Tag: Business – Decrypt

🟠 Strategy Raises $467M and Skips Bitcoin for a Third Straight Week

Strategy’s Bitcoin-buying machine stayed in neutral last week. The company raised $467M by issuing common stock and put all of it toward cash, lifting its USD Reserve to a record $3B. That means no Bitcoin purchase for the third week running.

The raise added roughly 18% to Strategy’s cash reserves in one move, giving the company more than 20 months of coverage on its $1.76B in annual dividend and interest obligations. So the entirety of the week’s capital markets activity went toward fortifying the balance sheet’s cash cushion. Since its last Bitcoin purchase on June 22, Strategy has generated about $215M from selling Bitcoin, less than half of what this single stock issuance brought in. So the stock sales are now doing more work than the Bitcoin sales.

Once again, Saylor is issuing common shares to fund dividend payments on his preferred stock. MSTR holders are being diluted to pay STRC holders. MSTR fell 4% Monday to around $90.80, down 18% on the month, though it has steadied since hitting a 28-month low of $81.81 in late June. STRC sits at $87.04, still below its $100 par value where it has lingered since mid-May while paying a 12% dividend. And with Bitcoin at $62,600 against an average cost of $75,476, the 843,775-coin stack is roughly $11B underwater.

The biggest open question right now is—why isn’t he buying Bitcoin at these prices? Raising cash to alleviate market concerns and to fund future debt payments made sense when he was mostly depleted, but now Saylor is flush. He’s got 20 months’ runway, and more importantly, he’s proven he can just dump MSTR shares whenever to raise cash. So why not buy BTC here 50% off ATH? Certainly any BTC buy in the low 60s makes more sense than buys in the 80s, 90s, or over 100k. So why aren’t they coming? And when will the next buys hit the tape? Hopefully we find out soon. Or Saylor will have some explaining to do…

🏛️ Democratic Opposition to the CLARITY Act Grows With Four Weeks Left

Senate Democrats are hardening against the CLARITY Act, and Trump’s $1.2 billion crypto fortune is (unsurprisingly) the wedge.

Elizabeth Warren wrote to Senate leadership on Monday, demanding the bill bar the president, vice president, senior officials, members of Congress, and their families from profiting off the crypto industry. Anything less would be a giveaway to the president and his family at the public’s expense, in her words. On Tuesday, more Senate Democrats including Chris Murphy and Chris Van Hollen are expected to hold a press conference against it, hitting both Trump’s crypto dealings and the argument that the bill weakens financial oversight built after the Great Depression.

The vote math is the tricky part here. CLARITY needs 60 votes, which means at least seven Democrats have to cross over, and possibly more given the Republican bench is thinner than expected (Mitch McConnell remains hospitalized, and Lindsey Graham died suddenly over the weekend). Trump argued Monday that the Senate should pass CLARITY in Graham’s honor, calling him a big supporter of crypto legislation. Graham was never involved in the negotiations, rarely spoke on the topic, and was the only Senate Republican to co-sponsor a 2023 anti-money-laundering bill that crypto groups called deeply hostile to the industry.

So where does this leave us? Less than four weeks remain before the August recess, and everyone involved agrees that missing it means the bill dies in the noise of the November midterms. The industry’s own forecasters have already marked it down, with Galaxy cutting its odds to 50% and prediction markets pricing it in the low 40s. The core problem is that the harder Democrats push for ethics language, the more they force Republicans to choose between passing the bill and protecting the president’s business interests. That’s a bad trade to have to make with a hard deadline bearing down.

🌎 Macro Crypto and Markets

• Crypto majors are mixed ahead of CPI; BTC even at $62.7k; ETH +1% at $1,795; SOL -1% at $75; HYPE -2% at $64
• HASH (+11%), INJ (+5%) and ENA (+5%) led top movers
• Oil +10% at $81; Gold -1% at $4,020
• Stock futures are mixed as oil spikes and June CPI data comes this AM; DOW -0.7%, Nasdaq +0.3%
• June CEX volumes fell 5% from May, while derivatives trading rose by 4%
• BTC and ETH social media posts have fallen to their lowest level since 2020
• Coinbase CEO Brian Armstrong admitted Base “messed up” on content coins, saying they didn’t work and that Base pivoted away early this year, with resources now going toward trading, payments, and AI agents
• BlackRock, Goldman Sachs, and JPMorgan joined a 54-firm UK tokenization taskforce backed by the government, starting with tokenized repo, with a Treasury report projecting up to £33 billion in annual economic output by 2035
• TeraWulf’s CEO said “not all megawatts are created equally” in the AI race, stating flatly that “we’re not involved in Bitcoin” anymore as its $19 billion Anthropic hosting deal completes its shift from miner to AI infrastructure company
• SBI Holdings’ blockchain initiative pivoted to Solana for tokenization and stablecoin issuance, moving off R3’s Corda and bringing the Solana Foundation into its joint venture, with plans including payment infrastructure for AI agents

