Senior Republican officials reportedly contacted Commerce Secretary Howard Lutnick after a crypto-linked super PAC signaled plans to spend $1.75 million in Texas. 

The planned spending would have supported Texas Attorney General Ken Paxton in a Republican Senate runoff against Sen. John Cornyn, according to Axios.

The filing drew attention because President Donald Trump had not taken a side in the race. Republican leaders viewed the planned move by Fellowship PAC as a possible disruption in a sensitive primary contest.

Fellowship PAC was seeded by Cantor Fitzgerald, the firm Lutnick led before joining the Trump administration. Lutnick divested his interests last year, and his sons now run the firm, Axios reported.

Planned ad buy did not move forward

Axios reported that Fellowship PAC did not place the ad buy listed in the Federal Election Commission filing. Republican leaders were later told the group had not aired and was not preparing to air pro-Paxton ads.

The report also said media-tracking data showed neither Fellowship PAC nor its ad firm had run political ads this cycle. It remains unclear whether Lutnick acted after the calls from Republican officials.

The National Republican Senatorial Committee criticized the reported filing after it surfaced. The concern centered on the PAC entering a race that GOP leaders were watching closely.

Crypto PAC draws national attention

The crypto angle has made Fellowship PAC one of the more closely watched political groups ahead of the 2026 midterms. Cantor Fitzgerald donated $10 million to the group, according to federal filings reported by Bloomberg and Yahoo Finance.

Fellowship PAC is chaired by Jesse Spiro, Tether’s head of government affairs. The group also received $1 million from Anchor Labs, a crypto infrastructure firm linked to Cantor, according to earlier reports.

Fellowship PAC had reportedly aimed to raise $100 million for the 2026 election cycle. By mid-April, it had brought in $11 million from disclosed backers.

Crypto spending remains under scrutiny

The episode comes as crypto political spending grows in Washington. Axios reported that crypto groups spent roughly $120 million to $130 million in the 2024 elections, including about $40 million from Fairshake.

The 2026 cycle is drawing more attention because the industry is also pushing for clearer digital asset rules. This week, more than 100 crypto companies and lobbying groups urged Congress to move forward on market structure legislation.

Fellowship PAC’s reported Texas filing shows how crypto-linked political spending can attract attention beyond digital asset policy. It also shows how party leaders may respond when outside groups move into contested races.

TLDR

• Meta is preparing to eliminate approximately 8,000 positions (10% of total staff) effective May 20, 2026
• The workforce reduction aims to help finance the company’s ambitious AI infrastructure budget of up to $135 billion this year
• An additional 6,000 unfilled positions are being eliminated from hiring plans
• The company has introduced a controversial internal monitoring system that captures keyboard and mouse activity for AI training purposes
• Shares of META declined 2.31% in response to the workforce reduction announcement

Meta has revealed intentions to eliminate approximately 8,000 positions — representing roughly 10% of its total employee base — scheduled to become effective on May 20. The announcement triggered a 2.31% decline in META shares.

The social media behemoth positioned the workforce reduction as a streamlining initiative, though the resulting cost savings are projected to be completely absorbed by the company’s ambitious artificial intelligence investment strategy. Meta has publicly committed to allocating up to $135 billion toward AI infrastructure development throughout 2026.

What distinguishes this round of job eliminations from earlier ones is the absence of compensatory hiring in alternative divisions. Meta is simultaneously eliminating 6,000 vacant positions from its recruitment pipeline. This approach suggests the reductions represent more than a simple reallocation of human capital.

Meta Platforms, Inc., META

In an internal communication, Janelle Gale, Meta’s chief people officer, recognized that providing a month’s advance notice before individual notifications would prove “incredibly unsettling” for the workforce. She explained that premature disclosure became unavoidable following information breaches.

Workplace morale at Meta has experienced a dramatic deterioration. Information from Blind, an anonymous professional networking platform for verified company employees, indicates that over 80% of Meta-related commentary posted this year has carried negative sentiment. By comparison, only approximately 20% of such posts were negative throughout 2024.

