Pathfinding in AI&Crypto

Your guide to crypto and AI. All the latest news in one place. Our Telegram:

[Pathfinding in AI&Crypto @Pathfinding_AI]

Polymarket is not a casino and not a bookmaker. This is a full-fledged exchange of probabilities, where the price is formed by the collective opinion of the market.

Figure out the mechanics before you invest your first dollar. 📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

We’ve already mentioned @Polymarket more than once in the news: liquidity from @wintermute_t, insider trading, CFTC regulation, and more.

But we’ve never explained how this thing works from the inside. Closing the gap. 🧵

[Pathfinding in AI&Crypto @Pathfinding_AI]

Now to practice - how to start trading:

1. Go to Polymarket. Preferably use a proxy to avoid an account ban by geo
2. Click Connect Wallet - connect MetaMask/Rabby Wallet. On the first connection the platform will require many signatures to create an account, this is normal
3. Click deposit and deposit through any convenient network and token. For depositing it is better to use stables, since there may be a significant fee on other tokens
4. Trading tip: place limit orders, not market orders. On many markets there is a large spread and little liquidity - a market order can execute at a high price. Also, limit orders have no fee from the platform itself

KYC is not required yet for most users.

[Pathfinding in AI&Crypto @Pathfinding_AI]

ETH at $250,000 - Tom Lee voiced such a scenario on June 2 at Proof of Talk in Paris, CoinDesk writes. From the current $1,765.79, that's roughly 141x.

Bitmine, where Lee is chairman, has 5,390,404 ETH - that's 4.47% of the supply. CoinGecko counts 7,428,358 ETH across 32 institutions, or 6.16% of the supply. Ethereum already accounts for more than 75% of tokenized RWAs and over 60% of the global stablecoin supply.

Recently we wrote that the SEC postponed relaxations for tokenized stocks - the market is only being assembled, and the network for it is already being chosen.📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

Grayscale launched a $HYPE ETF on Nasdaq on June 3 with built-in staking - ticker HYPG, fee 0.29%.

The fund launched with a seed investment of ~$115M in $HYPE tokens from Hyper Holdings Global LP

Why HYPG stands out:

➡️ Lowest fee. 21Shares THYP charges 0.30%, Bitwise BHYP charges 0.34%. Grayscale sets 0.29%

➡️ The only HYPE ETF with staking. The fund doesn't just hold tokens - it stakes them with a yield of ~2.2% annually. The income is reflected in the fund's value

➡️ Third HYPE ETF in a month. Three funds in that timeframe is a signal of serious institutional interest in Hyperliquid

$HYPE became the third altcoin after $SOL and $XRP to get a spot ETF in the US 🇺🇸. We've written about Hyperliquid multiple times: the token entered the top 10 by market cap, and the NYSE head called the project "bigger than Nasdaq."📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

ECB 🇪🇺 on June 2 published a report on global reserves, and there's a historic shift: gold has overtaken US 🇺🇸 Treasuries as the main reserve asset of the world's central banks for the first time since 1996.

Numbers:
➡️ Gold: ~27% of global CB reserves (was 20% a year earlier)
➡️ Treasuries: ~22% (was 25%)
➡️ The dollar overall still dominates: 42% of all reserves

Why did this happen? Two reasons. First - the price of gold rose ~60% over 2025, and what was already sitting in vaults became more expensive. Second - deliberate purchases: China 🇨🇳, Poland 🇵🇱, Turkey 🇹🇷, India 🇮🇳 are actively building up their reserves.

After the freeze of Russian reserves in 2022, it became clear: someone else's debt securities can be blocked, but physical gold in your own vault - can't. According to WGC data, in April 2026 CB purchases resumed: +17 tons after March sales.

Tether in 2025 bought over 100 tons of gold - more than any individual central bank. The ECB specifically noted this.

If CBs keep moving away from Treasuries, yields on them could rise - and that puts pressure on all risk assets, including crypto. That said, there's a nuance: the ECB emphasizes that at 2023 gold prices, Treasuries would still be in first place. So for now this is more of a revaluation effect than a mass dollar dump.📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

@omegadyor @MicroStrategy @Polymarket

[Pathfinding in AI&Crypto @Pathfinding_AI]

👀 @MicroStrategy sold 32 BTC for $2.5M before May 31, but only disclosed it in an SEC filing on June 1. On @Polymarket the question was: "Will Strategy sell any amount of BTC before May 31?" Trading volume on the contract grew to $143M.

The sale happened. But Polymarket twice resolved the bet as "No," stating that confirmation came after the deadline. The rules said "will sell," not "will announce the sale," but the platform added a clarification - after the main trading volume - that public confirmation was needed before the deadline. One trader lost ~$500,000 and wrote: "I just got scammed!"

