Samsung SDS, the IT services arm of Samsung Group, is exploring stablecoin infrastructure, digital asset systems and AI-based payment models with Dunamu, the operator of Upbit, one of the biggest local cryptocurrency exchanges.

Samsung SDS said it is discussing potential cooperation with Dunamu on stablecoin infrastructure, digital asset systems and AI-based payment business models, CEO Lee Jun-hee said during the company’s second-quarter earnings call on Thursday.

“We have already secured differentiated business capabilities in digital asset infrastructure through the Korea Securities Depository’s tokenized securities platform project and through end-to-end validation […] of the full stablecoin process from issuance to settlement,” Lee said. He said he expects the relationship with Dunamu will help Samsung SDS expand in the digital asset infrastructure market.

The news came days after Samsung Electronics unveiled plans to add stablecoin support to Samsung Wallet, broadening the company’s digital asset push.

Strategic investment targets digital finance

In May 2026, Samsung Securities, Samsung SDS and Samsung Card agreed to buy a combined 4% stake in Dunamu, deepening Samsung affiliates’ ties to South Korea’s digital asset sector.

In the latest Q2 call, Lee reportedly said the company’s investment in Dunamu is a strategic move rather than a financial investment, adding that both companies plan to refine potential business models for digital financial infrastructure.

Source: Samsung SDS

“By combining Samsung SDS’s IT services, cloud, and security capabilities with Dunamu’s blockchain expertise, we aim to lead this market,” the Q2 transcript said.

Related: South Korea report proposes stablecoin rules before crypto law

Samsung SDS did not immediately respond to Cointelegraph’s request for comment, while Dunamu declined to comment.

AI growth supports broader expansion

The digital asset initiative comes amid Samsung SDS’ ongoing expansion in AI and cloud services, which helped lift Q2 revenue 5.9% year on year to 3.72 trillion Korean won ($2.6 billion), according to the quarterly earnings presentation.

Cloud revenue increased 17% from a year earlier, with external cloud business revenue jumping 75%, driven by demand for Samsung’s cloud platform and graphics processing unit-as-a-service offerings.

The company also reportedly outlined ambitious plans to expand its AI infrastructure from 110 megawatts today to 230 MW by 2029 and more than 800 MW by 2031, underscoring its broader push to build AI infrastructure alongside digital finance services.

Magazine: Fears of AI-driven DeFi hack epidemic overstated for now — but not for long

Situational Awareness, the hedge fund founded in 2024 by ex-OpenAI researcher Leopold Aschenbrenner, has approached investors and lenders for fresh capital after suffering heavy losses in the recent artificial intelligence stock sell-off, the Financial Times reported Thursday.

The fund, which the Wall Street Journal said had around $20 billion in assets under management as of June 8, has also offered some investors the option to buy portfolio assets, according to the FT, citing people briefed on the discussions and a July 24 investor letter. 

The size of the losses and amount sought were not disclosed. Aschenbrenner’s fund had gained 439% after fees through June, according to the letter, but the FT said borrowing increased the size of the fund’s bets, driving up losses when AI stocks collapsed during July’s market rout. Aschenbrenner also reportedly argued in the letter that the sell-off had created attractive investment opportunities.

Cointelegraph previously reported that the fund had made a big bet in its portfolio around the power and data centers supporting AI, including Bitcoin (BTC) miners pivoting into AI computing. A filing with the US Securities and Exchange Commission in March showed about $1.11 billion in positions across seven Bitcoin miner stocks, including IREN, Core Scientific, Riot Platforms and CleanSpark.

Aschenbrenner wrote a series of essays on artificial general intelligence in mid-2024 around the same time he launched his Situational Awareness fund. In it, he predicted that AGI machines will outpace college graduates by the end of the decade.

Cointelegraph contacted Situational Awareness for comment but had not received a response by publication.

Related: Bitcoin mining’s 2026 reckoning: AI pivots, margin pressure and a fight to survive

This article is produced in accordance with Cointelegraph’s Editorial Policy and is intended for informational purposes only. It does not constitute investment advice or recommendations. All investments and trades carry risk; readers are encouraged to conduct independent research.

