Learning with UB

[Anatoli Kopadze @AnatoliKopadze]

Anthropic engineer:

"You're not supposed to prompt Claude. You're supposed to build a system that prompts itself."

In 45 minutes she breaks down how Anthropic builds agents that remember, learn from their mistakes, and get smarter with every run.

Worth more than any paid course you'll find on building agents.

Watch the session, then read the guide on building loops below.

[Learning with UB @utsavbatra]

....

[Learning with UB @utsavbatra]

Before we become a developed nation, we must remain a fair one.

For millions of Indians, exams like NEET, UGC-NET, CUET and Board Exams are not just examinations.

They are the great equaliser.

[Learning with UB @utsavbatra]

Not technology.

Not policies.

Trust.

And yet, year after year, we find ourselves discussing paper leaks, cancelled exams, evaluation controversies, data vulnerabilities and administrative lapses.

What's even more concerning is how quickly we have started accepting them.

[Learning with UB @utsavbatra]

Everyone is told they are running the same race.
They are not.

[Learning with UB @utsavbatra]

The son of a farmer, the daughter of a daily wage worker, the child of a small shopkeeper and the student from a metro city all sit in the same examination hall hoping that merit will decide the outcome.

That belief is what holds the system together.

Not buildings.

[Learning with UB @utsavbatra]

Can ordinary Indians still trust the system to be fair?

Imagine a race.

Thousands of students standing at the same starting line.

Each carrying years of hard work, sacrifice and hope.

All told that merit will decide the outcome.

[Learning with UB @utsavbatra]

A country aspiring to become a $10 trillion economy, a global innovation hub and a developed nation cannot afford institutional complacency in the very systems that determine opportunity.

Because every examination controversy raises a question far bigger than education:

[Learning with UB @utsavbatra]

It is to the idea that hard work alone is enough.

The day a student starts believing that connections, luck or loopholes matter more than effort, we have a much bigger problem than a leaked paper.

[Learning with UB @utsavbatra]

Who pays for the lost time, the mental stress and the opportunities that never come back?

In any functioning system, trust is earned through accountability.

In ours, accountability often seems to disappear into the system itself.

The real damage isn't to an examination.

[Learning with UB @utsavbatra]

A leak happens.

An investigation begins.

A committee is formed.

Statements are issued.

News cycles move on.

But where is the accountability?

Who takes responsibility when lakhs of students are affected?

[Hedgie @HedgieMarkets]

🦔Alphabet just announced an $80 billion stock sale to fund AI infrastructure. $10 billion from Berkshire Hathaway at a 5% discount to Friday's close, $30 billion in underwritten offerings, and $40 billion through an at-the-market program starting Q3. This comes on top of $30 billion in bonds from February, $11 billion from European markets, and a $25 billion bond sale last November.
Google has over $100 billion in cash on hand and still chose to dilute shareholders. The company revised its 2026 capex forecast to $190 billion.

My Take
This one rattled me a little. Google is one of the most profitable companies on the planet. If they can't fund their AI buildout from cash flow and existing reserves, what does that say about the cost of this race? They're spending $190 billion in capex this year. Alphabet, Microsoft, Meta, and Amazon combined are north of $700 billion, and Wall Street expects that to cross $1 trillion in 2027. These are numbers that stop making sense in any normal business context.

Berkshire got a 5% discount that retail investors will never see. Goldman, JPMorgan, and Morgan Stanley are on the deal. The stock slipped after hours. And the S&P 500 just hit a record last week on the back of 20 stocks, most of them in this exact space. So you have the companies driving the entire market rally now telling shareholders they need to sell equity to keep up with their own spending. If the AI returns show up, this looks aggressive and smart. If they don't, Google just diluted its shareholders to fund the most expensive capex cycle in corporate history with no guaranteed payoff, and the people who got the best terms on the way in are the ones who can most afford to lose.

Hedgie🤗

[Hedgie @HedgieMarkets]

🦔Oracle is cutting 30,000 employees by June 15, roughly 18% of its global workforce. Revenue grew 22% to $17.2 billion last quarter. Cloud revenue jumped 44%. The AI segment grew 243%. Net income was $3.7 billion. The company committed $50 billion in capex this year for AI data centers and is a key partner in Stargate alongside OpenAI and SoftBank. Severance is capped at 26 weeks, unvested stock is forfeited, and Oracle turned down workers who petitioned for better terms.