Both "No" outcomes have been disputed and sent to UMA voting, with the verdict expected June 4-5. We've written more than once about Polymarket's issues: insider trading, $520K theft, pressure on CFTC employees. The platform is growing, but its dispute resolution rules remain its weak spot.

Even if you're right on the facts, on prediction markets you can lose because of wording. 📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

@SprotelBitcoin @MicroStrategy @Polymarket 💯

[Pathfinding in AI&Crypto @Pathfinding_AI]

SEC 🇺🇸 on June 2 published a draft Strategic Plan for 2026-2030, where digital assets for the first time in history became a separate priority on par with investor protection.

SEC Chair Paul Atkins stated: "Blockchain and crypto technologies can transform America's financial infrastructure."

What's included in the plan:
➡️ Rules for tokenization - moving stocks and bonds onto blockchain. We've already been following this topic
➡️ Regulation of staking and custody of crypto assets
➡️ Joint token taxonomy with CFTC - back in March the two regulators divided: what's a security, what's a commodity, what's a stablecoin

At the same time, for now this is a draft - public comments are accepted until July 2. There are no specific timelines for individual rules in the document, and between the plan and actual laws it could take 1-2 years.📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

👀 @Google is issuing new shares for the first time in 20 years, and right away for $80B. All the money will go to AI: data centers, GPU clusters, model training. GOOGL shares dropped 3.5% on the first day.

A secondary offering is when a company prints new shares and sells them to investors. Existing shares get "diluted" in the process: their stake in the company shrinks. It's like if a pizza got more slices, but the pizza itself didn't get bigger.

Numbers:
➡️ Offering size: $80B - the largest secondary offering by a tech company
➡️ Berkshire Hathaway came in for $10B out of its $397B cash pile
➡️ Dilution for current shareholders - only ~1.8% at a ~$4.5T market cap
➡️ Google's CapEx for 2026: $180-190B - that's twice as much as in 2025

Institutions are rebalancing portfolios in favor of AI stocks. $2.3B was pulled from BTC ETFs in May. The five largest tech companies will spend over $600B on AI in 2026 alone - that's money that could have gone into other assets.

We already covered how Anthropic raised $65B and how @cz_binance warned about an AI bubble.📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

👀 @NEARProtocol co-founder Ilya Polosukhin announced that quantum-resistant signatures based on the FIPS-204 standard will appear on testnet by the end of June.

All blockchains currently use digital signatures that a regular computer can't crack. But in March 2026 Google Quantum AI showed that quantum computers need not 10+ million but around 500,000 qubits for cracking, and the threat horizon shifted to ~2029.

Bitcoin and Ethereum still have no transition plan. NEAR is the first L1 in the top 30 to begin implementation. Plus thanks to its architecture with human-readable account names and rotatable keys, users don't need to change their address - one transaction is enough to change the lock.

We already covered the quantum topic when the US Department of Commerce invested $2 billion in quantum technologies

That said, for now this is testnet, not mainnet. Hardware wallets (Ledger and others) don't yet support the new signatures. The $NEAR price already rose 28% in May on this news. Preparation for the future starts now. 📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

👀 @durov announced that the native token of the @ton_blockchain is being renamed to Gram. The blockchain itself, The Open Network (TON), keeps its name. The rebranding will take 2-3 weeks.

🧐 Gram is the original name for the token that Durov came up with back in 2018. At the time, @telegram raised $1.7B through an ICO, but the SEC blocked the launch. Telegram returned the money to investors and paid an $18.5M fine. The community picked up the code and launched the token as Toncoin in 2021. Now Durov is bringing back the original name.

Durov explains the renaming as part of the MTONGA plan (Make TON Great Again). The idea is to remove the legacy left behind by the era without Telegram: Toncoin was launched by the community, not by Durov's team. Now that Telegram has replaced TON Foundation, became the largest validator, cut fees by 6x, and sped up the network, it makes sense to bring back the original name too.

Market reaction: after the May MTONGA moves, TON has already grown 60-120% 📈 from its lows of ~$1.35. At the time of the news, the price is around $1.99. Gifts and memecoins in the TON ecosystem have also picked up, and activity in mini-apps has noticeably jumped.

If you're holding Toncoin, you don't need to do anything. The token, addresses, and balances stay the same - only the name and ticker on exchanges will change. We did a detailed breakdown of the TON ecosystem - check it out if you want to understand what's inside.📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

Week in crypto and AI: record outflows from BTC-ETF and $HYPE at a new all-time high. BTC loses ~5.5% over the week, while individual projects are setting records. Let's go through the highlights:

1. BTC-ETF: 9 days of outflows in a row
Investors pulled ~$2.8B from Bitcoin ETFs over 9 trading days - a record since launch. On May 27 alone, $733M left in a single day, of which $528M came from BlackRock IBIT. BTC closed the week around $72,500-74,000. Geopolitics and rising bond yields continue to weigh on risk assets.

