TheChek News Flow

Tiffany releases its ‘gravity defying’ diamond

The new setting is part of a collaboration with Louis Vuitton’s creative director Pharrell Williams

This week Tiffany & Co debuted the new Titan Setting, which creates the illusion of a hovering diamond. The effect is possible thanks to the Floeting “platform”, a patented creation from a New Zealand jewellery company that includes a micro-engineered round brilliant diamond and its accompanying mounting, set without prongs or bezels.

“The diamond appears to sort of defy gravity,” says Floeting’s co-inventor Chris Benham, speaking from Wellington.

The launch is the latest effort from the LVMH-owned house to elevate its standing in the high-end jewellery segment, where its promise has been slow to materialise despite the French conglomerate’s investment in its rebranding since 2021. The LVMH group does not report individual brand performance, but sales of its watches and jewellery division struggled to perform in 2024, closing the year with revenues down 2 per cent over 2023 on an organic basis. Despite an improvement in Tiffany’s performance, investment in the category weighed on operating profit.

Under LVMH, Tiffany in 2023 unveiled a complete redesign of its Manhattan flagship store, which has been described as “the biggest investment in the history of luxury”, and a number of collaborations with names such as Supreme, Fendi, Nike and Beyoncé. The Titan Setting itself is part of a new instalment of the Titan collection, a collaboration with Louis Vuitton men’s creative director, recording artist and producer Pharrell Williams that the house debuted last year.

Tiffany is the first jewellery house to use the Floeting platform, which took 17 years to perfect, with input from science and gemstone laboratories, and even an ex-Formula One engineer. The stone that comes with it features 89 facets, about 50 per cent more than your average round brilliant diamond, and ultra-precise micro grooves to connect the setting.

The development of the first round brilliant diamond is credited to Belgian diamantaire Marcel Tolkowsky in 1919, and the cut has long been Tiffany’s wheelhouse — its patented Tiffany setting engagement ring features a round brilliant diamond held by six prongs. But the shape today has become increasingly indistinguishable in the market and a harder sell as consumers seek more interesting and unusual fancy cut stones.

Tiffany’s move to partner with Floeting follows a trend among top jewellery houses seeking to strengthen brand identity by developing recognisable cuts, settings and designs. Tobias Kormind, managing director of 77 Diamonds, one of the largest online jewellers in Europe, says that “there’s a very strong commercial incentive right now to make everything about the brand — and that’s where the cuts really come in, [as they] allow brands to differentiate significantly”.

Most prominent in this effort is Louis Vuitton, which has cut stones into the star and flower shape of its iconic monogram print in both its high and fine jewellery collections. Francesca Amfitheatrof, artistic director for watches and jewellery at Louis Vuitton, calls the LV Monogram cuts the “holy grail” for clients, who are “totally seduced” by them, she says. Meanwhile, another LVMH stablemate, Chaumet, has also created a unique hexagon-shaped cut called Taille Impératrice (empress-cut in French), featuring 88 facets, which riffs off the house’s signature bee-themed jewellery collection.

Engineering new cuts, however, requires a considerable amount of time and investment. For this reason smaller independent jewellers have been designing around signature fancy cuts, such as Bond Street’s David Morris, who is known for its regular use of rose cut diamonds. Rather than embarking on the development of a completely new cut and setting, Tiffany has partnered with Floeting, which had already spent almost two decades working on its creation.

“The idea that things move and perform beyond explanation is fascinating,” says Williams of why he was drawn to the Floeting platform, which has been employed in a diamond earring with a fishtail setting, and a ring encircled by four Titan spears that highlight the central diamond. “For one to look and question ‘Wait a minute, how is that stone there?’”

The collection also introduces a unisex-style pendant, earrings, necklace, bracelet and ring punctuated with earthy grey Tahitian pearls over 11mm in diameter. Men are increasingly drawn to Tiffany Titan, according to the house and the Floeting novelty may further that interest. Christine Douglas, Floeting’s co-inventor, says that the setting has attracted a new segment of male clients who haven’t traditionally considered diamonds, in particular engineers, architects and those working in tech.

“We thought that people might be put off by the technology, but in fact it’s been the opposite,” she says. Tech bros who are embracing diamond jewellery with a touch of Pharrell magic? That could be the boost that Tiffany needs.