Corporate Treasuries & ETFs

Meme Coin Tracker

• Meme leaders were mostly red down 1-2%; DOGE even, SHIB -2%, PEPE -1%, PENGU -4%, TRUMP -1%, BONK -2%
• Robinhood chain memes were led by Cashcat +8% to $180M, WOOD +70% to $15M, Wishbone +45% to $9M
• Febu (+60%), three (+100%) and Cards +20% were notable Solana movers

📈 Myriad Market of the Day

💰 Token, Airdrop & Protocol Tracker

🚚 What is happening in NFTs?

• NFT leaders were mixed; Punks even at 32.4 ETH, BAYC +1% at 8.94 ETH, Pudgy -1% at 4.33 ETH; Hypurr’s -4% at 179 HYPE
• PXL NET (+60%) and Squiggles (+10%) led top movers
• New Robinhood NFT sets jumped including RH Miners (+1350%) and 8skullz (+340%)

OpenAI published a new prompting guide for GPT-5.6 Sol, its newly released flagship model, and the main message will feel wrong to anyone who spent the last year writing multi-page system prompts: stop writing so much. The core idea is outcome-first prompting. Define what good looks like, set the stopping conditions, and get out of the way.

Detailed how-to instructions, repeated style rules, examples that don’t change behavior—all of it is now considered noise.

OpenAI backs this with numbers: In internal coding agent tests, leaner system prompts improved evaluation scores by roughly 10–15% while cutting total tokens by 41–66% and costs by 33–67%.

GPT-5 vs. GPT-5.6: What actually changed

The GPT-5 prompting guide, published at launch in August 2025, was about adding scaffolding. You got XML persistence blocks telling the model to keep working until the problem was solved, detailed context-gathering templates that mapped exactly how to parallelize searches and when to escalate, and tool preamble scripts that narrated every step out loud.

The philosophy was calibrating eagerness—building explicit rails for when to go harder or stand down.

GPT-5.6 mostly doesn’t need those rails. The new guide tells you to trim: repeated rules, style instructions that don’t change behavior, examples that do nothing, and process steps the model already handles reliably. So basically, that “ block with its parallel search batches and early-stop criteria that used to help is now scaffolding the model has to parse around, not scaffolding that helps it.

What you actually keep is simpler: the user-visible outcome, success criteria, stopping conditions, and hard constraints. The guide’s model of a good prompt starts with “Resolve the customer’s issue end to end”—then specifies exactly what done looks like, what actions to complete before responding, and what to do when required evidence is missing. Not “be thorough.” Not “keep going.” Just: here is the destination.

The risk calculus also shifted. The guide warns that GPT-5.6 follows prompt contracts closely, and that “conflicting rules can create more instability than missing detail.”

An earlier model would pick one instruction when it hit a conflict. GPT-5.6 burns reasoning tokens trying to reconcile both, which is slower, more expensive, and often wrong. If your system prompt has overlapping rules—and most production prompts do—this is the thing to fix first.

Also OpenAI heavily advises against using the old trick of resorting to absolutes like “always do this” or “never do that” to steer the AI’s behavior in a specific direction.

Two concrete additions round out the difference. The first is the text.verbosity parameter: Because GPT-5.6 is already more concise by default than GPT-5.5, old “be brief” instructions now over-correct and make responses too short. Set a global default via the parameter, then override per task in the prompt. The second is a section on Programmatic Tool Calling—for bounded workflows where code handles filtering, batching, or aggregating large intermediate outputs and returns a compact result, offloading that work from the model’s judgment entirely.

But does it work?

We used the guide to optimize our prompt for TYPE OR DIE, the first-person typing survival horror game we build to benchmark a model’s coding abilities. The result was more polished: GPT-5.6 Sol tackled the auto-aim logic more efficiently than on previous runs, the visuals had more coherence, and the overall feel of the game was cleaner.

It took more time to build. The model didn’t jump straight to code—it mapped the entire problem first, planned each system before writing a line. That’s the guide working as intended. Define the destination; the model chooses the route.

The new prompt is available on our Github so you can check it out.

You can play the original GPT 5.6 game by clicking on this link.

The game created under the newer prompt, is available here.