2. Google engineer made $1.2M on insider info at Polymarket
The CFTC charged a Google engineer with using Google Year in Search data to place bets on Polymarket. We covered this case in detail - the first precedent of criminal prosecution for insider trading on a prediction market.

3. $HYPE hit a new ATH - $69.45
Hyperliquid keeps rising: a new all-time high, and ETF inflows into the token exceeded $136M. The NYSE chief called Hyperliquid "bigger than Nasdaq." The contrast with BTC-ETF outflows is telling: the market is not uniform.

4. Dimon vs. CLARITY Act 🇺🇸
JPMorgan CEO Jamie Dimon stated: "banks won't accept the law" if stablecoin issuers can pay yield without banking oversight. As a reminder, the CLARITY Act passed the Senate committee two weeks earlier. The battle between banks and crypto over stablecoins is heating up.

5. OKX launched Exchange OS
OKX opened a protocol that lets anyone create their own exchange - spot, futures, prediction markets - built on the L2 network X Layer. Major exchanges are starting to provide tools instead of holding onto their monopoly.

6. No CBDC in the US
Treasury Secretary Scott Bessent called CBDC "off the table." The administration views a digital dollar as a surveillance tool and is betting on stablecoins.

What does this mean for next week? Institutions are taking profits in BTC, but money isn't leaving crypto entirely - it's flowing into individual projects. Legislative shifts (CLARITY Act, CBDC rejection) are building a long-term foundation for the industry.📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

People who built @ethereum are selling ETH and leaving the foundation. This week several major news stories at once:

➡️ Eric Connor (former core developer) stated that he significantly reduced his ETH position

➡️ David Hoffman (Bankless co-founder) sold all his ETH on May 26

➡️ 8+ key researchers have left the Ethereum Foundation since the beginning of 2026

➡️ The Ethereum Foundation itself sold another 10,000 ETH (~$23M) - the second weekly sale in a row

➡️ Harvard exited its ETH ETF position worth $87M a quarter after buying

A week ago Vitalik wrote that he wants to reduce the foundation's sales. But at the same time former EF researcher Dankrad Feist proposed creating a new organization with $1B and a leader "who wants to fight" - essentially admitting that the current structure isn't coping.

The Ethereum network works fine. The question is about the token's economics: L2 networks use the infrastructure but generate little revenue for ETH itself.📝

[Pathfinding in AI&Crypto @Pathfinding_AI]

Most AI companies will go bankrupt, but AI itself will grow exponentially - that's what @cz_binance, the founder of @binance, stated on May 29, 2026, reports Bloomingbit.

The point isn't that AI is useless. The problem is in company economics: chips, data centers, clouds, engineer salaries, and constant model training cost a lot.

🧐 Burn rate - this is the speed at which a company spends money. If cash burns faster than revenue or new investments come in, sooner or later you have to cut costs, sell out, or go bankrupt.

According to SVB, an early-stage AI company spends about $5 to generate $1 of new revenue. Among startups valued over $1B, only 21% are profitable, although they grow profit by 72%.

Example - CoreWeave. In their 2025 report they have $5.1B in revenue, but $1.17B in net loss and about $14.7B in long-term debt. Growth is there, but it's bought with expensive infrastructure.

➡️ How companies like these break: investors give money for fast growth, the company takes on obligations, and then the market demands profit. If revenue doesn't catch up with expenses, the next round becomes expensive or doesn't happen at all.

➡️ Why this matters: AI as a technology can become stronger, but holders of stocks, tokens, or shares in weak projects can still lose money, just like in the early internet era.

[Pathfinding in AI&Crypto @Pathfinding_AI]

Your laptop runs on @intel or @AMD. It's been that way for decades. Next week, that might change: according to Axios, @nvidia and @Microsoft will unveil the first Windows computers powered by Nvidia chips.

The debut is expected June 1-3 at GTC Taipei and Microsoft Build. Among the devices - Surface from Microsoft and models from Dell. No prices or exact specs yet.

🧐 NPU is a special unit inside the processor, designed for AI tasks: translation, recognition, image generation - all right on the device, without the cloud.

Why Nvidia needs this:
The company dominates in server AI chips (data centers, cloud). Now it wants to enter personal devices too. The companies are teasing this as "A new era of PC," and media are linking the announcement to Arm-based N1/N1X chips.

We recently broke down how capital is flowing into AI infrastructure. Give it a read if you missed it.📝