Individual Economists

Meet The New AI Billionaires Of 2025

Zero Hedge -

Meet The New AI Billionaires Of 2025

In just a few short years, dozens of entrepreneurs and technologists have crossed into billionaire status thanks to the artificial intelligence gold rush.

To learn more about this new class of wealthy founders, Visual Capitalist's Marcus Lu highlights some of the most notable AI billionaires minted in 2025.

Data & Discussion

This visualization is based on reporting from CNBC, which tracks the emerging fortunes tied to AI companies around the world. Net worth data was collected from Forbes, as of August 2025.

*The average Anysphere founder’s stake value is $1.3B.

CoreWeave Co-founder: Michael Intrator

Michael Intrator, co-founder and CEO of CoreWeave, stands out with an estimated net worth of $6 billion. His company provides cloud infrastructure optimized for AI workloads, a sector that has seen explosive demand as AI models become increasingly compute-intensive.

The company launched its IPO in March 2025 at $40 per share, and has seen extreme volatility since then. Shares peaked at $183.58 on June 20, 2025, but are currently trading in the low $90 range.

CoreWeave has a strategic partnership with NVIDIA, which is not only its primary supplier of GPUs, but also a major investor.

Scale AI Co-founders: Alexandr Wang & Lucy Guo

Alexandr Wang and Lucy Guo, co-founders of Scale AI, are currently two of the world’s youngest billionaires.

The pair met when working for social question-and-answer website Quora, and started Scale AI in 2016. The San Francisco-based firm provides data annotation and other services that are used to build, test, and refine AI systems.

Since then, Guo has founded Backend Ventures (a tech-focused venture capital firm) in 2019, and later Passes (a content creator monetization platform) in 2022.

More recently in June 2025, Meta invested $14.3 billion into Scale AI, a deal which also placed Wang at the head of Meta’s Superintelligence Lab.

OpenAI Alumni: Dario Amodei, Ilya Sutskever & Mira Murati

Three of OpenAI’s most prominent alumni, Dario AmodeiIlya Sutskever, and Mira Murati all helm companies valued in the billions.

Amodei, formerly vice president of research at OpenAI, co-founded Anthropic in 2021. The company is responsible for Claude, which as of January 2025 had 105 million monthly users.

Sutskever, formerly chief scientist at OpenAI, co-founded Safe Superintelligence in 2024. The company is valued at over $30 billion, with investment from firms like Andreesen Horowitz and Alphabet.

Last but not least is Mira Murati, former chief technology officer at OpenAI. She founded Thinking Machines Lab in February 2025, raising $2 billion from a consortium of investors including Andreessen Horowitz, Nvidia, and AMD. The company is set to announce its first product later this year.

If you enjoyed today’s post, check out The Most Popular AI Chatbots in 2025 on Voronoi, the new app from Visual Capitalist.

Tyler Durden Wed, 09/10/2025 - 22:00

It Makes More Sense To Produce Hydrogen With Nuclear, Not Renewables

Zero Hedge -

It Makes More Sense To Produce Hydrogen With Nuclear, Not Renewables

Authored by Ross Pomeroy via RealClearEnergy,

Hydrogen is often thought to be linchpin of a future 100% renewable economy. To make up for wind and solar's deal-breaking intermittency and to rid industry of energy-dense fossil fuels, the surplus cheap electricity that renewables produce during times of abundance would need to be channeled into electrolyzers that split water molecules into hydrogen and oxygen. The hydrogen could then be collected, stored, transported, and eventually combusted for on-demand energy.

But is this scenario really the most cost-effective and environmentally-friendly option? Would it make more sense, for instance, to produce hydrogen from another carbon-free source – nuclear power?

A trio of scientists in the Department of Civil and Industrial Engineering at the University of Pisa in Italy explored that question. Utilizing data from the International Atomic Energy Agency Hydrogen Economic Evaluation Programme, the group performed a feasibility assessment to compare various methods of producing hydrogen from futuristic Gen IV nuclear reactors. Their findings are published in the journal energies.

