but the way you communicated shows honesty and transparency in your decision. Good separation package. Good luck to all who got impacted, life will give you different opportunity.

[Maruthi @maruthid1]

Totally worth it. Eagerly waiting for your upgrade to all flights.

[Qatar Airways @qatarairways]

We agree😉

[Maruthi @maruthid1]

@bindureddy It’s too early…future is obvious we will have Intelligent agents how this ecosystem getting build out is very important to watch. Starting from OpenAI every one building , New companies (with lot of capital ) with Agent first approach will take lead over existing Saas players

[Maruthi @maruthid1]

@gokulr @OpenAI It’s easy to draw these visions? Execution, selling to enterprises, building partner ecosystem those are complex things to solve. Microsoft, Salesforce , SAP every one thinks / have same vision it’s just that we will end up having different companies providing AI capabilities

[Maruthi @maruthid1]

@saxena_puru It seems more reaction to Fed chair announcement Kevin he is big critic of QE and crypto meltdown - due to quantum threat ?

[Maruthi @maruthid1]

Agree. $CRM, $NOW, $MSFT (Dynamics) are logical layers on top of databases for business ops. That software layer is exactly what AI agents disrupt and they can be built fast. Satya articulated this well. MSFT is already re-imagining the stack.

[IvanaSPEAR @IvanaSpear]

Software stocks are falling like a rock. $500+ billion gone over the past week.

◾ $MSFT beat revenue → stock was down 11%+

◾ $NOW crushed earnings → Stock Crashed 11%.

◾ $CRM reported growth → Down 6% …and keeps going

Is software dead?

Here is my view:

In mid-2024, the consensus was clear: Hardware was overhyped, and Software was the next AI beneficiary.

Software Companies spent billions preparing for AI.

Salesforce, ServiceNow, touted AI Agents as the next big thing, splurging into big acquisitions.

BUT…

It's taken these software giants years to develop tools that barely address customers' needs.

Generative AI can now do what traditional software was designed to do, just better and in real time.

Traditional software is built once, and then you train users or provide instructions.

AI models adapt dynamically to whatever needs to be done.

So why would anyone buy thousands of software seats when AI can build custom solutions instantly?

As a result, software multiples collapsed from 17x forward revenue in 2021 to 4.2x today - an all-time low!

The valuation logic supporting high multiples was high growth, recurring revenues, and predictable cash flows.

That predictability is now coming under attack.

However, not all software is created equal, and there are many opportunities that will come from investors throwing the baby out with the bathwater.

This is the biggest sector repricing since 2008, and it's not stopping.

Being on the right side of these rotations can make or break LT returns.

$SNOW $ADBE $MNDY $BILL $ASAN $HUBS $U

[Maruthi @maruthid1]

@Nithin0dha @zerodha It’s far away from simple process. I tried following up with your customer care for couple of weeks and stopped it.

[Maruthi @maruthid1]

Is it only cow dung? What about buffalo? Goat? Sheep? Etc etc

[Sridhar Vembu @svembu]

@truthin_tyagi There are now FDA approved fecal pills in the US. Go look them up.

Does that not suggest to you that "cow dung and cow urine are beneficial" is a scientific proportion worthy of investigation?

[Maruthi @maruthid1]

@EricLDaugh He is rocking World Economic Forum!

[Maruthi @maruthid1]

@wealthmatica bcos of simplicity and AI narrative they were able to sell to mid and smaller enterprises, it will face hurdles when they target larger enterprises. I’m long $zeta but they will hit roof pretty soon, rapid innovation in AI will be key to success

[Maruthi @maruthid1]

@wealthmatica Nice graphics and good explanation. What is stopping Adobe or Salesforce to add identity matching capability ? It’s intentional to maintain neutral platform position, regulatory and trust risk. Why would agencies tolerate vendor lockin? It would collapse their partner ecosystem

[Maruthi @maruthid1]

@elonmusk Many people of eagerly waiting for CyberSUV. What’s stopping you making SUV similar to Cybertruck?

[Maruthi @maruthid1]

@QualityInvest5 I would say sell $adobe small portion and buy $zeta. My strong hunch is zeta success will eat away some parts of Adobe business in Marketing cloud (not talking about its creative cloud)