If you want to push further, or are too lazy to memorize all these new guidelines, you can build your own custom GPT and feed it the full guide as its knowledge base. Configure it to analyze any prompt you throw at it, understand the underlying logic, and rewrite it in GPT-5.6 style. You end up using prompt engineering to engineer better prompts.

Promptception. You’re welcome.

New Hampshire Governor Kelly Ayotte helped make the state into one of the crypto-friendliest in the nation when she signed HB 639 into law last week. 

Known as the The Blockchain Basic Laws act, the bill provides protections for cryptocurrency innovation and use in the state while also allowing for the creation of a special blockchain dispute docket in the superior court. 

“With Governor Ayotte’s signature on HB 639, New Hampshire has once again demonstrated that it intends to lead the nation in blockchain innovation,” said New Hampshire Representative Keith Ammon, the bill’s primary sponsor, in a statement. 

“The Blockchain Basic Laws protect one of the most fundamental rights in the digital economy—the right of individuals to control their own digital assets through self-custody,” he added. “They also provide clear legal protections for blockchain developers, miners, validators, entrepreneurs, and businesses building the next generation of financial technology.”

The state’s latest blockchain legislation follows its passing of a strategic Bitcoin reserve last year. The bill, signed by Ayotte in May 2025, allows the state’s treasurer to invest up to 5% of its public funds in the leading crypto asset, alongside precious metals like gold and silver. 

Ammon, who played a key role in that bill’s passage, told Decrypt at the time it was “one little way our state could hedge against inflation in the future.”

“Today, with the signing of HB 639, we have taken another major step by enacting one of the most comprehensive blockchain rights laws in the country,” he said in a statement. “Entrepreneurs, investors, developers, and innovators across America should know that New Hampshire is open for blockchain business.”

Despite its advances, the state’s executive council last week blocked a proposal that would have allowed the New Hampshire Business Finance Authority to facilitate a Bitcoin-backed municipal bond.

“The Odyssey” director Christopher Nolan famously doesn’t use a smartphone, so it’s perhaps unsurprising that he isn’t on board with the latest tech buzzphrase.

On the press tour for his latest effort, Nolan told The Telegraph that young audiences are “utterly rejecting” generative AI, adding that, “I’ve never seen a more rapid wholesale dismissal of a supposedly foundational jump in technology in my lifetime.”

He pointed to the reaction of his four children, in their late teens and early 20s. “Their judgment of AI slop has been immediate and harsh,” Nolan said. “They see it for what it is very quickly—and it’s much easier for them to identify it—because it grew out of an online world they know really well.”

The technology has hit at “exactly the wrong time” for filmmaking, he argued, pointing to a “renewed interest in more tactile, more real forms of storytelling” after a glut of films featuring virtual environments.

Nolan’s own blockbusters have become famous for their spectacular in-camera effects work, whether crashing a 747 into a building for “Tenet,” landing a Spitfire on a beach for “Dunkirk” or planting entire crops of corn for a chase scene in “Interstellar.”

That said, he’s not averse to using computer-generated VFX himself, like Two-Face’s scarred visage in “The Dark Knight,” and he conceded that not every aspect of generative AI is necessarily “useless or meaningless.”

Generative AI and filmmaking

Generative AI has caused a rift in the film industry, with some creatives openly coming out against the technology and others embracing it. Among the former camp are “Pan’s Labyrinth” director Guillermo del Toro, who has led chants of “fuck AI” on stage, and Steven Spielberg, who has dismissed the technology as an “empty chair with a laptop on it.”

On the other side of the fence are the likes of Martin Scorsese, who has joined AI firm Black Forest Labs as an adviser, and “Terminator 2” director James Cameron, who sits on the board of Stability AI. Ben Affleck has become a convert to the technology, having sold his AI startup InterPositive to Netflix after previously doubting that the technology would be able to “write anything meaningful” or create films “from whole cloth.”

While filmmakers debate among themselves, AI companies are pressing ahead with video creation tools like Utopia’s PAI, which aims to maintain consistency of output across cuts and scenes.

Publicly traded Ethereum treasury firm BitMine Immersion Technologies added another $49 million worth of ETH to its stash last week, acquiring 27,801 ETH. 

The firm now holds 5,770,038 ETH, nearly 4.8% of the token’s circulating supply, valued around $10.1 billion as Ethereum trades around $1,780 on Monday. 

BitMine’s latest purchase comes amid a strong week for the underlying Ethereum network, according to its chairman Tom Lee, who pointed to the strong public debut for Robinhood’s Ethereum layer-2 network, Robinhood Chain. (Disclaimer: Tom Lee is an investor in Dastan, the parent company of Decrypt). 