Just days ago, leaked internal documentation unveiled a newly implemented software system that captures employee keyboard inputs, cursor positioning, and click patterns. The company states this information will serve to train artificial intelligence systems in executing routine computing operations. Participation is mandatory for all employees, with personal email usage subject to the same monitoring protocols.

The leaked documentation spread rapidly across social platforms and generated substantial criticism on Meta’s internal communication channels. A highly-rated employee comment asked: “This makes me super uncomfortable. How do we opt out?”

A Vision of Smaller Teams

Andrew Bosworth, Meta’s chief technology officer, circulated an internal position paper outlining two distinct operational models currently functioning within the organization. The first maintains conventional practices — expansive teams, comprehensive documentation, formalized evaluation processes. The second operates with minimal headcount, accelerated timelines, and AI-integrated workflows.

“These teams are tiny. They move extremely quickly,” Bosworth explained. He characterized 2025 as feeling “like 100 years ago” considering the rapid transformation enabled by AI-assisted productivity.

Meta CEO Mark Zuckerberg has become progressively more outspoken regarding artificial intelligence’s capacity to reduce team sizes. “We’re starting to see projects that used to require big teams now be accomplished by a single very talented person,” he stated during January remarks.

The organization has already reorganized segments of its engineering division with extremely flat hierarchical structures featuring 50-person-to-one-manager ratios. Meta is additionally building what it describes as a “CEO agent” designed to assist Zuckerberg in accessing and synthesizing information from throughout the enterprise.

Spending Concerns Persist

Previously, investors responded favorably to Meta’s workforce reductions. The company’s elimination of 21,000 positions throughout late 2022 and early 2023 catalyzed significant share price appreciation. However, the market’s current response has been considerably more reserved.

The primary concern centers on the likelihood that savings generated from workforce reductions will simply be reallocated toward AI capital investments, which have already reached unprecedented levels. Meta’s projected AI expenditure of up to $135 billion for 2026 may face upward revision when quarterly financial results are disclosed.

Meta Superintelligence Labs has recently unveiled a next-generation AI system. The organization indicated that the keystroke-monitoring technology will support that division in teaching its models fundamental computer proficiencies including dropdown menu navigation and keyboard shortcut utilization.

TLDR:

• EIP-8182 proposes a shared shielded pool built directly into Ethereum as a native system contract.
  
• Fewer than 1 in 10,000 Ethereum transactions were private in 2025, still below the 2020 peak.
  
• The pool has no admin key or governance token and upgrades only through Ethereum’s hard-fork process.
  
• Users can swap tokens on a DEX and reshield funds in one transaction while keeping their identity private.

EIP-8182 is a draft proposal that could bring private transfers directly into the Ethereum protocol. Currently, nearly every Ethereum transaction is fully public, exposing balances, payment amounts, and counterparties.

The proposal aims to address this by embedding a shared shielded pool into Ethereum itself. Fewer than 1 in 10,000 Ethereum transactions were private in 2025, remaining below 2020 levels.

The Core Problem With Ethereum’s Current Privacy Landscape

Existing privacy solutions on Ethereum face a structural challenge known as the anonymity-set chicken-and-egg problem.

Privacy on Ethereum works by pooling funds together, making individual transactions harder to trace. Larger pools offer stronger privacy for all users. Smaller, fragmented pools weaken privacy across the board.

New privacy applications cannot offer meaningful privacy to early users. Without sufficient privacy, new users have little reason to join.

Once a pool grows large enough, users are reluctant to leave, even for a better product, because migration reduces their privacy protection.

This dynamic means the largest pool tends to stay dominant, regardless of quality. More competing apps mean smaller individual pools and worse outcomes for users overall. A shared standard has therefore been absent from the ecosystem.