Stefano Buono récupère les missions de Ludovic Vandendriesche, directeur général de Newcleo SA, tout récemment rebaptisée Newcleo opérations.

Après un automne déjà bien rempli, la 􀀁n de l’année 2024 a été riche en événements pour Newcleo. L’entreprise qui planche sur un projet de petit réacteur nucléaire modulaire de 4e génération (Advanced modular reactor - AMR), refroidi au plomb, a perdu l’un de ses hommes forts en décembre. Ludovic Vandendriesche, directeur général depuis septembre 2023 de la 􀀁liale lyonnaise, Newcleo SA, a quitté ses fonctions. La gre􀀂e n’a pas pris avec cet ancien DG du spécialiste des solutions et services de sécurité incendie DEF, également passé par la CNIM et le groupe d’ingénierie Onet technologies. Son départ a été entériné par le conseil d’administration le 9 décembre.

Ses missions sont reprises par le fondateur de Newcleo, Stefano Buono, déjà PDG de la holding, qui a annoncé la 􀀁nalisation du transfert de son siège social du Royaume-Uni vers la France le 30 septembre. Objectif : rapprocher l’entreprise de l’Europe et notamment de ses investisseurs.

D’autant qu’après la clôture de sa première levée de fonds (535 millions d’euros), Newcleo entend en ouvrir une seconde, en France. La maison mère, qui compte l’ex-ministre des armées Florence Parly à son conseil d’administration, a par ailleurs décidé de changer l’appellation de sa 􀀁liale Newcleo SA pour celle de Newcleo opérations. Autre changement dans la gouvernance de l’entité lyonnaise : le directeur administratif et 􀀁nancier Matthieu Houmeau en a été nommé administrateur.

Techniquement, Newcleo a également franchi une nouvelle étape ces dernières semaines. La société, qui revendique 700 actionnaires dont le fonds de pension italien Inarcassa, des family o􀀃ces ou encore des sociétés industrielles et d’ingénierie comme la transalpine Walter Tosto et la française Ingérop, a annoncé le 16 décembre avoir déposé son dossier d’option de sûreté (DOS) - examen des options de sûreté préalable à une demande d’autorisation de création d’installation nucléaire de base - auprès de l’Autorité de sûreté nucléaire et de radioprotection (ASNR). Sa demande porte sur son projet d’installation d’essais d’assemblages combustibles.

SoftBank would lead $40 billion round for the ChatGPT maker, some of which would go to Stargate AI infrastructure venture

OpenAI is in early talks to raise up to $40 billion in a funding round that would value the ChatGPT maker at $340 billion, according to people familiar with the matter.

SoftBank would lead the round and is in discussions to invest between $15 billion and $25 billion, The Wall Street Journal reported Wednesday.

The Japanese company is helping to assemble investors for the rest of the round, one of the people said.

OpenAI was last valued at $157 billion in October, when it raised $6.6 billion. A near-doubling of its value in just a few months would be extraordinary even by the standards of Silicon Valley’s current AI boom.

The funding will be used in part to help OpenAI fulfill its roughly $18 billion commitment to Stargate, a joint venture with SoftBank and others to finance the construction of new data centers in the U.S. powering OpenAI’s technology across the country. The startup also expects to use the cash to fund its money-losing business operations.

Spinning off the “non-core asset” will net Kering proceeds of about 350 million euros.

As Kering focuses on its core luxury brands — including Gucci, Saint Laurent, Bottega Veneta — the company is handing off some of its landlord duties to an American expert.

Simon Property Group bought 100 percent of Kering’s The Mall Luxury Outlets, which it established in 2001. The Mall operates two outlet centers in Italy — one near Florence in Leccio and another on the Italian riviera in Sanremo.

The deal nets Kering proceeds of about 350 million euros.

According to a statement released by both parties: “Kering’s brands will maintain a presence in these two very high-end shopping villages, the strategy implemented by Kering aiming at gradually concentrating its outlet distribution to a limited number of exclusive venues.”

Kering, which will report on its fourth-quarter results next month, has been tightening its belt and cutting costs, partially as its Gucci brand has struggled to find its footing and saw organic sales drop 25 percent in the third quarter.