Impermanent loss has long been one of the biggest challenges facing liquidity providers (LPs) in decentralized finance (DeFi). While automated market makers (AMMs) have revolutionized decentralized trading, they expose LPs to the risk of earning less than simply holding their assets whenever prices diverge significantly.

As artificial intelligence becomes increasingly integrated into DeFi protocols, many investors are asking an intriguing question:

Can AI finally eliminate impermanent loss?

The short answer is not entirely—but AI can dramatically reduce its impact. Let’s explore how.

---

Understanding Impermanent Loss

Impermanent loss occurs when the price ratio between two assets in a liquidity pool changes after you deposit them.

For example:

• You provide ETH and USDC to a liquidity pool.
• ETH doubles in price.
• Arbitrage traders rebalance the pool.
• You end up holding less ETH and more USDC than if you had simply held both assets.

Although trading fees can offset these losses, they aren’t always sufficient during periods of high volatility.

This is why many LPs hesitate to provide liquidity despite attractive yields.

---

Why Impermanent Loss Exists

Impermanent loss isn’t a bug—it’s a consequence of how AMMs maintain liquidity.

Traditional AMMs like constant-product pools automatically adjust token balances according to mathematical formulas.

These formulas:

• Keep markets liquid
• Allow permissionless trading
• Remove the need for order books

But they cannot predict future prices.

As a result, liquidity providers essentially sell appreciating assets and accumulate depreciating ones automatically.

---

Enter Artificial Intelligence

AI introduces something AMMs have never possessed:

Prediction.

Instead of relying solely on fixed mathematical curves, AI can analyze:

• Historical price behavior
• Market volatility
• On-chain liquidity movements
• Whale wallet activity
• Trading volume
• Cross-chain capital flows
• Social sentiment
• Macroeconomic events

This allows protocols to make smarter liquidity decisions.

---

AI Can Optimize Liquidity Placement

Concentrated liquidity protocols require LPs to choose price ranges.

Selecting the wrong range often leads to:

• Reduced fee generation
• Inactive liquidity
• Greater impermanent loss

AI can continuously monitor markets and recommend—or automatically adjust—the optimal liquidity ranges based on:

• Expected volatility
• Trend strength
• Volume concentration
• Support and resistance zones

Instead of manually repositioning liquidity, AI agents could perform these adjustments in real time.

---

Predictive Risk Management

Machine learning models excel at identifying patterns humans often miss.

Imagine an AI system detecting:

• A surge in exchange inflows
• Whale selling activity
• Rising options volatility
• Negative sentiment across crypto social platforms

The AI could recommend temporarily withdrawing liquidity before significant price swings occur.

After volatility subsides, liquidity could be redeployed.

This proactive strategy reduces exposure to major impermanent loss events.

---

Dynamic Portfolio Allocation

Rather than placing all assets into a single pool, AI can intelligently diversify liquidity across multiple pools.

For example:

• Stablecoin pools during uncertain markets
• ETH/BTC pools during lower volatility
• Emerging token pools when momentum increases
• Yield-generating vaults when volatility spikes

Capital continuously shifts where risk-adjusted returns are highest.

This resembles how institutional portfolio managers rebalance investments—only AI can do it every minute.

---

Adaptive Fee Strategies

Some modern AMMs feature dynamic trading fees.

Instead of fixed fees, AI can estimate:

• Expected volatility
• Arbitrage intensity
• Liquidity demand

The protocol can then automatically increase fees during turbulent periods.

Higher fees help compensate LPs for taking on greater risk.

This doesn’t eliminate impermanent loss, but it can significantly offset it.

---

AI-Powered Hedging

One of AI’s greatest strengths may lie outside the liquidity pool itself.

An intelligent system could automatically hedge LP positions using:

• Perpetual futures
• Options
• Synthetic assets
• Volatility products

For instance:

If AI predicts ETH is likely to experience extreme price movement, it could open a corresponding hedge that offsets potential impermanent loss.

Today, these strategies require sophisticated traders.

