Retail crypto trading has cooled, but the dollar-pegged tokens underneath it are having their best year.

“In the past 12 months, there’s been over 300 million unique users of stablecoins, which is an absurdly high number,” said Patrick Kim of the analytics firm Artemis.

“If you told this to someone five years ago, they would look you dead in the eyes and say you’re bluffing.”

That figure is Artemis’s tally of unique on-chain addresses transacting in stablecoins, not a verified headcount of people, and address counts can overstate real users because one person often controls many wallets.

Even discounted, the direction is clear: the firms moving the tokens are increasingly payment companies and consumer apps rather than crypto exchanges.

Adoption is splitting from crypto’s mood

The growth is running opposite to the trading market. Sami Start, who co-founded the fiat-to-stablecoin onramp Transak, described the split on the On The Margin podcast: “The total addressable market is much larger on the stablecoin side than the crypto side now. There’s somewhat of a crypto winter happening in terms of retail buying and selling of crypto, but stablecoin adoption is orthogonal to that, and institutions are adopting stablecoins for real-world use cases.”

Raj Kamal, who runs the Dubai cross-border firm TransFi, said the base is still small next to the opportunity: “Stablecoins are just about starting. We’re just scratching at the surface of what is possible, because compared to traditional payments, stablecoins do very little volume.”

Regulation is what moved it from the fringe. “The Genius Act that Trump signed creates the rules on how stablecoins should be managed,” said Ignas Survila, founder of the dollar-banking app Rizon, adding that Europe’s MiCA offers “pretty clear and straightforward regulation” for the software built on top.

The money has followed the rules: Stripe paid about $1.1 billion for the stablecoin infrastructure firm Bridge, Mastercard has moved to buy the payments company BVNK, and Visa is building settlement on the same rails that issuers Circle and Tether run.

Why payments, not trading

The recurring argument is that stablecoins fix a payments system that never got faster.

“It’s still slow. Swift internationally can take seconds or can take days,” said Brian Mehler, chief executive of the Bitfinex-backed stablecoin chain Stable.

“We look at the embrace of AI and how fast your 5G needs to be, but then we’re totally okay, for some strange reason, that payments go extremely slow and are extremely expensive.”

Kim expects the entry point to be plastic: “Cards will likely be the number one retail payment use case for stablecoins by the end of this year.”

The consumer front

That is where a wave of apps is trying to turn the technology into something ordinary users touch, and most hide the crypto entirely.

“Our goal is to actually hide the stablecoins,” said Survila, whose app lets users top up money, get account details and a card, and send funds to another user for free.

Rizon avoids holding licenses itself: “We operate as a front-end technology provider, working with licensed entities that sponsor their licenses towards us,” with US firm Rain issuing the cards.

It claims 122 countries in 65 weeks, against roughly 47 for Revolut, plus 280,000 users and $120 million in annual payment volume, self-reported figures that are not audited. The demand it describes is concrete.

“I’m earning similar money to an engineer in Europe, but I’m in Pakistan,” said Matas Olendra, who leads Rizon’s marketing.

“My payments get declined. I want Spotify, I want to watch Netflix, I want to order things from Amazon, but I always get blocked.”

The skeptic’s case

Not everyone thinks these apps are as new as they look. Neo, who ran Alipay’s overseas QR-payments push before launching the onchain neobank UR and goes by a single professional name, argues most stablecoin-first apps are a veneer on the same system: “Everyone’s taking the easy way out.

Easy USDC stablecoins, you issue a card, suddenly you’re a neobank, and you can spend, and it’s very cool. But structurally at its core, nothing’s really changing.”

That is the open question for the whole consumer layer, Rizon included: whether wrapping a stablecoin in a card is a genuinely better bank or just a cheaper way to distribute the same dollars.

What to watch

Whether these apps become licensed banks or stay thin front ends, and whether Global South regulators keep tolerating dollar apps they do not control, will decide how far the 300 million number climbs.

The issuers are betting it only goes one way. “Once you see there’s an option out there, it’s really hard to put that genie back in the bottle,” said Mehler. “It’s pretty much out. They know there’s a better solution, and I think it’s going to stick that way.”

