A headline about Tesla in 2013.

Shares have since risen by 6,173%.

✔️ Subscribe on @venture

🔍 Nvidia linked to pirated books used for AI training

A controversial story has surfaced around Nvidia and the use of pirated book datasets to train AI models.

In early 2024, a group of authors sued Nvidia, claiming the company trained models on Books3, a known pirated dataset containing hundreds of copyrighted books. Nvidia responded at the time that this fell under fair use.

The case is still ongoing, but a new filing has added fresh detail. Newly revealed correspondence shows contact between an Nvidia employee and Anna’s Archive, a large shadow library hosting pirated books and academic papers.

What the messages show:

🖱 An Nvidia data strategy employee asked Anna’s Archive what content was available and how to get fast access to a large book corpus

🖱 Anna’s Archive explicitly warned Nvidia that the data was illegally sourced and asked whether the employee had internal approval to use it

🖱 About a week later, Nvidia management reportedly approved the request, citing competitive pressure

🖱 Access was then granted

The exact scale and payment details are unknown, but reports estimate the dataset at roughly 500 terabytes, potentially millions of books.

The situation may go further. The plaintiffs claim Nvidia likely used other shadow libraries such as LibGen, Sci-Hub, and Z-Library, and may have distributed internal scripts allowing corporate clients to automatically download similar datasets. These claims have not yet been proven.

The lawsuit continues.

✔️ Subscribe on @venture

❓ What’s going on with the money at OpenAI

Over the past few days, headlines around OpenAI finances have intensified. The core narrative is simple. Cash burn is accelerating, and the gap between costs and revenue keeps widening.

Here is what the numbers say.

🖱 The Information reports, citing internal forecasts, that OpenAI could lose $14–17B in 2026. That is roughly three times higher than earlier estimates. By the end of 2028, cumulative net cash outflow could reach $44B, based on internal documents.

🖱 Deutsche Bank is more pessimistic. Their estimate puts total negative cash flow before profitability at $143B. They note that no startup in history has operated at losses of this scale. Around $15M per day is reportedly spent on Sora alone.

🖱 Even under optimistic assumptions, cumulative spending by the end of the decade could approach $200B. That assumes the company keeps operating without disruption.

🖱 Based on the current balance sheet, cash would run out around mid-2027. This does not factor in the $134B lawsuit filed by Elon Musk.

🖱 Against this backdrop, investor George Noble publicly criticized OpenAI, saying the company is “falling apart” and advising AI founders to sell near peak hype, while investors reduce exposure to the AI trade.

🖱 A near-term cash crunch still looks unlikely. Sam Altman is currently raising around $50B at a reported valuation of $750–830B, according to Bloomberg. Talks are largely with Middle East investors.

The real issue is not access to capital today. It is scale. If revenue does not grow by an order of magnitude while costs keep compounding, even very large checks only buy time.

✔️ Subscribe on @venture

✔️ Subscribe on @venture

📹 TikTok in the US is now officially American-controlled

The sale of TikTok’s US business was approved by both the US and Chinese governments. The asset was valued at $14B, with additional fees paid to the US government for brokering the deal.

Ownership after the transaction:
🖱 New US investors hold 50%, including Oracle and Revolution
🖱 Existing investors retain 30.1%
🖱 ByteDance keeps 19.9%

The company will be run by Adam Presser, currently deputy CEO of global TikTok.

Structurally, this resolves years of regulatory pressure. TikTok’s US operations now sit under American control, while the product and brand stay intact.

✔️ Subscribe on @venture

💰 OpenAI wants a cut of what its models help create

OpenAI CFO Sarah Friar said the company is exploring business models beyond subscriptions. The reason is cost. Compute spending keeps rising, and flat fees do not scale with outcomes.

What OpenAI is considering:
🖱 Commerce layer fees from purchases made directly inside ChatGPT
🖱 Discovery layer participation tied to value created with its models

The second part is the shift. In some cases, OpenAI would take a share of intellectual property or future revenue. One example mentioned was drug discovery, where an AI-generated molecule could trigger a claim on patent value or sales.