Tomorrow, AI agents could execute them autonomously.

---

Reinforcement Learning for AMMs

Researchers are exploring reinforcement learning, where AI continuously learns from market outcomes.

Instead of relying on static formulas, AI-powered AMMs could adapt their behavior based on:

• Trader activity
• Liquidity utilization
• Historical performance
• Market efficiency

Each market cycle provides new data, enabling the system to improve over time.

Eventually, liquidity allocation could become increasingly optimized with every transaction.

---

AI and Intent-Based DeFi

The next generation of DeFi may be driven by intent-based systems.

Instead of manually selecting pools, users simply specify their goals:

• Maximize yield
• Minimize impermanent loss
• Preserve capital
• Earn stable income

AI agents then determine:

• Which protocols to use
• When to move liquidity
• How to hedge positions
• When to rebalance

Liquidity management becomes autonomous rather than manual.

---

The Challenges

Despite its promise, AI cannot eliminate impermanent loss entirely.

Several obstacles remain:

Market Uncertainty

Even advanced AI cannot predict black swan events with certainty.

Unexpected news, protocol exploits, or geopolitical developments can quickly invalidate predictions.

Data Quality

AI is only as effective as the data it receives.

Incomplete or manipulated on-chain data can lead to poor decisions.

Execution Costs

Frequent rebalancing introduces:

• Gas fees
• Slippage
• MEV exposure
• Operational complexity

Sometimes the cost of optimization outweighs the benefits.

Smart Contract Risk

AI strategies still depend on secure smart contracts.

If the underlying protocol is compromised, optimization becomes irrelevant.

---

The Future: AI as a Liquidity Manager

Rather than replacing AMMs, AI is likely to become their intelligent layer.

Future liquidity providers may no longer choose pools manually.

Instead, autonomous AI agents will:

• Monitor markets 24/7
• Rebalance liquidity automatically
• Hedge risky positions
• Optimize fee generation
• Reduce capital inefficiencies
• Continuously learn from market behavior

Providing liquidity could eventually resemble hiring an AI portfolio manager.

---

Conclusion

AI is unlikely to eliminate impermanent loss because the phenomenon is rooted in the mechanics of automated market makers and the unpredictability of financial markets. However, it has the potential to substantially reduce its impact through predictive analytics, dynamic liquidity allocation, automated hedging, adaptive fee optimization, and continuous portfolio rebalancing.

As AI agents become more sophisticated and intent-based DeFi matures, liquidity provision could shift from a passive activity to an actively managed, intelligent strategy. The future may not be one where impermanent loss disappears—but one where it becomes far more manageable, allowing liquidity providers to earn more efficiently while taking on less unnecessary risk.

REQUEST AN ARTICLE

This month, analysts have repeatedly pointed to multiday inflows into the U.S.-listed crypto exchange-traded funds as evidence of the return of institutional demand. Zoom out, though, and the institutional story still looks bleak.

Bitcoin spot ETFs have pulled in just $205 million in net inflows in July, the lowest monthly total on record, according to SoSoValue data. While there are still two trading days left, the figure marks anemic recovery from the heavy red ink of prior months, which saw $2.43 billion exit in May and $4.52 billion in June.

Ether has fared somewhat better. ETH ETFs have attracted $342.85 million in July, almost as much as in April and outperforming bitcoin and other crypto funds. XRP is on track for a fourth consecutive month of inflows, though the sum remains a paltry $13.61 million. Solana ETFs sit at $13.82 million.

Together, the numbers paint a picture of limited institutional appetite at best. Ether’s stronger haul is consistent with its price performance against bitcoin. The Binance-listed ether-bitcoin pair has surged by 11% this month.

The basic multiple-to-Net Asset Value (mNAV) that investors are willing to pay for Michael Saylor’s Strategy has declined by two-thirds over the past two years from 2x to 0.68x.

Exactly 24 months ago, Strategy common stock was worth exactly twice the value of its BTC. For every $1 of BTC the company owned, MSTR traded at $2.

In 2024, there was optimism about Strategy’s ability to positively accrete BTC for shareholders through successful business operations.

