Business, Deals & Funding
Claude Code Changelog
v2.1.170
Version 2.1.170 introduces Claude Fable 5, described as a 'Mythos-class' model made safe for general use with capabilities exceeding any previously generally available model. The update also fixes a bug where sessions were not saving transcripts and not appearing in the --resume option when launched from the VS Code integrated terminal or any shell that inherited Claude Code environment variables.
Why it matters
The bug fix for session transcripts not saving when launched from VS Code's integrated terminal is a genuinely useful improvement that likely affected many users, since VS Code is a very common development environment. The introduction of Claude Fable 5 as a new model tier is notable, though the marketing language ('Mythos-class') is vague without concrete benchmarks or capability descriptions. Overall, this is a meaningful update combining a practical bug fix with a significant model upgrade.
TechCrunch AI
Meta signs first AI data center deal in India with Reliance
Meta has signed its first AI data center deal in India, partnering with Reliance Industries to develop a 168-megawatt AI-enabled data center in Jamnagar, Gujarat. The facility will support Meta's global AI computing needs and can be expanded over time. It will be powered by renewable energy and cooled using desalinated seawater, with Meta covering all energy and water costs. This deepens a relationship that began with Meta's $5.7 billion investment in Jio Platforms in 2020 and a $100 million AI joint venture launched last year. The deal comes amid a broader rush of AI infrastructure investment into India by companies including Microsoft, Amazon, Google, OpenAI, and Blackstone-backed AirTrunk. India's data center capacity has grown from 375 MW in 2020 to about 1.5 GW in 2025, with projections of over 8 GW by decade's end. Meta also contracted nearly 1 GW of new renewable energy capacity…
Why it matters
This is a strategically significant move for both Meta and India's AI infrastructure ecosystem. For Meta, diversifying its data center footprint into India makes sense given rising demand for AI compute, geopolitical considerations around concentration risk, and India's improving policy environment including tax exemptions for foreign cloud providers. For Reliance, this validates its ambition to become a full-stack AI infrastructure provider, offering everything from construction to renewable p…
Guardian AI
Do we really need gigantic, noisy, water-guzzling datacentres ruining our communities? In this economy? | First Dog on the Moon
This is a cartoon editorial by First Dog on the Moon (a regular Guardian cartoonist) criticizing the proliferation of large data centers, highlighting their negative community impacts including excessive noise, water consumption, heat generation, pollution, and contribution to climate change. The piece sarcastically questions whether these environmental and social costs are justified for trivial AI uses like generating images of 'a funny dancing cabbage.'
Why it matters
The cartoon takes a strongly critical stance against the expansion of data centers, particularly those driven by AI demand. It frames them as environmentally destructive and community-harming infrastructure that disproportionately serves frivolous purposes. The tone is satirical and dismissive of the tech industry's data center buildout, implying the trade-offs between environmental costs and the actual utility of AI-generated content are deeply unfavorable.
Guardian AI
Chinese activist in UK told by X that abusive deepfakes do not breach rules
A Chinese activist based in the UK, Apple Peiqing Ni, founder of the China Dissent Network, was targeted with deepfake posts on X (formerly Twitter) depicting her as a sexually promiscuous drug addict after she posted about Tiananmen Square. When she reported the abusive content to X, the platform told her the posts did not breach its rules. UK police had advised her to file the complaint with the US-headquartered platform.
Why it matters
This case highlights a deeply troubling intersection of technology-enabled harassment, transnational repression, and platform accountability failures. Deepfake abuse designed to discredit and silence a political dissident is a serious form of harassment that should clearly violate any reasonable content policy. X's response that this does not breach its rules is alarming and suggests the platform's moderation standards are woefully inadequate, particularly for protecting vulnerable individuals…
Guardian AI
How do you want to live? Both James Valentine and the pope have offered a challenge to humanity at a crossroads | Peter Lewis
The article draws a parallel between the late Sydney radio presenter James Valentine and the pope, both of whom have challenged humanity to actively shape the world they want future generations to inherit. Valentine's death in April 2026 prompted widespread grief, but his spirited approach to life and departure left a lasting impression. The piece frames both figures as offering guidance at a critical crossroads for humanity, likely touching on themes of technology, AI, and preserving human values in a rapidly changing world.
Why it matters
The article appears to be a reflective opinion piece that uses the juxtaposition of a beloved Australian media personality and the pope to explore deep questions about human agency, technology (particularly AI, as suggested by references to 'Magnifica Humanitas'), and the kind of future we are building. The author seems to advocate for intentional, values-driven engagement with the challenges of modernity rather than passive acceptance of technological and social change.