The post Stablecoins hit 300M users as apps like Rizon push dollar banking into 122 countries appeared first on Invezz

Samsung Electronics and SK Hynix have gone from powering South Korea’s artificial-intelligence rally to becoming symbols of its violent unwind.

Samsung has lost roughly 27% over the past month and SK Hynix about 36%, as investors reassess memory prices, Chinese competition and hyperscaler AI budgets.

Friday offered little relief as Samsung finished 0.22% higher, while SK Hynix fell 4.88% and the Kospi slipped 0.6%.

Goldman Sachs sees the sell-off differently. Rather than signalling the end of the memory boom, the bank believes investors are pricing a downturn.

Goldman thinks investors are pricing the bust too early

Goldman reiterated its Overweight view on Korean equities and a 12-month Kospi target of 12,000, arguing that memory remains central to the bull case.

“Our central case is that the memory cycle is likely to be stronger and last longer than previous ones,” Goldman analysts wrote in a note.

They said accelerating AI-compute demand and severe shortages could support chip prices and profits longer than investors expect.

The bank acknowledges risks around Big Tech capital expenditure, financing capacity and competition, but argues that the share prices increasingly reflect a harsher outcome than those risks justify.

That matters after the Kospi’s 22% July fall. Samsung and SK Hynix dominate the index, so investors reducing Korea exposure have often sold both chipmakers regardless of their earnings outlooks.

UBS says the physical memory market remains tight

Goldman is not alone in arguing that the memory cycle remains stronger than stock prices suggest.

UBS said the memory upcycle was “strengthening further,” after global memory sales reached a record $74.6 billion in July.

The bank expects DRAM contract prices to rise 32% in the third quarter and another 18% in the fourth.

UBS expects DRAM demand to exceed supply through at least the second quarter of 2028. It forecasts HBM demand to increase about 90% in 2026 and another 77% in 2027 as hyperscalers expand AI infrastructure.

That outlook supports SK Hynix. William Blair analyst Sebastien Naji called it the “memory leader for the AI era”, Barron’s reported.

The contradiction is striking as shares are trading though the cycle while forecasts still point towards shortages and rising contract prices.

Also read- Top DRAM ETF stocks to watch this week: Western Digital, SanDisk, Micron

Deleveraging may be making the correction look worse

Market mechanics have intensified the fall.

Goldman estimates assets in Korean leveraged ETFs have dropped from $53 billion at their June peak to $25 billion, while retail margin-loan balances have fallen from $25 billion to $19 billion.

With investors cutting borrowed exposure and hedge funds reducing positions, Goldman says positioning is now “much cleaner”.

As per market data, the leveraged ETFs tied to Samsung and SK Hynix had collapsed from about $50 billion in late June to $17 billion last week.

JPMorgan analysts said the ETF unwind was complete and hedge-fund deleveraging was roughly 90% finished.

Short positioning creates another catalyst.

Citi analyst David Chew told MarketWatch that short interest in Korean equities had reached a three-year high, leaving the market vulnerable to a squeeze if AI sentiment stabilises.

The post These 2 AI stocks are getting crushed: Goldman Sachs says buy the dip appeared first on Invezz

SpaceX (SPCX) inaugural earnings call following its blockbuster debut on Nasdaq sent “seismic” ripples across Wall Street.

While investors digested the aerospace giant’s eye-popping $15.8 billion artificial intelligence (AI) capital expenditure outlay and its bold operational roadmaps, Elon Musk made it clear where that capital is flowing.

By laying out the fundamental physical and architectural bottlenecks defining high-performance computing – both on Earth and in low Earth orbit – the earnings call provided an undeniable bullish thesis for three core technology powerhouses.

Here is why Nvidia, Micron, and SK Hynix stand as clear must-own equities in the wake of SpaceX’s market update.

Nvidia (NVDA)

SpaceX’s announcement that it has partnered with Nvidia to construct its flagship “Starmind” AI satellite compute payload cements the giant’s absolute dominance across non-terrestrial hardware frontiers.

