
Published on

In this new edition of the Alpha Recap, we take a look back at the week's most important crypto market insights: major news, yield and airdrop strategies, key developments, and concise analyses to help you look beyond the noise.
The Alpha Recap aims to bring you the most important Alphas from the crypto market each week. Every Friday, we deliver a condensed version of the most valuable information from our Alpha Feed.
Exclusive to OAK Premium members, the Alpha Feed gathers insights, yield and airdrop strategies, as well as key market information. In other words, it embodies OAK Research's DNA: giving you filtered content that cuts through the market noise.
Hyperliquid confirmed this week that the deployment of HIP-4 prediction markets will soon become permissionless, starting with a testnet phase before a mainnet rollout. In practice, the mechanics largely mirror what we've seen with HIP-3: deployers will have to lock up 500,000 HYPE, roughly $30.5M at current prices, an amount that can also be slashed in the event of manipulation or improper resolution on a given market.
Validators will define standardized templates, stored immutably, which deployers will have to use to create their markets. Alongside this, each deployer will receive an allocation of 100 "issues", meaning the possible resolutions for markets. Once a market resolves, the deployer recovers its tokens and can launch a new market. On the functional side, deployers will be able to capture up to 50% of the fees generated by their markets, all of which must be denominated in USDC in line with AQAv2.
The stakes around going permissionless are high, because HIP-4 markets remain very underdeveloped. Daily volumes currently amount to just a few million dollars, with open interest hovering between $20M and $30M, a far cry from the hundreds of millions processed by Polymarket or Kalshi.

→ In our dedicated Alpha, we look at the reasons behind this still-limited adoption, the platforms best positioned to benefit from the opening, and above all the implications of this new vertical for HYPE.
Trade 100+ perps with up to 40x leverage on a fully decentralized exchange.
Alphabet reported results far above expectations, with revenue of $119.8B (+24% year over year), $40.8B in operating income and a Google Cloud segment that keeps accelerating. With $24.8B in revenue (+82%), Google Cloud saw its operating margin climb from 20.7% to 35.6%, a sign that the massive AI investments are starting to translate into profitability as well.
At the same time, fears of a decline in the search engine in the face of ChatGPT still aren't materializing. Google Search continues to grow at 17% year over year, while YouTube posted 13% growth. In fact, according to Alphabet, embedding AI features across its products is actually helping sustain that momentum.
These performances also come with an unprecedented shift in the group's financial profile. For the first time since its IPO in 2004, Alphabet is reporting negative free cash flow, to the tune of $5.9B. More specifically, the company generated $39B in operating cash flow but invested $45B in its infrastructure, and now expects up to $205B in capex in 2026, versus $91.5B a year earlier. To fund that effort, it raised close to $50B in equity, issued $20B in debt and suspended its share buyback program.
→ Our Alpha breaks down how to read the headline +298% net income figure, why the return on these hundreds of billions is far from guaranteed, and the parallel with the semiconductor thesis we recently shared with our Premium members.
Since Noxa halted new token launches on July 11, all but admitting that its infrastructure could no longer handle the success it had encountered, the battle to succeed it has quickly taken shape. Two protocols now dominate the landscape, Pons and Long.xyz, with what are in fact almost opposing visions of how memecoins should launch.
Pons has taken hold at a spectacular pace, its market share jumping from 5.7% on July 14 to 60.9% on the 22nd. In a single week, the protocol saw more than 66,000 tokens launched for close to $380M in cumulative volume. In practice, its model remains deliberately minimalist, with no bonding curve, liquidity locked from deployment onward and a buyback & burn mechanism funded by protocol revenue. A V2, announced this week and which we studied in detail in our Alpha, could nonetheless reshuffle part of the deck.
Long.xyz takes a radically different approach. The protocol relies on a dynamic Dutch auction mechanism, but its true distinguishing feature lies in "memestocks", memecoins quoted directly against tokenized equities such as NVIDIA or SpaceX. Somewhat paradoxically, memecoins are thus becoming a liquidity engine for the Robinhood Chain's tokenized equities, at a time when those very assets had struggled to find a genuine use case.

→ Our Alpha details how both models work, analyzes Pons' conversion figures behind its volumes and the hopes riding on its token, and finally revisits the strengths and the limits of each of these two approaches.
Unlock all our research and get the right insights, at the right time.