As of today, that measure of optimism has declined by exactly two-thirds.

Saylor’s company now holds 843,775 BTC at an average cost of $75,476, uncomfortably higher than the current market below $65,000.

For most of the past two years, its shares commanded a premium to that pile. As that original premium flipped to a discount, the company redefined the term mNAV twice in an attempt to keep it above 1x.

Its latest iteration is holding on by a thread. It read 1.03x yesterday.

New and updated market metrics are live at https://t.co/yIv7IimRdf. As Digital Credit becomes a larger portion of our balance sheet, we’ve sharpened our precision based on investor feedback. This video walks through what’s new and why.
00:00 – Intro to Strategy’s new and updated… pic.twitter.com/ndCoDc9PDW

— Strategy (@Strategy) July 23, 2026

Three definitions for mNAV

Basic mNAV is arithmetic a teenager can do. Divide the company’s market capitalization by the dollar value of its BTC. Above 1x means that investors are paying more for the stock than for the BTC. Below it, less.

BitcoinTreasuries.net, which still runs that original calculation, calculates Strategy’s basic mNAV at 0.64x.

Best to dust that number under the rug.

Strategy’s first attempt to redefine mNAV became enterprise value mNAV. This folded in the value of debt and preferred stock into the numerator alongside the market cap of MSTR, the common stock.

Conveniently, loading liabilities onto the top of a fraction made the result larger. The enterprise value-adjusted mNAV therefore sat above 1x long after the basic one had sunk below 1x, and only crossed below 1x in late June 2026.

Predictably, a new-and-improved version of mNAV arrived in July.

Strategy’s dashboard now begins its calculation from a reserve of roughly $57.7 billion in BTC plus dollars, subtracts $6.8 billion of debt and $15.4 billion of preferred stock, and calls the remaining $35.5 billion a net reserve.

It then divides that so-called net reserve by the share count, compares it against a share price of $95.32, and the multiple lands at 1.04x. Voilà.

Read more: How Michael Saylor replaced ‘bitcoin’ with ‘credit’

Advertisement

Redefinitions didn’t stop MSTR from crashing

Saylor pitched the overhaul on the grounds that “Bitcoin capital markets require a new financial language.”

Strategy CEO Phong Le credited investor feedback for an “upgraded” metric that “establishes a 1x threshold for accretive MSTR issuance.”

The redefinition also orphans the company’s entire back catalogue. Strategy’s glossary advises, “Prior to July 23, 2026, the company’s use of the term mNAV referred to a different metric so references to the company’s mNAV calculated prior to that date are not comparable to the company’s mNAV calculated after that date.”

Every mNAV Saylor has ever tweeted prior to this month is, by the company’s own account, incomparable to the one on the company’s own website today.

Advertisement

The same page concedes that despite the label, “it is not equivalent to ‘net asset value’ or ‘NAV’ or any similar metric in the traditional financial context,” and should be used “only by sophisticated investors who understand its limited purpose and many limitations.”

Saylor himself needed seven minutes to define the old mNAV on stage at BTC Prague in June.

Strategy briefly promised shareholders in July 2025 that it wouldn’t dilute MSTR shareholders below 2.5x mNAV, then sold $14.3 billion of stock beneath that line anyway.

On Wednesday, it was 0.68x, and the metric that produced both figures no longer officially exists.

Advertisement

The price of MSTR — the thing that actually matters to shareholders and cannot be redefined — has declined 38% year to date and 76% over the past 12 months.

Got a tip? Send us an email securely via Protos Leaks. For more informed news and investigations, follow us on X, Bluesky, and Google News, or subscribe to our YouTube channel.

The crypto market is showing resilience on Thursday, with bitcoin BTC$64,452.50 trading little changed at $63,915 and ether dropping just 0.25% since midnight UTC after what might be seen as a difficult session for global risk assets on Wednesday.

The calm, however, masks erratic back-and-forth price swings around the Federal Reserve’s interest rate meeting that flushed out leveraged futures bets, triggering heavy liquidations.