My Take
Oracle just posted some of the best numbers in its history and fired 30,000 people in the same breath. Revenue up 22%, cloud up 44%, AI up 243%, and the response was to cut nearly one in five employees so the savings could go toward data center capex. A $3.7 billion quarter and the company still chose infrastructure over the people who built the business that produced those numbers.

Oracle is also pumping capital into Stargate with OpenAI and SoftBank while those 30,000 workers hear their roles no longer fit. Remaining performance obligations jumped 325% to $553 billion, the largest contracted backlog in the company's history. Workers asked for better severance from a company that cleared $3.7 billion in 90 days and Oracle refused. You can have record revenue, record backlog, and record AI growth, and still lose your job because the capex budget needed your salary more than the company needed you.

Hedgie🤗

[Hedgie @HedgieMarkets]

🦔IBM CEO Arvind Krishna says AI is "not a bubble" then estimates the industry needs $6 to $8 trillion in total capex for data center and chip buildout. To recover that over seven years, companies would need $1 to $2 trillion in new annual revenue. Krishna says he doesn't believe that revenue exists. He also thinks only two or three companies will succeed at building leading AI models. Everyone else is spending to stay in a race most of them will lose.

My Take
"Not a bubble" but $6 to $8 trillion chasing revenue that the guy doing the estimate says probably isn't there. I don't know what definition of bubble Krishna is using, but by mine that qualifies. He's the CEO of IBM, a company that sells AI infrastructure services to enterprises. If he's flagging that the payback doesn't work, and he benefits from the spending continuing, that should carry more weight than a random skeptic saying the same thing.

Google just sold $80 billion in equity to fund its buildout. Oracle cut 30,000 workers to redirect salaries into capex. Anthropic filed its S-1 the same day GitHub Copilot's token billing collapsed and users fled. Krishna put a number on what everyone else is dancing around. The spending has outrun any realistic estimate of the revenue it can generate, and the companies writing the checks know it. They're just betting they're one of the two or three that survive long enough for it to pay off.

Hedgie🤗

[Brian Krassenstein @krassenstein]

BREAKING: It appears that a U.S. helicopter was shot down as it was searching for the missing pilot from the jet Iran had earlier shot down. This video was released by Iranian media.

What a complete and utter sh-tshow.

[abhay jain @abhayjainp]

Notes on small cap companies i found worth studying:

Aarti Pharmalabs: scaling up high margin cdmo business which recently hit 40 commercial projects and growth guidance of 30-40%. new atali greenfield plant will drive volume growth for cdmo and intermediates into fy27. xanthine capacity is doubling to 9000 tons by the end of this fy. they have a massive trigger here as china just withdrew its export rebate on caffeine which should push prices up. Key metric to track is CDMO as % of sales.

Aeroflex Industries: precision manufacturer of metallic flexible hoses & assemblies, pivoting into liquid cooling skid assemblies for AI data centers via an exclusive 5 year India agreement with a US partner. Targeting 1000+ cr consolidated revenue by FY29. EBITDA margin target of 23-25%. value added product mix already at 54% and climbing. targeting 300 to 350 crores in peak revenue from liquid cooling skid assemblies by FY29.

Quality Power: HV electrical equipment sector experiencing supply constrained demand driven by grid modernization, data centers. targeting revenue of 700-800 crores for FY26 with margins around 20 22 percent. expanding the sangli coil factory to increase capacity by 9x by june 2026, targeting peak revenue potential of 1500-2000 crores post expansion. mehru capacity is expanding by 45%.

Time technoplast: transitioning towards a higher margin product mix with value added products targeted to reach 35% of sales in two years. expanding cng cascade capacity to generate 800 cr in revenue by 4q 26, launching 14.2 kg composite lpg cylinders for omcs, developing hydrogen cylinders for drones, and supplying hdpe pipes for large infrastructure projects like the amaravati smart city. aiming to be completely debt free in the next 6 months.

disc: no reco.

[ValueEquity @EquityValueIn]

Sai Life sciences Capex tracker , mapping out phase wise ramp up

disc: no position

[The AI Colony @TheAIColony]

Breaking: GOODBYE POWERPOINT.

CLAUDE can now create a presentation in 120 seconds.

Use these prompts instead and see the magic:

[Akshay Jogani @kshayjogani]

JustDial is now on ~₹5,400 Cr of cash against a market cap of ~₹4,800 Cr, effectively trading at a negative EV.

A powerful reminder that promoter's perceived capital allocation intent + cred matters a lot when market assigns a multiple.

Disc: don't own / no advice!!