In October, François-Henri Pinault, chairman and chief executive officer of Kering, said, “Our absolute priority is to build the conditions for a return to sound, sustainable growth, while further tightening control over our costs and the selectivity of our investments.”

Simon is one of the world’s largest luxury mall operators and is expected to step in without interruption to The Mall’s operations.

This is Simon’s second big outlet move this week.

The real estate firm said on Wednesday that it agreed to purchase a large site in Thompson’s Station, Tenn., to develop a “luxury shopping and lifestyle destination” known as Nashville Premium Outlets.

Simon expects to break ground on the 325,000-square-foot, mixed-use center next year with plans for roughly 75 “best-in-class” retailers, restaurants and a hotel.

This is all familiar territory for Simon, which operates the mammoth Woodbury Common Premium Outlets in upstate New York, housing Gucci, Prada, Loro Piana, Saint Laurent, Brunello Cucinelli and many others.

Chinese AI lab DeepSeek provoked the first Silicon Valley freak-out of 2025 after releasing open versions of AI models that compete with the best technology OpenAI, Meta, and Google have to offer.

DeepSeek claims to have built its models highly efficiently and quickly (though some are skeptical of these claims), and is providing these models at a fraction of the price American AI companies charge. The development has rattled not only tech giants but the highest levels of the U.S. government, which fear that China is pulling ahead in the AI arms race.

“I wouldn’t be surprised if a lot of AI labs have war rooms going on right now,” said Robert Nishihara, the co-founder of AI infrastructure startup Anyscale, in an interview with TechCrunch.

The rise of DeepSeek marks an inflection point for Silicon Valley’s AI landscape. AI CEOs, founders, researchers, and investors tell TechCrunch that DeepSeek’s models have major implications for American AI policy. Moreover, these experts say, the models serve as an indicator of the accelerating rate of AI progress.

“Of course [DeepSeek] was over-hyped,” Ravid Shwartz-Ziv, an assistant professor at NYU’s Center for Data Science, told TechCrunch. “But it’s still very interesting, and there’s a lot we can take from it.”

New ways to get AI thinking

One of DeepSeek’s key innovations in creating its R1 model was “pure reinforcement learning,” a trial-and-error approach, according to Workera CEO and Stanford adjunct lecturer Kian Katanforoosh.

Katanforoosh compared DeepSeek’s breakthrough to a kid figuring out not to touch a hot plate by accidentally burning themselves.

“[A kid] might touch a hot plate, get burned, and quickly learn not to do it again,” Katanforoosh said via text. “That’s pure reinforcement learning — learning from trial and error based on feedback […] DeepSeek’s method is all about letting the model learn through experience alone.”

DeepSeek seems to have relied more heavily on reinforcement learning than other cutting edge AI models. OpenAI also used reinforcement learning techniques to develop o1, which the company revealed weeks before DeepSeek announced R1. OpenAI’s upcoming o3 model achieves even better performance using largely similar methods, but also additional compute, the company claims.

Reinforcement learning represents one of the most promising ways to improve AI foundation models today, according to Katanforoosh. The term “foundation models” generally refers to AI models trained on massive amounts of data, like images and text from the web. It seems likely that other AI labs will continue to push the limits of reinforcement learning to improve their AI models, especially given the success of DeepSeek.

Just a few months ago, AI companies found themselves struggling to boost the performance of their foundation models. But the success of methods such as reinforcement learning and others, like supervised fine-tuning and test-time scaling, indicate that AI progress may be picking back up.

“R1 has given me a lot more confidence in the pace of progress staying high,” said Nathan Lambert, a researcher at Ai2, in an interview with TechCrunch.

A turning point for AI policy

R1, which can be downloaded and run on any machine that meets the hardware requirements, matches or beats o1 on a number of AI benchmarks. While it’s not the first time we’ve seen the performance gap narrow between “closed” models like that of OpenAI and openly available models, the speed with which DeepSeek did it has taken the industry aback.

This may push the U.S. to increase its investment in open, or even fully open source, AI in order to compete with China. Martin Casado, a general partner at Andreessen Horowitz (a16z), tells TechCrunch that DeepSeek proves just how “wrongheaded” the regulatory rationale of the last two years has been.