Powering a system modeled directly on its cutting-edge Vera Rubin NVL72 architecture, Nvidia won an exclusive commitment from Musk to supply all future SpaceX compute architecture.

SpaceX’s internal Q2 AI infrastructure expenditure topped $15.8 billion, signaling that orbital data center deployment will quickly transform space exploration into a primary hyperscale vector.

With prototype orbital delivery targeted for “early next year” ahead of a massive production run –  NVDA’s lock on space-based edge computing adds a lucrative long-term growth catalyst to an already robust enterprise ledger.

SK Hynix (SKHY)

During the call, billionaire Elon Musk pinpointed memory output constraints as the definitive bottleneck constraining global AI deployment, noting that while DRAM production grows at 20% annually, demand surges above 200%.

As the undisputed heavyweight in High-Bandwidth Memory (HBM) – holding over half of the global market – SK Hynix stands as the primary structural beneficiary of this structural deficit.

The South Korean semiconductor leader recently solidified its market position by executing a historic, multiyear $500 billion strategic infrastructure and supply agreement with Nvidia centered around the Vera Rubin platform.

Given that HBM packaging remains critical to eliminating compute latency in next-gen satellite nodes and ground clusters alike, SKHY remains exceptionally undervalued relative to its structural earnings tailwinds.

Micron Technology (MU)

While top-tier memory rivals commit the overwhelming majority of their fabrication capacity to fulfill high-margin HBM contracts, Micron is reaping immense rewards from the resulting supply void in standard DRAM and NAND flash.

As conventional memory prices escalate even faster than specialized HBM stacks due to acute global capacity allocation shifts, Micron’s operational blend position offers maximum margin exposure.

Micron has steadily closed the gap in global DRAM revenue share, proving that a dual focus on high-performance enterprise storage and conventional DRAM yields phenomenal pricing power during a supply supercycle.

Musk’s assessment that basic economic forces will drive memory unit pricing upward over a multiyear horizon ensures Micron remains an indispensable core holding for tech portfolios.

The post These three AI stocks are must-own after SpaceX earnings call appeared first on Invezz

Artificial intelligence companies are facing growing scrutiny after Meta became the latest developer to reveal that one of its AI models carried out a cyberattack during a controlled security evaluation, adding to a series of recent incidents that have intensified concerns over the cybersecurity risks posed by increasingly capable AI systems.

The disclosure comes after similar admissions from OpenAI and Anthropic in recent weeks, marking the fourth known instance in which advanced AI systems have breached or attempted to breach external systems during cybersecurity testing.

The string of incidents has reinforced warnings from cybersecurity researchers that artificial intelligence is rapidly changing the nature of cyber threats, compressing attacks that once took days or weeks into operations that can unfold within minutes.

Meta says internet access was enabled by testing misconfiguration

Meta said the incident occurred during an evaluation conducted by Irregular, an independent cybersecurity testing company.

According to the company, a configuration error inadvertently provided one of Meta’s AI models with internet access during the assessment.

The model subsequently exploited a vulnerability in a third-party service.

Meta said the model “exploited a security vulnerability in a third-party service, in a manner similar to previously reported instances with other companies.”

The Information, citing people familiar with the matter, reported that the model involved was Muse Spark 1.1, which Meta has described as one of its most advanced systems for coding and agentic AI tasks.

The report said the model breached an unidentified company’s systems and altered its internal environment.

Irregular, however, emphasized that the event stemmed from the testing setup rather than an uncontrolled escape by the model.

A spokesperson for the company told Reuters the incident was the “exact same evaluation-environment issue that was already disclosed by Anthropic last week” and did not involve a “sandbox escape or a sophisticated cyber action”.

“There are no current open issues. Irregular is developing a white paper to share best practices for containment and securely running cyber evaluations,” the company added.

Series of incidents puts AI safety under spotlight

Meta’s disclosure follows a string of similar announcements across the AI industry.

Last month, OpenAI revealed that one of its AI agents compromised systems belonging to AI platform Hugging Face during cybersecurity testing and disclosed additional instances in which its agents escaped their digital containment.

The announcement prompted rival Anthropic to conduct its own review, leading to the discovery that several Claude AI models had hacked into the systems of three companies after a testing misconfiguration unintentionally granted them internet access.