Two methods of producing hydrogen from nuclear power rose to the top. First, engineers could construct an attached electrolyzer system just like with renewable energy. Since a nuclear reactor is almost constantly running as "baseload" power, plant operators could simply divert power to the electrolyzers when grid demand wanes. The researchers estimate the cost of hydrogen with this setup would be 2.71 USD/kg with paltry carbon emissions of 0.3 kgCO2e/kgH2. 

Second, the authors envisioned a system where futuristic Gen IV reactors operating at high temperatures (between 550 and 1000 °C)  can produce hydrogen through the hot steam they emit. This high-temperature steam electrolysis is similar to how hydrogen is produced from steam reforming via natural gas. They predict costs here to be 3.57 USD/kg with slightly higher emissions of 0.8 kgCO2e/kgH2. Costs are higher because it a more novel system, even though it is "the most efficient coupling since it better exploits the electrical and thermal energy resources produced by the reactor," the researchers write.

Costs and carbon emissions for both methods compare favorably with current costs of hydrogen produced from fossil fuels and renewables. In Europe in 2023, hydrogen made via methane reforming cost 3.76 EUR/kg (4.39 USD) and produced emissions of at least 11.6 kgCO2e/kgH2. Hydrogen made from a direct connection to renewables cost 6.61 EUR/kg (7.71 USD) with emissions similar to production from nuclear.

The researchers' assessment is highly preliminary, of course. There's only one commercial nuclear reactor in operation today that matches the reactor they modeled. It's in China. Their cost estimates could also be overly rosy, and it's likely that the cost to produce hydrogen with renewables will come down over time as solar panels grow more efficient. 

The world currently produces 52.6 million tons of hydrogen per year, used mostly to make ammonia for fertilizer. The process of making this hydrogen accounts for two percent of the world's total energy consumption and contributes roughly the same proportion of global carbon dioxide emissions. Even if hydrogen doesn't find wider use in industry, transportation, and grid storage, we still need a lot of it to feed the world, preferably produced in a far cleaner manner than it is currently. Nuclear energy could provide it in abundance.

Source: Buzzetti, R.; Lo Frano, R.; Cancemi, S.A. Sustainable Hydrogen Production from Nuclear Energy. Energies 2025, 18, 4632. https://doi.org/10.3390/en18174632

Tyler Durden Wed, 09/10/2025 - 19:30

Eight Years Of Consumer AI Deployment In One Giant Timeline

Zero Hedge -

Eight Years Of Consumer AI Deployment In One Giant Timeline

What unfolds in this sweeping spiral graphic, made possible by Visual Capitalist and Made Visual Daily, is a story of how the LLM ecosystem has exploded in size, complexity, and ambition. Each arc represents a model release, each colored path traces a different entity, from OpenAI to Meta, Anthropic, Baidu, and beyond.

When eight years of LLM evolution are collapsed into one view, what’s revealed is not merely the pace of releases, it’s how competition, curiosity, and capital have fused into a self-sustaining cycle. Each launch seems to fuel the next with even greater velocity, pushing breakthroughs into deployment at unprecedented speed and scale. It’s a vivid testament to how far the field has come — and how high the stakes now are.

Key Players in LLM Innovation

While OpenAI maintains a commanding lead in user base and public attention, the wave-like graphic underscores a crucial point: this is not a one-player game. The ecosystem is vast and highly competitive.

OpenAI: The most recognizable name in the space. With ChatGPT and its API integrations, OpenAI has become the face of generative AI. GPT-4 and GPT-5 remain benchmarks for capability and deployment scale.

Meta: Meta’s LLaMA series has steadily evolved into one of the most important open-weight model families. LLaMA 2 helped popularize open access; LLaMA 3 is now used in major consumer products like Instagram and WhatsApp, and LLaMA 4 is on the horizon.

Google: With Gemini, Google has rebranded its Bard chatbot into a full-spectrum AI platform. It’s integrated across Workspace tools, Search, and Android, signaling a deep push into multimodal and productivity-focused use cases.

DeepSeek: Based in China, DeepSeek has quickly emerged as a major contender in open-weight LLMs. Its DeepSeek-V2 and DeepSeek-Coder models are among the top-performing open-access models, especially in code generation and multilingual tasks. It reflects China’s growing ambition to compete in foundational AI research.