This would not apply to regular users. The focus is enterprise contracts in areas like pharmaceuticals or energy, where AI output has direct commercial value.

The model moves OpenAI closer to revenue sharing based on results, not usage.

✔️ Subscribe on @venture

Value of $5,000 invested 15 years ago:

🖱Nvidia: $1,650,000
🖱Tesla: $1,455,844
🖱Broadcom: $587,824
🖱Lam: $222,916
🖱Micron: $200,666
🖱KLA: $179,635
🖱ASML: $177,594
🖱AMD: $168,521
🖱Netflix: $159,942
🖱Amazon: $132,339
🖱Alphabet: $108,312
🖱Apple: $107,320
🖱Applied Mat: $106,372
Microsoft: $80,500

✔️ Subscribe on @venture

This chart looks obvious in hindsight.

But living through it wasn’t easy.

Microsoft survived missed cycles, brutal bear markets, antitrust lawsuits, bad products, leadership changes, and years where the stock went nowhere.

Most investors would’ve sold long before the payoff.

This is what real compounding actually looks like - slow, frustrating, boring… until suddenly it isn’t.

The question isn’t “could you buy Microsoft?”

It’s could you have held it?

✔️ Subscribe on @venture

“i have no faith in this company for good reason. anyways….who would like to pay for it?”

✔️ Subscribe on @venture

Finding a high-paid job in the US now looks statistically brutal

🖱 The odds sit around 0.000029%, roughly 1 in 3,400,000. That is the implied probability based on current openings versus demand.

🖱 For context, becoming a unicorn company is estimated at about 0.00006%, or 1 in 1,700,000.

By the numbers, creating thousands of jobs is more likely than landing one of the top ones.

✔️ Subscribe on @venture

"For quality of life, it is better to err on the side of being an optimist and wrong, rather than a pessimist and right"

- Elon Musk

✔️ Subscribe on @venture

🎞 5 YouTube interviews on AI from industry leaders

A short list of long-form conversations worth time. These are not news clips. Each one explains how people building AI think about risk, limits, and direction.

🖱 Geoffrey Hinton
The godfather of neural networks reflects on AI safety and why he underestimated the risks early on.

🖱 Elon Musk
A blunt take on how current AI can already replace a large share of white-collar work. Focus is on scale and near-term impact.

🖱 Aravind Srinivas
The Perplexity CEO talks about what intelligence actually means and how machines differ from humans. Less hype, more first principles.

🖱 Terence Tao
A mathematician explains why large language models are simpler than they look, and why their effectiveness is still poorly understood.

🖱 Yann LeCun
One of the founders of modern AI explains why LLMs alone will not lead to AGI and where current approaches fall short.

Together, these interviews map the gap between what AI can already do and what it still cannot.

✔️ Subscribe on @venture

⭐️ On this day in 1964: Nike founded.

It was first called Blue Ribbon Sports.

🖱 Sales (1964): $8,000 for the year

🖱 Sales (2026): $8,000 each 5 seconds

✔️ Subscribe on @venture

💸 Jensen Huang once sold Nvidia stock to buy a Mercedes

Jensen Huang shared a story at the World Economic Forum about a gift that became very expensive over time.

🖱 After Nvidia’s IPO in 1999, he sold part of his shares to buy his parents a Mercedes S Class S600 with a 6 liter V12. The car cost about $135,000. Nvidia was valued at $300 million and the stock traded at $12.

🖱 The sale involved 11,250 shares. Since then, Nvidia has completed 7 stock splits. Those shares would now equal about 5.4 million shares.

🖱 At a current price near $187, that stake would be worth over $1 billion. For scale, the Rolls Royce Droptail costs about $32 million and only 4 exist.

Huang joked that it was the most expensive car in the world.
His parents still own it.

His remaining Nvidia stake is now valued at roughly $123 billion.

✔️ Subscribe on @venture

🌴AI domain boom fills Anguilla’s budget

🖱 Anguilla is seeing a sharp revenue lift from the global rush into AI. The reason is simple. The .ai domain zone belongs to the small Caribbean island.