TechCrunch AI
Google just fired a warning shot in the AI subscription price wars
Google has cut the monthly price of its Google AI Plus subscription from $7.99 to $4.99 while doubling included storage from 200GB to 400GB. Originally launched in January as the most affordable paid AI subscription in the U.S., the tier includes features like video generation via Omni Flash, Google Flow creative studio, and NotebookLM. Venture capitalist Chi-Hua Chien of Goodwater Capital sees this as a signal that AI infrastructure is entering a commoditization era, drawing parallels to how web-era infrastructure companies like Cisco and Lucent were eventually commoditized. He argues that companies like OpenAI and Anthropic, both of which have filed confidentially to go public, will face increasing margin pressure as vertically integrated giants like Google leverage their distribution and bundling advantages. The price war had already been playing out in emerging markets like India, w…
Why it matters
This is a significant and predictable development that validates what many have long suspected: raw AI capability is rapidly becoming a commodity, and the companies best positioned to win are those with existing distribution, vertical integration, and the ability to subsidize AI services through other revenue streams. Google is perfectly suited for this kind of price war — it can absorb lower margins on AI subscriptions because it monetizes users across search, ads, cloud, and its broader ecosy…
Guardian AI
Seattle enacts year-long ban on new AI datacenters
Seattle's city council unanimously passed a year-long moratorium on the construction of new AI datacenters, making it the largest US city to enact such a ban amid growing national backlash against the energy consumption of AI infrastructure. The move is notable given that Seattle's metro area is home to major tech companies Amazon and Microsoft.
Why it matters
This is a significant development that reflects mounting tensions between the rapid expansion of AI infrastructure and local communities concerned about energy consumption and environmental impact. While a one-year moratorium is temporary, Seattle's status as a major tech hub makes this a symbolically powerful move. It signals that even cities deeply intertwined with the tech industry are willing to pump the brakes on unchecked AI growth. The unanimous vote suggests broad political consensus on…
The Verge AI
I tried Siri AI, and so far it actually works
The Verge's Allison Johnson tested Apple's newly upgraded Siri AI in iOS 27, announced at WWDC, and found that it actually works for practical tasks like adding multiple calendar events from emails, diagnosing plant problems, creating shopping lists, and providing context-aware answers by referencing personal email and calendar data. While acknowledging these are basic AI assistant capabilities, Johnson emphasizes that the fact they work reliably is significant for Apple. However, she notes these features lag behind what Google's Gemini has offered on Android for over a year, which makes sense given that the new Siri is built on Gemini models. Apple reportedly has proprietary technology working alongside these models both on-device and in the cloud.
Why it matters
The article strikes a cautiously optimistic but measured tone. Johnson is genuinely impressed that Siri finally works for practical everyday tasks, particularly for parents managing busy schedules, but she's clear-eyed about the fact that these capabilities are table stakes in 2026 and that Google's Gemini has been doing the same things for much longer. The framing of 'baby's first AI assistant stuff' paired with 'it's huge that it actually works' captures the sentiment perfectly — Apple is lat…
From X/Twitter
- Chase Dimond says Claude Fable 5 one-shotted a full ecommerce email mockup for MVMT with zero edits.
- Trevin argues your DESIGN.md should follow the Google spec with frontmatter, not just prose describing your design system.
- Claude doesn't count messages — it counts tokens. Here are 10 tricks to stretch your daily limit without upgrading.
- Carolina Milanesi makes the case that Apple's most interesting WWDC move was what it didn't announce — no moonshots, no autonomous agent demos, just incremental polish.
- One WWDC graphic quietly outlines the shift from an app-based OS to an intent-centric operating system — and it might describe a new class of devices.
- Early Fable 5 impressions: it thrives on big complex tasks, feels slow even at low effort, and ultracode spawns a swarm of parallel agents.
From Reddit/HN/YC
- [Hacker News] PowerToys 0.100 ships with a new shortcut guide, command palette improvements, and more.
- [Hacker News] WebCLI asks a fun question: what if the browser was just another Unix command?
- [Hacker News] Linux developers are vibe-coding with Copilot to keep vintage AMD GPUs alive — giving the HD 2000 through HD 6000 series a second life.
- [Hacker News] Germany is pouring €100B into making its trains run on time.
- [Hacker News] The case for purpose-built local AI agents over general-purpose ones.
- [Hacker News] Loom — an open-source delivery harness for coding agents.