“One of the biggest crypto success stories in 2026 is the breakaway success of the Robinhood Chain L2 mainnet on July 1, built on Arbitrum,” said Lee in a statement.

“Already, dollar volumes have exceeded $1 billion, and Robinhood Chain now has more trading volume than any other decentralized exchange (DEX), demonstrating the outstanding utility and product market fit for Ethereum, which is the underlying chain,” he added. 

DEX volumes on the chain in the last week have surpassed $3 billion according to data from DeFi Llama, but it still trails more established networks like Ethereum and Solana, which have registered $7.27 billion and $12.34 billion worth of DEX volume over the same time period, respectively. 

Robinhood Chain, which finalizes on Ethereum and uses ETH as its native gas token, has blossomed in the early going, creating small fortunes for some early meme coin traders on the blockchain—like one individual who turned $85 into more than $2 million in paper gains.

“Robinhood’s 27 million users are paying crypto fees denominated in ETH,” said Lee. “In other words, everyday users are starting to see ETH as money,” he added. 

While the mobile brokerage may boast millions of users, data gathered by Token Terminal points to just 788,000 active addresses on Robinhood Chain thus far. 

Despite the activity, ETH has dropped around 2% in the last 24 hours and has only gained around 1.3% in the last week of trading. That marks significant outperformance of shares in BitMine (BMNR), though, which have dipped 5.7% in the last five trading days. 

BitMine shares are trading around $14.65 shortly after the opening bell on Monday, down more than 2.2% since trading began. 

Strategy’s Bitcoin-buying machine remained in neutral last week as the firm continued growing its cash reserves, forgoing acquisitions of the digital asset for a third straight week.

The company raised $467 million during the period by issuing common stock, lifting the balance of its so-called USD Reserve to $3 billion, according to an announcement.

Shares of Strategy were down 4% following the opening bell, changing hands around $90.80, according to Yahoo Finance. Although the firm’s stock price has tumbled 18% over the past month, it has steadied since hitting a 28-month low of $81.81 in late June.

Strategy’s flagship preferred stock, Stretch (STRC), had edged down to $87.04, after approaching its highest point in nearly a week in pre-market trading. Since mid-May, the product that currently offers a 12% annual dividend has lingered below its $100 par value, while notching record lows.

The company’s latest move underscored its commitment to ensuring that it can fulfill preferred stock dividend payments and debt interest obligations, padding its cash cushion to record levels following the adoption of a capital management framework weeks ago.

In a note shared on Monday, Benchmark-StoneX Managing Director and Senior Research Analyst Mark Palmer shared that Strategy added around 18% to its cash reserves in a single move, providing the firm with more than 20 months’ worth of coverage for its annual dividend and interest obligations of $1.76 billion.

“The entirety of the company’s capital markets activity during the week was channeled toward fortifying the balance sheet’s cash cushion,” he added.

The framework marked a significant shift for Strategy, formalizing conditions under which the world’s largest corporate holder of Bitcoin could sell the digital asset. On Monday, the company’s stockpile of 843,775 Bitcoin was valued around $53 billion.

Since Strategy reported its last Bitcoin purchase on July 22, the company has generated around $215 million in proceeds from selling the digital asset. Those funds were earmarked for dividends and debt, mirroring the intent of its latest fundraising efforts.

Before Strategy formalized its new approach, some analysts voiced concerns that the company’s USD Reserve had worn too thin, intensifying scrutiny on the sustainability of “ballooning” costs tied to products such as Stretch that receive routine payouts.

“Orange dots tell only part of the story,” Strategy co-founder and Executive Chairman Michael Saylor said in an X post on Sunday, hinting at the company’s shifting scope alongside a chart of the Bitcoin-buying firm’s recent purchases.

The company’s willingness to tap Bitcoin as a source of liquidity for its cash reserves represented a reversal of Saylor’s buy-and-never-sell mantra, but some analysts say the shift toward “two-way capital allocation” is ultimately in the company’s best interest.

On Monday, Bitcoin had fallen 2.3% over the past 24 hours to $62,600, according to CoinGecko. With an average purchase price of $75,476 per Bitcoin, that meant Strategy’s stockpile remained roughly $11 billion underwater.

Thailand’s central bank has turned data-analytics tools on the country’s stablecoin market, scanning high-volume trades for signs of illicit money as part of a widening crackdown on the shadow economy.