Anthropic: Backed by Amazon and Google, Anthropic’s Claude models are designed with “constitutional AI” in mind—prioritizing safety and steerability. Claude 3 is one of the top performers across reasoning and coding benchmarks.

Beyond these leaders, the graphic is crowded with rising names. Mistral, Cohere, Baidu, Stability AI, xAI, and Alibaba are all carving unique paths—whether through specialized models, regional focus, or open-weight releases.

Contextualizing the Wave: From Transformers to Today

The roots of today’s LLM explosion trace back further than 2017’s breakthroughs:

  • Pre‑Transformer era: N‑gram and statistical models dominated (“web as corpus,” IBM’s early work).

  • 2017: The advent of the transformer (“Attention Is All You Need”) completely redefined architecture.

  • 2018–2020: Landmark models emerged — BERT, GPT‑1, GPT‑2 — steadily increasing capabilities.

  • 2020 onwards: GPT‑3 exploded public attention and utility. GPT‑3.5, GPT‑4 followed, each more capable.

  • Latest: LLaMA 3 and 4, with massive scale, multimodality, and instruction tuning, showcase how quickly the frontier keeps moving.

From this timeline, one thing is clear – innovation no longer flows from a single source, but from dozens of labs racing to define what’s next. The result? An LLM landscape that’s more competitive, creative, and global than ever before.

Explore what’s coming next for LLM-makers on Voronoi: What is coming next for LLM makers?

Tyler Durden Wed, 09/10/2025 - 19:05

Why The Democrats' Latest Framework For Crypto Market Structure Could Hurt Financial Privacy

Zero Hedge -

Why The Democrats' Latest Framework For Crypto Market Structure Could Hurt Financial Privacy

Authored by Frank Corva via BitcoinMagazine.com,

This morning, 12 Senate Democrats published a six-page framework for digital asset market structure legislation in which they outlined their plan to combat illicit finance while protecting users’ financial privacy.

Senator Ruben Gallego and 11 of his colleagues recently released a crypto market structure framework that includes vague language around financial privacy.

The group of Democrats, which included Senate Banking ranking member Ruben Gallego (AZ), Kirsten Gillibrand (NY), and Catherine Cortez Masto (NV), stated in the first page of the document that digital assets legislation should be guided by certain values, include “protecting financial privacy while denying bad actors access to the financial system.”

In the fifth section of the framework, they outlined what this looks like.

The outline included the following points:

  • Require digital asset platforms to register with FinCEN as “financial institutions” under the Bank Secrecy Act (BSA), while adopting anti-money laundering/combatting the financing of terrorism (AML/CFT) policies

  • Address bad actors’ use of DeFi platforms to circumvent illicit finance controls

  • Ensure that crypto platforms serving U.S. customers comply with sanctions and AML/CFT requirements, even if domiciled abroad

  • Shape ecosystems to isolate non-compliant platforms that enable illicit activity

Intentionally Unclear Language?

What does “addressing bad actors’ use of DeFi platforms” look like? What role will U.S. regulators play in “shaping ecosystems to isolate non-compliant platforms”?

These questions remain unanswered as per this framework, which is notably less detailed and comprehensive than the draft of the CLARITY Act that the Senate Banking Committee recently released.

What comes to mind I think of the U.S. government “shaping digital asset” ecosystems is its pushing for the implementation of a digital ID that only allows for “good actors” to transact. Former CFTC Chair Tim Massad said he is in favor of such a scheme in an interview I conducted with him earlier this year.

Luckily, this would be nearly impossible to technically implement for bitcoin. Smart contract blockchain networks, on the other hand, would be more easily susceptible to the censoring of transactions, as the government could mandate that the smart contracts on the network include certain rules and stipulations within the code that prohibit bad actors from transacting.

What Democrats Should Do

If Democrats are looking to engage in Bitcoin and crypto regulation in good faith, they should be more specific about their intentions and include further details about how they plan to combat illicit finance while still preserving user privacy.

And in their next round of messaging, they should clearly define how they plan to “shape ecosystems” as well as provide clear definitions for what they mean by terms like “platforms”. For example, when they say “platforms” are they referring to centralized entities that hold users’ private keys like Coinbase or Kraken, or does the term also encompass services like Samourai Wallet or Tornado Cash?