“For AI, I think this just shows us that [the United States] is not alone in our technical capability,” Casado said in an interview. “Very competitive solutions can come from anywhere, but in particular, China. Rather than hampering U.S. innovation, we should invest strongly in it. Open source does not in some way enable China. In fact, disallowing our companies from doing open source means that our technology doesn’t proliferate as much.”

Casado seemed to be referring to former President Biden’s recently-repealed AI Executive Order and the vetoed California bill SB 1047, both of which a16z aggressively opposed. A16z has argued both measures prioritized preventing “outlandish” AI doomsday scenarios over American innovation. More broadly, Silicon Valley generally had success tamping down the “AI doom movement” in 2024. The real concern around AI, a16z and others have repeatedly said, is America losing its competitive edge to China.

That scenario seems much more tangible in light of DeepSeek’s rise.

Not for nothing, a16z is heavily invested in many of the open AI world’s largest players, including Databricks, Mistral, and Black Forest Labs. The VC firm may also play an outsized role advising the Trump Administration on AI. Former a16z partner Sriram Krishnan is now Trump’s senior policy advisor for AI.

President Trump said on Monday that DeepSeek should be a “wakeup call” for American AI companies, while praising the Chinese AI lab for its open approach. That lines up pretty closely with a16z’s stance on AI.

“DeepSeek R1 is AI’s Sputnik moment,” said a16z co-founder Marc Andreessen in a post on X, referencing the launch of the Soviet Union’s Earth-orbiting spacecraft decades ago that pushed the U.S. to seriously invest in its space program.

The rise of DeepSeek also appears to have changed the mind of open AI skeptics, like former Google CEO Eric Schmidt. Just last year, Schmidt expressed concern about the proliferation of Western open AI models around the globe. But in an op-ed published Tuesday, Schmidt said DeepSeek’s rise marks a “turning point” in the global AI race, and called for further investment in American open AI.

Looking ahead

It’s important not to overstate DeepSeek’s accomplishments.

For example, some analysts are skeptical of DeepSeek’s claim that it trained one of its frontier models, DeepSeek V3, for just $5.6 million — a pittance in the AI industry — using roughly 2,000 older Nvidia GPUs. The Chinese AI lab did not sprout up overnight, after all, and DeepSeek reportedly has a stockpile of more than 50,000 more capable Nvidia Hopper GPUs.

DeepSeek’s models are also flawed. According to a test by information-reliability organization NewsGuard, R1 provides inaccurate answers or non-answers 83% of the time when asked about news-related topics. A separate test found that R1 refuses to answer 85% of prompts related to China, possibly a consequence of the government censorship to which AI models developed in the country are subject.

Then, there are the claims of IP theft. OpenAI says that it has evidence that DeepSeek used its AI models to train its own, using a process called distillation. If true, this would be a violation of OpenAI’s terms, and would also make DeepSeek’s accomplishments less impressive. For example, Berkeley researchers recently created a distilled reasoning model for just $450. (Of course, OpenAI is currently being sued by a number of parties for allegedly committing copyright infringement in training its own models.)

Still, DeepSeek moved the needle with more efficient models — and it innovated. Lambert noted that, unlike o1, R1 reveals its “thinking process” to users. Lambert has observed that some users trust or believe AI reasoning models more when they see their internal process, during which they “explain their work.”

Now, we’ll have to see how America’s policymakers, and AI labs, respond.

Supporting digital assets that could marginalise the dollar is ‘dangerous’, investor letter says

The Trump administration’s embrace of cryptocurrencies is helping fuel a speculative mania that could cause “havoc” when prices collapse, hedge fund Elliott has warned.

The $70bn-in-assets firm took aim at the US government’s apparent enthusiasm for assets that have soared in price but have “no substance”, and also at politicians who are supportive of cryptocurrencies that could eventually become a rival to the US dollar, according to an investor letter seen by the Financial Times.

The fund “has never seen a market like this”, wrote Elliott, referring to the speculative investor frenzy it believes is currently gripping financial markets. It pointed to the artificial intelligence boom and high equity market valuations as signs of investors “acting like a crowd of sports bettors”.

“Crypto is ground zero” for the speculative surge across markets, not only due to the size it has grown to but also because of its “perceived proximity to the White House”, it added.