Bank of Thailand Governor Vitai Ratanakorn said the central bank has begun screening abnormally large stablecoin transactions, particularly in Tether’s USDT, and has already flagged some that appear designed to sidestep disclosure requirements or move funds outside normal banking channels, according to Thai outlet Thansettakij.

Because the Securities and Exchange Commission, rather than the central bank, directly regulates digital assets in Thailand, the Bank of Thailand is coordinating with the SEC, which holds the authority to act on the findings. USDT is the largest stablecoin and the most widely used trading pair on crypto exchanges, according to CoinGecko data.

A wider net

The stablecoin screening is one strand of a broad campaign against what Thai officials call the “grey economy,” an effort the governor cast as a long haul requiring several measures running in parallel rather than any quick fix. Since April, banks have had to check the purpose of cash withdrawals of 5 million baht (about $150,000) or more, a rule the central bank says cut large cash withdrawals by roughly 35%. From the fourth quarter, depositors bringing in 5 million baht or more in cash may have to declare where it came from.

Regulators are also tightening controls on high-value banknote exchanges and gold trading, where officials noticed buyers ordering gold through an app in the morning and collecting it from shops in the afternoon. Suspicious activity is now reported to Thailand’s Anti-Money Laundering Office, and monthly gold withdrawals have fallen from about 4,000 kilograms to around 700. Banks have separately closed thousands of “mule” accounts linked to online gambling.

Crypto in Thailand

Thailand has become a hotspot for crypto-enabled crime, and its agencies have been chasing it aggressively. Thai police recently traced a romance-scam laundering network in which a single wallet moved more than $122.5 million in 10 months through cross-chain swaps, as part of Interpol’s Operation First Light. In recent months, investigators have also widened a mining probe into a $300 million Chinese laundering network and seized $8.6 million in illegal mining rigs powering scam compounds.

At the same time, the country is courting legitimate crypto: the SEC’s three-year plan pushes tokenization and crypto ETFs, and the central bank says it is working to develop a baht-backed stablecoin as part of a broader financial-infrastructure overhaul.

Apple has sued OpenAI and two former employees, accusing the ChatGPT maker of using stolen trade secrets for its consumer hardware efforts.

The complaint, filed Friday in the U.S. District Court for the Northern District of California, names former Apple senior system electrical engineer Chang Liu and former iPhone and Apple Watch design executive Tang Yew Tan, along with OpenAI Foundation, OpenAI Group PBC, and io Products.

Apple alleges Liu, who left the company in January after eight years, failed to return a company laptop and later accessed Apple’s internal systems through an authentication bug.

“While employed by OpenAI, Mr. Liu also exploited a rare, previously unknown authentication bug to access Apple’s shared network folders,” Apple’s attorneys said in the complaint. “Upon discovering that he had this unauthorized access to Apple’s systems, Mr. Liu did not report it, return his stolen Apple-issued work laptop, or delete the program that allowed the access.”

Apple alleges Liu downloaded dozens of confidential hardware files, including information about unreleased products, engineering presentations, technical specifications, and proprietary project data.

The company also alleges Tan, who spent 24 years at Apple before becoming OpenAI’s chief hardware officer, used confidential information from his time at Apple to benefit OpenAI.

The complaint claims Tan used Apple’s internal project names during OpenAI interviews and asked about unreleased products. Apple also alleges candidates were told to bring “actual parts,” for “show and tell.”

Apple further claims OpenAI’s recruiting process requested “CAD/design artifacts,” prototypes, supplier information, and details about employees’ work on Apple hardware.

Apple and OpenAI did not immediately respond to a request for comment by Decrypt.

The lawsuit follows OpenAI’s $6.4 billion acquisition of io Products, the hardware startup founded by former Apple designer Jony Ive. Ive is not named in the complaint.

According to the filing, OpenAI’s hardware division has hired more than 400 former Apple employees. Apple claims it contacted OpenAI in February with concerns about confidential information entering the company but did not receive a response.

The news comes after a separate trade secret dispute between OpenAI and Elon Musk’s xAI. In September, xAI sued OpenAI, alleging the ChatGPT maker recruited former employees to obtain confidential source code, training methods, and data center strategies.

OpenAI denied the allegations, and a federal judge dismissed the lawsuit in June, finding xAI failed to show OpenAI encouraged a former employee to disclose confidential information.

The lawsuit is a stark pivot from Apple and OpenAI’s earlier relationship.

In 2024, Apple tapped OpenAI to bring ChatGPT to Siri as part of its Apple Intelligence initiative. However, earlier this year, Apple turned to Google’s Gemini to power its next generation of AI models after delays stalled the rollout.