While the rate remained unchanged, three committee members voted for an increase. Higher rates reduce the attractiveness of risky assets.

About $286 million in positions were liquidated in 24 hours, according to CoinGlass. Longs accounted for $186 million and shorts $100 million, a balance that signals a market that moved hard in both directions and settled back where it started.

Hours after the FOMC decision, Iran launched multiple ballistic missiles at U.S. troops, prompting President Donald Trump to vow to hit Iran “hard” in response.

Oil surged, erasing Monday’s declines, and U.S. equities fell. Still, S&P 500 and Nasdaq index futures are now slightly positive. Microsoft (MSFT) and Meta (META) earnings after the bell today could swing sentiment once again.

Derivatives positioning

• Crypto futures long/short ratio flips slightly bearish: The crypto futures long/short taker volume ratio has flipped slightly bearish, with shorts at 51%, according to Coinglass data. The market appears to be in stasis with both open interest (OI) and trading volume mostly unchanged from yesterday.
• Uniswap’s UNI among top performers: Uniswap’s UNI token is one of the best performers among the top 100 coins. Open interest in its futures, however, has pulled back to 65.80 million tokens, suggesting an unwind of positions. The 24-hour cumulative volume delta (CVD) isn’t encouraging either. The negative print points to leadership among those shorting futures at market orders rather than passive limit orders.
• Bitcoin open interest holds steady: Bitcoin’s OI remains around 750K BTC, as it has since early June. In other words, participation in the coin’s price bounce remains limited. Ether’s OI, too, has pulled back to under 14 million tokens from a six-week high of 14.53 million ETH. This again shows limited demand for leverage despite ether outperforming bitcoin this month.
• Bear dominance in the altcoin market: Most of the 25 largest coins, except BTC, ADA, TRX, CRO, ZEC, have negative 24-hour OI adjusted CVDs. It seems the altcoin market price action is still dominated by bears.
• Bitcoin’s BVIV drops below 38%: Bitcoin’s BVIV, the 30-day implied volatility index, has fallen back below 38%, nearing levels that have historically served as floors. This calls for caution, as volatility is mean-reverting and can snap back higher from the floor.
• BTC and ETH calls in demand: Options listed on Deribit show BTC calls at strikes $70,000 and $75,000 leading the 24-hour volume rankings. Calls offer an asymmetric upside exposure in the underlying asset (like a lottery ticket). The same is true for ether: The top five most traded bets are all call options.

Token talk

• Injective INJ$4.7966 was the 24-hour standout on Thursday, rising 6.95% as DeFi tokens outperformed. Uniswap (UNI) added 4.46% and FET recovered 3.28% after a bruising week.
• Zcash (ZEC) extended its recent run, adding 1.54% since midnight UTC to $474, continuing to outpace privacy coin peers. Monero (XMR) gained 0.4%.
• Hyperliquid (HYPE) slipped 0.47% to $53.64, extending a retreat from last month’s highs that has now unwound roughly 30% from its peak.
• Lighter (LIT) gave back 4.22% over 24 hours, though it is up 0.50% since midnight, suggesting sellers may be running out of steam after a correction that has erased nearly 25% from its July peak.
• Jupiter (JUP) fell 1.48% since midnight after a brief recovery on Wednesday, with daily trading volume continuing to dwindle to $23 million despite regularly topping $50 million earlier this year.

Robinhood reported record second-quarter results, highlighting strong growth in transaction-based revenue overall while acknowledging a meaningful pullback in cryptocurrency-specific earnings. In its latest earnings report, the online brokerage said crypto transaction revenue fell to $100 million, down from roughly $160 million a year earlier.

Despite the decline in crypto-related income, the company delivered an overall quarter marked by rising profitability and expanding platform activity. Total revenue climbed 32% year-over-year to $1.31 billion, and net income increased 48% to $573 million, according to the earnings release.