A second problem compounds this: app-level privacy systems require upgrade mechanisms controlled by specific parties — multisig holders, token holders, or DAOs. Public transfers on Ethereum carry no such trust requirement, and a private-transfer default cannot either.

How EIP-8182 Addresses These Structural Issues

EIP-8182 places a shared shielded pool directly into Ethereum as a system contract at a fixed address. It also introduces a ZK proof-verification precompile. The pool has no admin key, no governance token, and no on-chain upgrade mechanism.

In April 2025, Ethereum co-founder Vitalik Buterin called for privacy tools to be built into existing wallets. He wrote: “Wallets should have a notion of a shielded balance, and when you send to someone else, there should be a ‘send from shielded balance’ option, ideally turned on by default.” A year on, that integration has not materialized at scale.

Any wallet integrating EIP-8182 connects to one shared anonymity set. Every new user strengthens privacy for all existing participants. Applications can then compete on user experience, proving speed, and developer tooling rather than pool size.

The pool evolves only through Ethereum’s hard-fork process — the same mechanism governing all other protocol changes. This removes the need to trust any third party for upgrades.

What Developers Can Build Using EIP-8182

Recipients use standard Ethereum addresses and ENS names. No separate privacy-specific address format is required, and no off-chain coordination step is needed. A recipient registers once, and private sends work to their existing address thereafter.

EIP-8182 separates transaction authorization from proof generation. Users sign transaction details in their existing wallet and can optionally send them to a remote prover.

As the proposal notes, “the prover has the power to compute but not the power to decide,” meaning altered transaction parameters will simply fail verification.

Private funds can also leave the pool, interact with any public Ethereum smart contract, and return — all within one transaction.

This pattern supports swapping one token for another on a decentralized exchange while keeping the user’s identity and destination private.

EIP-8182 is currently in draft status. The proposal is open for review at eip8182.com, where a full specification and reference implementation are also available.

For bitcoin BTC$78,344.33 traders, the direction of the Dollar Index (DXY), a measure of the greenback’s strength against a basket of other currencies, hasn’t mattered this much in nearly four years.

That’s because the 30-day correlation coefficient between the two now stands at -0.90, according to TradingView, the most negative reading since September 2022. A reading below 0 indicates an inverse relationship: When the dollar weakens, bitcoin gains, and vice versa.

Keep in mind, though, that the reading, while widely tracked, can be influenced by bitcoin’s 24/7 trading structure, particularly weekend price action that is not mirrored in the Dollar Index’s weekday-only trading.

The coefficient of determination, or correlation squared, comes in at 0.81, implying that roughly 81% of bitcoin’s short-term price moves are statistically associated with moves in the index.

Notably, bitcoin’s rally has stalled since hitting highs above $79,000 on Wednesday. This comes as DXY bounced to 98.75 from the April 17 low of 97.63.

This is an excerpt from CoinDesk newsletter ‘Daybook.’ Sign up here, if you haven’t already.

The outlook for the Dollar Index appears supported by broader macro risks, including elevated oil prices tied to the tanker traffic disruptions in the Strait of Hormuz and a continued U.S.-Iran standoff over ceasefire negotiations.

“Macro is still trying to lean against it [BTC’s continued rally]. Oil has risen for five straight sessions and Hormuz remains effectively constrained. That should be a headwind because it keeps the inflation channel alive and keeps risk premia from fully unwinding,” analysts at Marex said in an email.

One positive is the sustained inflows into the U.S.-listed spot exchange-traded funds (ETFs). While those are keeping prices supported, industry leaders are still taking a cautious approach.

Anthony Scaramucci, founder of SkyBridge Capital, said bitcoin may not see a meaningful recovery until October or November, and the current price action aligns with BTC’s four-year reward halving cycle. He said that whales, who hold large numbers of BTC, and long-time holders have continued to sell into ETF-driven demand. Stay alert!

Read more: For analysis of today’s activity in altcoins and derivatives, see Crypto Markets Today . For a comprehensive list of events this week, see CoinDesk’s “Crypto Week Ahead.”