They are questions that must be answered if Bitcoin and crypto advocates and enthusiasts are to trust that Democrats do in fact want to enshrine into law the right for Bitcoin and crypto users to preserve their privacy.

Tyler Durden Wed, 09/10/2025 - 18:40

MIT President Sally Kornbluth Is A Magnet For Research Fraud And Medical Misconduct

Zero Hedge -

MIT President Sally Kornbluth Is A Magnet For Research Fraud And Medical Misconduct

Authored by Paul D. Thacker via The DisInformation Chronicle,

The Boston Globe published an incredible investigation highlighting yet another corruption scandal in academic research, this time involving two prominent research institutions—Duke University and MIT—tied together by Sally Kornbluth, the former head of research at Duke, who MIT picked as president in 2022. The Globe article adds to Kornbluth’s headaches as she has also been caught up in a dragnet of prestigious universities investigated by the Trump administration for alleged antisemitism and liberal political bias.

The White House spotlight likely spurred the Globe reporters to delve into Kornbluth’s past at Duke, uncovering court documents and sworn depositions by prominent Duke scientists. The Globe reported that, while at Duke, Kornbluth ignored scientific criticism and a whistle-blower pointing to corrupt, harmful research by a Duke physician treating patients dying from cancer.

But did anything happen to Kornbluth? No!

As the Globe documents, Kornbluth kept failing up, gaining ever more prominent positions at Duke, even as the university dealt with an increasing series of unethical research problems. (The Globe didn’t report on several other examples of Duke research misconduct, which are detailed below.)

The Boston Globe’s reporting doesn’t plow much new ground. In fits and starts, several media outlets covered the cancer research scandal overseen and ignored by Kornbluth while she was at Duke. The Globe just put all the pieces together into a comprehensive narrative (The Globe article is behind a paywall, but an MIT professor sent me a copy which you can read here).

Much of the news about Duke’s cancer research corruption was broken by The Cancer Letter, such as a whistle-blower complaint by a medical student, who Kornbluth and other Duke administrators tried to shut down: “Duke Officials Silenced Med Student Who Reported Trouble in Anil Potti’s Lab.” The Cancer Letter has an entire series devoted to the Duke Scandal which you can find here.

The federal government later charged Duke’s cancer physician with research fraud, finding that he had submitted false information in his grants to the National Institutes of Health (NIH). Several of his papers published in premier research journals were retracted and he was banned for several years from receiving NIH grants.

But it doesn’t end there.

Kornbluth evaded any consequences from this highly detailed example of research fraud that harmed dying patients. On the contrary, Duke promoted Kornbluth to Provost, the chief academic officer overseeing the university’s teaching and research mission. “The Provost also engages with issues concerning admissions, financial aid, information technology, and all other facets of university life touching on academics,” explains Duke on their website.

Duke made national news for more fake scientific studies in 2019, while Kornbluth was Duke’s academic provost.

The Globe didn’t report this, but the Justice Department forced Duke to pay $112.5 million in 2019 because university officials submitted bogus data to win federal grants on a mouse study project.

“This settlement sends a strong message that fraud and dishonesty will not be tolerated in the research funding process,” said one federal official. “We will continue to take appropriate legal measures to ensure a fiscally sound system that protects grant funds.”

The Boston Globe missed another Kornbluth medical scandal that stayed hidden until I reported it in 2022—the same year MIT picked Kornbluth as their university president. Yes, Kornbluth is ensnared in even more academic sleaze.

While Duke Provost, Kornbluth ignored a 2019 complaint filed by Massachusetts’ Attorney General Maura Healey that charged Duke physician Ralph Snyderman and other board members of Purdue as co-defendants with the Sackler family for addicting millions of Americans on opioids. “Defendants Peter Boer, Judith Lewent, Cecil Pickett, Paulo Costa, and Ralph Snyderman took seats on the Board and knowingly advanced the Sacklers’ scheme.” The Daily Mail later named Ralph Snyderman in their report on this lawsuit.

As I reported in 2022, the Massachusetts Attorney General’s complaint runs 274 pages, with Duke’s Ralph Snyderman named 76 times. “Together with the Sacklers, they controlled the unfair and deceptive sales and marketing tactics Purdue used to sell its opioids in Massachusetts.”