Key takeaways

• Crypto transaction revenue fell 38% to $100 million, even as Robinhood’s broader transaction-based revenue rose.
• Quarterly revenue and earnings reached records: revenue rose to $1.31 billion and net income grew to $573 million.
• Crypto notional trading volume totaled $40 billion: $18 billion on the Robinhood app and $22 billion via Bitstamp.
• Robinhood is scaling beyond trading: it launched parts of Robinhood Chain and introduced tokenized U.S. stocks and a decentralized lending product.
• Management reiterated a tighter cost outlook for 2026: it narrowed adjusted operating expenses and share-based compensation guidance.

Crypto revenue dips as overall transaction business grows

Robinhood’s results show a split between the performance of its crypto segment and the rest of its transaction engine. While cryptocurrency was the only major transaction category to decline during the quarter, the company pointed to strength in other areas that more than offset the weakness.

Crypto transaction revenue decreased to $100 million from about $160 million a year earlier, the company said. At the same time, transaction-based revenue rose 44% to $776 million, supported by higher contributions from categories outside digital assets.

The company also reported $40 billion in crypto notional trading volume for the quarter. Of this total, $18 billion came from the Robinhood app, down 35% year-over-year, while $22 billion came from Bitstamp. Robinhood acquired Bitstamp in June 2025, as referenced in earlier coverage here.

Market pricing reflected the mixed nature of the quarter ahead of the report: shares were down about 3.15% on Wednesday before the earnings release, based on Yahoo Finance data.

Robinhood’s crypto playbook shifts toward infrastructure and new products

Even with lower crypto transaction revenue, Robinhood used the quarter to push forward with its wider digital asset strategy. The company said it completed its acquisition of WonderFi, a Canadian crypto platform, continuing its efforts to broaden what it offers beyond pure trading.

After the quarter ended, Robinhood also rolled out additional ecosystem components. It unveiled the public mainnet of Robinhood Chain after previously reporting activity around bridged assets. The company further introduced tokenized U.S. stocks to eligible users in more than 120 countries and debuted its first decentralized lending product, Robinhood Earn.

On the Ethereum layer-2 side, DefiLlama data shows Robinhood’s new network posted $348 million in total value locked as of Thursday. The same dataset referenced stablecoins exceeding $500 million and more than $1 billion in bridged assets, illustrating that the platform’s activity is not limited to trading fees.

For investors, this matters because it reframes what “crypto performance” can mean for a brokerage. Transaction revenue can soften when market activity slows or user behavior shifts, but an expanding chain ecosystem—particularly one involving bridges and lending—can create alternative revenue pathways over time.

Non-crypto categories and platform metrics keep momentum

Robinhood’s earnings report emphasized that its broader product suite absorbed the crypto slowdown. The company said growth in event contracts, options, and equities more than offset the weakness in digital assets.

It reported that event contract revenue surged more than tenfold to $156 million, options revenue rose 29% to $342 million, and equities revenue jumped 95% to $129 million.

At the platform level, Robinhood posted record net deposits of $21.7 billion during the quarter. Total platform assets rose 32% year-over-year to $369 billion, and funded customers grew 7% to 28.4 million.

These metrics suggest that the company’s customer and balance-sheet expansion continued regardless of the crypto segment’s year-over-year revenue decline. For traders and users, that combination indicates a continued push to keep engagement broad—spreading attention across multiple asset classes and contract types rather than relying primarily on crypto transaction activity.

Guidance narrows as adjusted EBITDA grows

Robinhood also addressed expenses and profitability guidance. The company lowered and narrowed its 2026 outlook for adjusted operating expenses and share-based compensation to between $2.675 billion and $2.775 billion, down from a previously provided range of $2.7 billion to $2.825 billion.

On profitability, adjusted EBITDA rose 35% to $741 million. Total operating expenses increased 33% to $734 million, reflecting continued investment while still targeting more controlled growth at the operating level.

When viewed alongside the company’s digital asset expansion, the tighter expense guidance suggests management is trying to balance growth in new areas—like tokenization and decentralized lending—while keeping cost discipline in focus.

Going forward, investors will likely watch whether Robinhood Chain’s early traction translates into sustained engagement and monetization, and whether crypto transaction revenue stabilizes as broader platform growth continues to diversify away from purely crypto-dependent earnings.