What’s trending

• Pentagon email floats suspending Spain from NATO, reassessing UK’s Falklands claim over Iran war rift (Reuters): A memo circulating at high levels in the Pentagon lays out options to punish NATO allies that denied access, basing and overflight rights for the Iran campaign.
• Morgan Stanley launches Stablecoin Reserves Portfolio, a money-market fund for issuers (CoinDesk): Morgan Stanley Investment Management unveiled MSNXX, a $1 NAV government money market fund holding only Treasuries and government repo, built to meet the Genius Act’s reserve requirements.
• Wisconsin sues Kalshi, Coinbase, Polymarket, Robinhood and Crypto.com over prediction markets (CoinDesk): Attorney General Josh Kaul’s complaints allege sports event contracts are unlicensed gambling, citing the platforms’ own marketing.
• DOJ arrests Special Forces soldier who made $400K on Polymarket betting on Maduro’s capture (ABC News): The master sergeant was involved in the January operation and placed around $33,000 in bets hours before Trump announced the capture, netting more than $400,000. This is believed to be the first U.S. insider-trading prosecution tied to a prediction market.

Today’s signal

The chart shows daily swings in the ether-bitcoin (ETH/BTC) ratio in candlestick format since July last year.

This week, the ratio fell nearly 3% to 0.02965, its lowest since March 15. The move has two bearish implications.

First, it confirms a downside break from the short-term ascending channel that had guided the recovery from early February lows. Second, it pushes the ratio back below the broader downtrend line that has defined the decline since August.

This breakdown reinforces bearish momentum and increases the likelihood of further downside or extended consolidation in the ETH/BTC pair, that is, it points to continued underperformance of ether relative to bitcoin ahead.

Riot Platforms has sent another 500 BTC to an NYDIG deposit address, according to on-chain data cited by Lookonchain. The transfer was worth about $39 million at the time of reporting.

The latest move adds to a series of Bitcoin transfers from Riot over the past two weeks. The miner has reportedly sent regular batches of 60 BTC to 125 BTC to NYDIG execution wallets almost daily.

Riot also made another 500 BTC deposit two weeks earlier. The repeated transfers show that the company has continued to reduce part of its Bitcoin reserves through institutional trading channels.

Q1 report showed large Bitcoin sales

Riot had already disclosed Bitcoin sales in its first-quarter 2026 operational report. The company sold 3,778 BTC during the quarter and generated $289.5 million in proceeds.

The firm reported an average sale price of $76,626 per Bitcoin. These sales placed Riot among the major public miners using Bitcoin reserves to support business needs.

Riot remains one of the largest listed Bitcoin mining companies. Its recent activity comes as miners manage tighter margins after the latest Bitcoin halving.

Moreover, the Bitcoin halving cut miner block rewards by 50%, reducing the amount of new BTC miners receive for each block. This has raised pressure on companies with high energy and infrastructure costs.

Mining difficulty has also continued to rise. Higher difficulty means miners need stronger and more efficient machines to produce the same amount of Bitcoin.

Many firms are upgrading ASIC fleets and expanding facilities to stay competitive. These costs can push miners to sell Bitcoin, especially when they need cash for operations, debt, or equipment.

Other public miners also sell BTC

Riot is not the only miner selling Bitcoin this year. MARA has sold more than 15,000 BTC for about $1.1 billion after revising its 2026 treasury policy to allow ongoing sales for operational needs.

CleanSpark also sold 405 BTC at spot prices and another 500 BTC. Core Scientific announced the sale of 1,900 BTC earlier this year and said it planned to exit its Bitcoin holdings by the end of the first quarter.

The sales show a broader shift among miners after years of holding Bitcoin reserves. Public mining firms now appear more willing to convert BTC into cash as market and operating conditions change.

