Wafer, an AI inference startup with just eight employees, has raised $40 million in Series A funding at a valuation exceeding $200 million. The company turned down acquisition offers from multiple cloud and inference service providers. Wafer's inference business reached $8 million in annualized recurring revenue within three months of launch. The company's technology uses an AI agent to automatically optimize model performance on existing hardware, including NVIDIA and AMD GPUs. In internal tests, running GLM 5.2 on AMD MI355X achieved about 80% of the throughput of NVIDIA B200 at less than half the cost. Customers include Vercel and Inworld AI. After the funding, Wafer began hiring for four roles, all requiring five-day in-office work in San Francisco, with engineering offering a base salary of $250,000 plus equity.

Matt Palmer, head of developer experience and education at Cursor, outlined how Grok Bot works in an early beta write-up, describing it as an AI agent with its own always-on cloud computer. Rather than acting only inside a chat window, the system can open webpages and apps, access files, take screenshots, reuse browser sessions, and hand control back to the user when it hits login walls such as SSO, 2FA, captchas, or payments. Palmer said the setup felt closer to what a personal agent should actually be, compared with earlier tools that required too much deployment work, account management, and constant maintenance. The article breaks the product into several parts: persistent cloud execution on a Linux VM, layered memory across users, agents, and projects, trigger-based automation through messages, schedules, Slack, and git events, plus integrations with the same connectors and skills used by Cursor, including Notion, Slack, GitHub, MCP servers, Cursor Cloud Agents, and GitHub MCP. Palmer also described guardrails such as natural-language rules, a separate review agent, and allow/block lists. He gave examples ranging from demo generation and internal content monitoring to grocery comparison across Instacart and Amazon delivery, and DoorDash group-order navigation. After several weeks of use, he said the system had gradually earned his trust.

Base, the Layer 2 network built by Coinbase, has introduced the Builder Grant Program to support teams developing on its ecosystem. The program offers up to $5,000 in seed funding, alongside go-to-market (GTM) support, product feedback, faster technical assistance, and access to Base ecosystem resources. It targets projects in prediction markets, token launchpads, new asset creation, consumer apps, AI agent applications, and DeFi — including lending, leverage looping, yield generation, vaults, and tokenized stocks. Eligible applicants must have a live product, be focused on Base development, and seek help with product optimization and marketing.

PANews on Aug. 5 published its Top 10 ranking of column articles for July 2026, a list compiled from overall content quality and readership across all columns on the platform. The outlet framed July as a month of structural adjustment across macro, technology, and crypto markets: AI narratives entered a valuation reset phase, storage stocks pulled back, Bitcoin ended the month at about $62,900 with a 7.3% gain, while spot trading activity weakened and on-chain trading share climbed. Against that backdrop, AI and robotics dominated reader attention, accounting for seven of the 10 most popular articles. The top five entries came from animajoe, Think AI, MSX Research Institute, Tiger Research, and EX.IO. Topics ranged from AI hardware supply chains and humanoid robots to storage-sector valuations, AI agent wallets, and whether an AI bubble could mirror the 2000 dot-com collapse. The remaining ranked pieces covered stablecoin business models, second-half AI allocations, Bitcoin mining under U.S. grid stress, volatility in South Korean equities, and bottlenecks shaping the next leg of AI speculation. PANews also used the ranking announcement to invite more writers to open columns on its platform.

POAP founder Isabel said the project will begin winding down after operating for more than five years, bringing an end to a platform that issued millions of digital collectibles across hundreds of communities. According to her statement, POAP worked with global organizations including Coinbase, American Express, Warner Music Group and Bayer, and also saw large-scale adoption during Devcon, where more than 8,000 collectibles were minted in a single week. The project also launched an Airport Rally campaign that allowed collectors around the world to claim POAPs at airports. Isabel said the shutdown reflects the difficulty of building a sustainable business model without compromising the project’s core principles, citing crypto funding cycles, distribution models, fast-changing technical infrastructure and broader market cycles as key pressures. She added that she has not yet decided on her next step, but is now thinking about how go-to-market models for companies may evolve in the AI era.

POAP founder Isabel said the project will gradually wind down after operating for more than five years, closing a chapter for a protocol that issued digital collectibles across hundreds of communities. In her remarks, Isabel said POAP minted millions of digital collectibles over the past several years and worked with global organizations including Coinbase, Amex, Warner Music Group and Bayer. She also pointed to major activation moments during Devcon, where POAP minted more than 8,000 collectibles in a single week, and to the Airport Rally campaign, which let collectors around the world claim POAPs at airports. Despite that reach, Isabel said building a sustainable business model without compromising the project’s core principles became difficult because of crypto funding cycles and distribution patterns. She added that fast-changing technical infrastructure and market cycles made operations harder. Reflecting on the project, Isabel said user communities remain one of the most undervalued assets for companies, that products with real value can spread a brand through users, and that connecting users should not be overshadowed by short-term technology hype and price speculation. She said she has not decided on her next step and is now thinking about how go-to-market models may evolve in the AI era.

Erik Torenberg argues venture capital is shifting from judgment-driven investing to a model centered on winning deals, with scaled platforms better positioned to serve founders in the AI era.

Castle Labs used this week’s Chronicle to make a broader point about the current crypto cycle: headline activity metrics are no longer enough on their own. Investors are asking where protocol revenue goes, how durable it is, and whether token holders, LPs, stakers or treasury structures actually capture any of it. That shift is visible across several cases covered in the report. Maple, despite posting a record first half with $4.6 billion in AUM, 81% year-over-year growth, and $4.4 million in Q2 revenue, is facing questions over the design of its new rules-based buyback framework. The protocol plans to tie buybacks to monthly gross revenue, but tokens repurchased under the plan are set to flow into the SYRUP Strategic Fund, leaving open questions about whether the mechanism behaves more like treasury management than a direct value-accrual model for SYRUP holders. Uniswap has now activated protocol fees on V4, reviving a familiar debate around LP economics and UNI utility. Sky, meanwhile, reported a record June revenue run rate of $419 million and more than $250 million in USDS yield distribution, yet its Prime Agent structure makes it harder to determine what SKY holders are actually entitled to. The report also highlights Theo’s integration of Fidelity International’s FILQ into thBILL, OndoPerps’ launch with tokenized equities as collateral, and several themes Castle Labs is tracking, including Circle Agent Stack, Robinhood Chain and an SEC roundtable on IPO access.