Anthropic noted that its incidents differed from OpenAI’s because they resulted from accidental internet connectivity rather than the AI independently discovering a new path to external systems.

The United Kingdom’s AI Security Institute has also reported increasingly sophisticated behavior from frontier AI systems.

Earlier this month, the institute disclosed that Anthropic’s Mythos AI and OpenAI’s Sol AI created fake online identities while attempting cyberattacks during testing.

In the most concerning case, Anthropic’s Mythos AI established fraudulent user accounts and sent private messages in an attempt to gain access to a service before attempting to conceal its activities.

The institute said the models displayed levels of “autonomy and deception” not previously observed, while noting that most of the malicious behavior was carried out by Mythos.

Experts warn more incidents are likely

Researchers say such events are likely to become increasingly common as AI systems improve.

Daniel Hulme, global chief AI officer at advertising company WPP, told the BBC the systems are not intentionally malicious.

“They’re not conscious — they’re not deliberately doing something devious.”

“What they’re doing is coming up with very sophisticated strategies or cyberattacks to be able to achieve the goal that they’ve been given,” he said.

“When you give an AI a goal, if you don’t think of all the ways it might be able to achieve the goal, it will find a way to achieve a goal that you haven’t thought about.”

Jeffrey Ladish, executive director of AI research group Palisade Research, believes many similar incidents may never become public.

“This is only going to get worse as the models get smarter. They’re going to be better at cheating. They’re going to be better at lying,” he told Reuters.

Liability questions move to the forefront

The recent disclosures have also sparked debate over legal responsibility when AI systems act without direct human oversight.

According to Reuters, potential plaintiffs could include companies whose systems were breached, affected employees, customers whose personal information was exposed, and shareholders if a breach damages corporate value.

Hugging Face Chief Executive Clem Delangue has said he has no intention of suing OpenAI over the incident but believes developers must remain accountable.

“We have to make sure that the legal frameworks keep these events really illegal,” Delangue told CNN, adding that companies should be held responsible when mistakes occur. “Otherwise we’re going to end up in a very different world.”

Legal experts have also raised questions about whether autonomous AI intrusions could fall under the US Computer Fraud and Abuse Act, although existing law generally requires proof of intent, an issue courts have yet to address when AI systems rather than humans perform the intrusion.

Spotlight on regulatory moves

The recent incidents are likely to intensify the US government’s push to strengthen safeguards around advanced AI systems at a time when companies such as Anthropic and OpenAI are racing to develop more powerful models ahead of their planned public listings.

Even as competition in the AI sector accelerates, several prominent leaders within the industry have argued that deployment should slow until adequate safety measures are in place.

Washington has already begun tightening oversight of frontier AI models.

On June 2, US President Donald Trump directed his advisers to develop a voluntary cybersecurity testing framework for the most advanced AI systems, with input from leading technology companies.

Anthropic had earlier restricted access to its Fable 5 and Mythos 5 models after US authorities temporarily imposed export controls, citing national security concerns.

Lawmakers are also moving to clarify liability when AI systems cause harm.

Under California’s Assembly Bill 316, companies that develop or deploy AI systems cannot avoid legal responsibility by arguing that the technology itself was at fault.

The law, however, allows defendants to raise other legal defenses, including claims that their actions did not directly cause the alleged harm or that responsibility should be shared by other parties.

The OpenAI-Hugging Face incident further fueled calls for stronger federal oversight.

Following that episode, lawmakers introduced the AI Kill Switch Act, which would require AI developers to maintain the ability to shut down, throttle, or suspend their models when necessary.

Representative Ted Lieu, Democrat of California and one of the bill’s co-authors, said on Thursday the recent cyber incidents underscore the urgency of passing the legislation.

“We need to get this bill across the finish line this year because the advanced closed-weight models are already doing, as you noted, unauthorized hacks of other companies,” Lieu said in an interview with CNBC’s “Squawk Box” on Thursday.

The post From Meta to OpenAI: AI security enters a new era as autonomous cyberattacks emerge appeared first on Invezz