Jane Street Group LLC, a prominent quantitative trading firm, has moved to dismiss a lawsuit brought by the administrator of Terraform Labs’ bankruptcy estate. In a Manhattan federal court filing on Thursday, the firm urged the court to throw out the case, which accuses Jane Street of insider trading that allegedly worsened the Terra ecosystem’s collapse.

The motion frames Terraform’s complaint as an attempt to recover funds from Jane Street to cover a fraud Terraform allegedly perpetrated on the market. The filing argues that Terraform’s claims rest on a mischaracterization of the firm’s trading activity and that any alleged wrongdoing by Terraform itself has already been prosecuted and resolved.

Terraform’s court-appointed administrator, Todd Snyder, filed the lawsuit in February, naming Jane Street, Terra founder Do Kwon’s associates, and two Terraform employees. The complaint accuses the trading firm of acting on material nonpublic information from Terraform insiders to profit from Terra-related tokens as the project unraveled.

In their motion to dismiss, Jane Street’s lawyers contend that the firm’s Terra-linked trades were motivated by ordinary market signals and public information, not any nonpublic tips. The court papers emphasize that Terraform’s collapse became widely visible to investors as the market pricing deteriorated, and that Jane Street acted to sell a deteriorating investment during the downturn.

The filing also notes that Terraform’s own representatives have publicly linked the collapse to ongoing market dynamics, arguing that the timing of Jane Street’s TerraUSD sales does not align with any disclosed material nonpublic information. The motion points to Terra’s transition to a new liquidity pool in early May 2022 and asserts that Terraform failed to identify any specific nonpublic information that Jane Street allegedly learned, despite extensive pre-suit discovery.

For context, Terraform’s dramatic downfall occurred in May 2022 when its algorithmic stablecoin TerraUSD briefly lost its peg to the U.S. dollar, triggering a broader crash in the Terra ecosystem. The implosion sent the price of Luna token sharply lower and erased roughly $40 billion in market value, a specter that still shapes regulatory and legal scrutiny of crypto markets.

Jane Street’s submission argues that the fundamental questions about Terra’s collapse—and who bears responsibility—had already been addressed through criminal prosecutions. The motion points to the guilty pleas of Do Kwon on conspiracy and wire fraud charges, which culminated in a 15-year prison sentence, as evidence that the broader fraud narrative has been adjudicated by the courts.

On the factual point at the heart of the case—the timing of Terra-related trades—the motion asserts that Terraform’s complaint is self-defeating. It notes that Terraform claimed Jane Street’s largest TerraUSD sale occurred roughly 10 minutes after “material nonpublic information” allegedly became visible to the market, a sequence the filing says is inconsistent with the asserted information flow and timing. Jane Street also contends that Terraform failed to identify any specific nonpublic information the firm allegedly obtained, despite pre-suit discovery efforts.

The motion requests that the court dismiss the suit with prejudice, meaning Terraform could not refile a claim against Jane Street on the same grounds in the future. Court filings offer a window into the procedural chess game unfolding as the parties navigate whether crypto market activity can be treated the same as traditional securities markets in insider-trading disputes.

As the litigation unfolds, the dispute raises broader questions about how insider trading claims will be treated in the relatively uncharted territory of crypto markets. The case pits a well-resourced market-maker against a bankruptcy administrator aiming to recover value for Terra creditors. The outcome could influence how other market participants respond to similar allegations in the wake of high-profile collapses.

In addition to the core claims against Jane Street, the lawsuit named Terra’s co-founder and individuals connected to the project. The legal maneuvering reflects a broader pattern in which investors and authorities scrutinize trading activity around controversial crypto events, especially those tied to failed guarantees, liquidity shifts, or ecosystem transitions. The balance between public information and alleged nonpublic tips remains central to the legal debate.

Observers will be watching closely to see how the judge weighs the timing of Terra-related disclosures against the process by which nonpublic information might circulate in crypto markets. The court’s ruling could provide a blueprint for how similar insider-trading theories are evaluated when the assets in question are algorithmic stablecoins, cross-chain tokens, or other crypto instruments that lack traditional centralized disclosure regimes.