🖱 As AI projects launched, demand for .ai domains surged. By early 2026, the number of registered .ai domains passed 1 million. Growth accelerated in 2023, matching the rise of mainstream AI adoption.

🖱 According to Sherwood, Anguilla earned about $32 million from .ai domains in 2023. That accounted for roughly 22% of total government revenue. Estimates for last year put domain income near $70 million, close to half of all revenue.

AI hype ended up reshaping a national budget.

✔️ Subscribe on @venture

True Wealth by Nassim Taleb.

✔️ Subscribe on @venture

🚀 Space Beyond founder plans $249 space funerals with a SpaceX launch

🖱 Ryan Mitchell, founder of Space Beyond, plans a launch in late 2027 after securing a deal with SpaceX. The price point is $249 per customer, far below the usual cost that runs into the thousands.

🖱 The service sends 1 gram of ashes into orbit using CubeSats. These small satellites were originally developed at NASA and measure 10 x 10 x 10 cm, with multiple payloads packed into one unit. The satellite stays in orbit for about 5 years before burning up in the atmosphere, with its path trackable from Earth.

🖱 Mitchell is an engineer who previously worked at NASA and Blue Origin, leaving in 2025 to start the project. Falling launch costs and shared satellite missions make it possible to bundle many orders into a single flight.

Cheap access comes from sharing space rather than owning it.

✔️ Subscribe on @venture

📣OpenAI sells ChatGPT ads at NFL level prices

🖱 OpenAI has started selling ad placements inside ChatGPT at prices close to NFL broadcast video ads, according to The Information. Rates are around $60 per 1,000 impressions, which is well above typical pricing from Meta and Google.

🖱 The analytics offered to advertisers remain basic. OpenAI reports total impressions and total clicks, without showing the query context where ads appeared or what users did after clicking.

🖱 This leaves advertisers paying premium prices with limited visibility. A fuller ad stack would require deeper tracking and tooling, which does not exist yet.

✔️ Subscribe on @venture

Be Elon

🖱 Choose the hardest business on Earth: rockets
🖱 Start SpaceX
🖱 Join Tesla as a startup, take control, push out founders
🖱 Run both companies to the edge in 2008
🖱 Survive by one successful rocket launch that wins a ~$1.6B NASA contract
🖱 Figure out how to make rockets reusable
🖱 Launch a car into space. Because you can
🖱 Turn EVs from a joke into the world’s most valuable car company
🖱 Buy Twitter for $44B. Save free speech on the internet
🖱 Fire most people
🖱 Rename it after the company you lost 25 years ago
🖱 Start an AI company to race the one you helped create
🖱 Run six companies. CEO of four at the same time
🖱 Decide saving humanity is a systems problem
🖱 And treat it like it’s your full-time job

✔️ Subscribe on @venture

🛒 Waabi secures $1B to scale autonomous trucking

🖱 Waabi raised $1B to expand its self driving technology for trucks, with plans to adapt it for passenger cars later. Today, its vehicles operate with a safety driver on board.

🖱 The company says a fully autonomous truck is already ready and is expected to enter public roads this year. The system is built first for long haul freight, where routes are more predictable and utilization is higher.

🖱 $250M of the round comes as milestone based payments from Uber. The funding supports deployment of 25,000 Waabi robotaxis on the Uber platform.

Freight remains the entry point for autonomy at scale.

✔️ Subscribe on @venture

💰 Hugging Face turns down $500M from Nvidia to stay independent

🖱 Hugging Face declined a $500M investment offer from Nvidia that implied a valuation of about $7B. The talks surfaced a few months ago, with the company currently valued closer to $5B.

🖱 The reason was control. Hugging Face did not want a single large investor to gain outsized influence over its strategy and long term direction. The team said its goal is to keep AI broadly accessible rather than align it with one major player’s interests.

🖱 The company can afford that stance. In 2025, Hugging Face was profitable, driven by enterprise customers paying for advanced features. Roughly 3% of users fall into that category.

Independence over a premium check.

✔️ Subscribe on @venture