The “inevitable collapse” of the crypto bubble “could wreak havoc in ways we cannot yet anticipate”, it said.

Elliott’s criticism of the White House’s stance on crypto comes despite the fact that the firm’s founder, Paul Singer, is a longtime Republican donor who donated $56mn to conservative candidates in the 2024 election cycle, according to website OpenSecrets.

Singer, who has long warned of the dangers of crypto, has in the past been critical of Donald Trump’s economic policies. However, he donated $5mn to the “Make America Great Again” political action committee in the last election, whose goal was to support Trump.

Cryptocurrency prices have rallied hard since Trump’s election in November, after he promised he would make the US “the bitcoin superpower of the world”. Bitcoin surged from about $70,000 before the election to a record high of more than $100,000.

After his inauguration Trump signed an executive order to promote US “leadership in digital assets and financial technology while protecting economic liberty” and tasked a working group with evaluating the creation of a national digital asset stockpile.

Elliott’s letter also points to the “immense advantage” the dollar enjoys as the world’s reserve currency, and questioned why the US government would encourage the take-up of alternatives at a time when other countries are already trying to wean themselves off the greenback.

For any elected official to support “marginalising the dollar” was “profoundly dangerous”, wrote Elliott, highlighting the hundreds of millions of dollars spent helping politicians who were sympathetic to crypto get elected.

For instance, the Fairshake political action committee, which works to get pro-crypto candidates elected on both sides of the political aisle, spent $173mn in the 2023-2024 election cycle, according to OpenSecrets.

Trump’s own businesses have extended their links with the cryptocurrency markets. Alongside his sons and longtime business partners, Trump backed a crypto platform called World Liberty Financial last year, and earlier this month he and first lady Melania Trump both launched memecoins — a type of cryptocurrency with no underlying business model or cash flow.

Trump Media, in which the president owns a majority stake, said on Wednesday it would expand beyond social media and launch a financial services business that would invest up to $250mn into cryptocurrency and other assets.

Elliott, founded by Singer in 1977, is known as a fearsome activist investor, which takes on companies and even countries in heated boardroom or legal battles.

Elliott declined to comment.

Move over, DeepSeek. There’s a new AI champion in town — and they’re American.

On Thursday, Ai2, a nonprofit AI research institute based in Seattle, released a model that it claims outperforms DeepSeek V3, one of Chinese AI company DeepSeek’s leading systems.

Ai2’s model, called Tulu3-405B, also beats OpenAI’s GPT-4o on certain AI benchmarks, according to Ai2’s internal testing. Moreover, unlike GPT-4o (and even DeepSeek V3), Tulu3-405B is open source, which means all of the components necessary to replicate it from scratch are freely available and permissively licensed.

A spokesperson for Ai2 told TechCrunch that the lab believes Tulu3-405B “underscores the U.S.’ potential to lead the global development of best-in-class generative AI models.”

“This milestone is a key moment for the future of open AI, reinforcing the U.S.’ position as a leader in competitive, open-source models,” the spokesperson said. “With this launch, Ai2 is introducing a powerful, U.S.-developed alternative to DeepSeek’s models — marking a pivotal moment not just in AI development, but in showcasing that the U.S. can lead with competitive, open-source AI independent of the tech giants.”

Tulu3-405B is a rather large model. Containing 405 billion parameters, it required 256 GPUs running in parallel to train, according to Ai2. Parameters roughly correspond to a model’s problem-solving skills, and models with more parameters generally perform better than those with fewer parameters.

According to Ai2, one of the keys to attaining competitive performance with Tulu3-405B was a technique called reinforcement learning with verifiable rewards. Reinforcement learning with verifiable rewards, or RLVR, trains models on tasks with “verifiable” outcomes, like math problem solving and following instructions.

Ai2 claims that on the benchmark PopQA, a set of 14,000 specialized knowledge questions sourced from Wikipedia, Tulu3-405B beat not only DeepSeek V3 and GPT-4o, but also Meta’s Llama 3.1 405B model. Tulu3-405B also had the highest performance of any model in its class on GSM8K, a test containing grade school-level math word problems.

Tulu3-405B is available to test via Ai2’s chatbot web app, and the code to train and fine-tune the model is on GitHub. Get it while it’s hot — before the next benchmark-beating flagship AI model comes along.