For now, the case sits at a crossroads of market behavior, regulatory scrutiny, and the evolving standard for what constitutes actionable insider information in crypto markets. The docket remains active, and future filings will likely shed additional light on how the courts interpret these complex dynamics as crypto trading continues to mature into a regulated, litigated landscape.

Readers watching this case should note the docket referenced in the motion, which is publicly accessible. The filing material can be reviewed in the docket entry for Snyder v. Jane Street Group LLC on CourtListener: https://www.courtlistener.com/docket/72321910/29/snyder-v-jane-street-group-llc/

Key Highlights

• DeepSeek unveiled two open-source AI models: V4-Pro (1.6 trillion parameters) and V4-Flash (284 billion parameters)
• Each model features a 1-million-token context window, rivaling Google’s Gemini capabilities
• V4-Pro achieves performance comparable to OpenAI’s GPT-5.4 in coding tests and ranks second only to Gemini in reasoning tasks
• The company emphasizes significantly lower computational and memory requirements versus competitors
• News arrives amid reports of Tencent and Alibaba negotiating investment deals valuing DeepSeek above $20 billion

Chinese artificial intelligence firm DeepSeek unveiled preview editions of its newest flagship open-source AI system, V4, this past Friday. According to the company, this latest iteration delivers enhanced reasoning capabilities, cost efficiency, and an exceptionally large context processing capacity.

The firm introduced two distinct variants: V4-Pro and V4-Flash. The Pro edition features 1.6 trillion parameters, while the Flash variant represents a streamlined alternative containing 284 billion parameters, engineered for superior efficiency and cost-effectiveness.

Each variant supports processing up to one million tokens simultaneously. This capability enables them to analyze substantial volumes of text in a single operation, positioning them competitively alongside Google’s Gemini in this dimension.

The company noted that current models handle text exclusively. DeepSeek confirmed development is underway to incorporate multimodal functionality, which will enable future versions to analyze images and video content.

Performance Against Competing Systems

In MMLU-Pro testing, a standard industry benchmark, V4-Pro delivered results equivalent to OpenAI’s GPT-5.4. Performance placed it marginally below Google’s Gemini and Anthropic’s Claude Opus 4.6. For reasoning benchmarks specifically, V4-Pro secured second place behind only the most recent Gemini release.

DeepSeek highlighted that V4 has been fine-tuned for integration with AI agent frameworks including Claude Code, OpenCode, and CodeBuddy.

The organization characterized V4’s context capacity as “world leading with drastically reduced compute and memory costs.” Industry analyst Zhang Yi identified it as an “inflection point,” suggesting ultra-long context capabilities could transition from experimental research environments into mainstream commercial applications.

AI industry expert Max Liu characterized the launch as a “milestone” for China’s artificial intelligence sector, drawing parallels to the market impact when DeepSeek’s R1 initially debuted.

Financial and Strategic Landscape

This marks DeepSeek’s first significant new-generation model launch since R1 emerged in early 2025. That previous release sent ripples through global technology markets, affecting companies like Nvidia and Meta, by demonstrating that an economical, efficient model could rival expensive proprietary alternatives.

DeepSeek has not disclosed which semiconductor chips powered V4’s training process. Earlier in the year, U.S. authorities alleged the company utilized restricted Nvidia Blackwell chips. Subsequently, a report from The Information indicated training occurred on Huawei chips instead.

Huawei verified that its Ascend supernode infrastructure, utilizing Ascend 950 AI processors, would provide complete support for DeepSeek’s V4 systems.

The model debut follows closely after reports emerged that Tencent and Alibaba are pursuing investment discussions with DeepSeek at a valuation exceeding $20 billion. DeepSeek ranks among China’s six premier AI unicorn companies.

A preview build of V4 is currently accessible through Hugging Face. DeepSeek has not yet specified a timeline for the complete public release.