According to the complaint, Snyderman voted with the Sacklers to hire hundreds more sales reps to sell opioids; to implement incentive compensation policies that aggressively drove opioid sales; and, to pay out millions of dollars to already convicted criminals to help the Sacklers keep their loyalty.

In November 2020, Purdue pleaded guilty in federal court to several felonies, admitting that it marketed and sold dangerous opioid products, lied to the Drug Enforcement Administration about steps it had taken to prevent the diversion of opioids, and that it paid kickbacks to encourage prescribing.

Yet, Kornbluth did nothing to address Snyderman’s corrupt behavior.

When I contacted Duke in 2022 to ask if they were looking into the Attorney General’s complaint against Snyderman, a Duke spokesperson emailed me, “I can’t comment on Ralph Snyderman. I’ll decline comment.”

Who knows what other research scandals MIT’s Sally Kornbluth has tried to bury or will ignore in the future?

The DisInformation Chronicle is a community-supported publication. To receive new posts and support this work, consider becoming a free or paid subscriber.

Tyler Durden Wed, 09/10/2025 - 17:00

Nepal Army Takes Over Capital As Politicians Flee By Helicopter, Mayhem Worsens

Zero Hedge -

Nepal Army Takes Over Capital As Politicians Flee By Helicopter, Mayhem Worsens

The collapse of the Nepal government situation has gone from bad to worse overnight and into Wednesday. Parliament has gone up in flames, Prime Minister KP Sharma Oli has resigned, there have been dozens of deaths and injuries - including among some government officials or their families. 

Government ministers have been seen fleeing the capital, chased by enraged mobs of mostly youth, sick of government corruption and following the latest attempt to outright ban a large number of popular social media sites, including Facebook, X, Instagram, WhatsApp and YouTube.

Via Reuters/DW

But apparently the social media ban days ago was simply the straw that broke the camel's back. "The unrest started in early September, when a group of young Nepalis, fed up with seeing politicians’ children posting about their designer handbags and luxury travel while most people struggle to make ends meet, organized a peaceful protest," CNN reviews.

"Anger had been brewing for years about the country's worsening youth unemployment crisis and lack of economic opportunities, exacerbated by what many viewed as a growing disparity between the country’s elite and regular people," the report adds.

Residents of top politicians in Kathmandu have been reported attacked and in some cases damaged or set on fire, including the home of the now former prime minister of the country.

Army helicopters were seen ferrying government ministers to safety:

The Finance Minister was beaten by a mob, with some barely escaping the wrath of the rioting...

The moment at which police opened fire on crowds was the tipping point, with at least 19 demonstrators killed, as we detailed Tuesday. Even after the social media ban was reversed, the rioting became more intense. The police had merely described using riot control measures like rubber bullets and tear gas.

Part of the outrage among youth was an increasingly feeling of economic isolation from the rest of the world, and little hope for progress, in a tiny mountainous country where personal remittances accounted for over 30% of Nepal’s GDP.

Nepal’s parliament, which was built in 1903, engulfed in flames...

Recent World Bank figures have put youth unemployment of individuals aged 15 to 24 at just over 20%. So when the most popular social media apps were recently banned by the central government, popular rage sparked across the country.

Now, Kathmandu has been taken over by the army, with an effective state of martial law and indefinite curfew on until authorities can restore order. The Army is warning against looting:

The man in the spotlight is General Ashok Raj Sigdel, the Chief of the Army Staff, who has appealed to protesters to engage in talks to find a peaceful way out.

The 58-year-old General, who took over the top post last year, addressed the shaken nation in a televised address last night. "We appeal to the protesting group to halt protest programs and come forward for dialogue for a peaceful way out for the nation. We need to normalize the present difficult situation and protect our historical and national heritage and public as well as private property, and to ensure safety to the general public and diplomatic missions," he said.

The Kathmandu Post

People have been ordered to stay indoors. There are reports of tanks in the streets, with some of the youth claiming that the army troops are on their side, against the police.

Tyler Durden Wed, 09/10/2025 - 16:40

Pages