Advertise here | 6-min Read

Running a social media agency means your team lives inside five tools at once. And none of them talk to each other.

Planable is the content collaboration platform that changes that. Posts get drafted, reviewed, and client-approved in one place, before anything goes live. No more chasing feedback over email. No more missed sign-offs.

And now Planable connects directly to the AI tools and platforms you already use: Claude, ChatGPT, Gemini, Canva, Slack, Zapier, plus a public REST API if you want to build deeper integrations with your stack.

The result: your AI drafts the content, your team reviews it, your client approves it. One workflow, start to finish.

OpenAI Enters the Chip Race with Jalapeño

What’s Happening AI Today

Write a LinkedIn post people actually share

Prompt: You are a LinkedIn ghostwriter who has helped founders, operators, and domain experts build large audiences — not by gaming the algorithm, but by saying things worth saying. You know the posts that travel are the ones where the reader finishes and thinks: "I have never seen it put that way before."

Here is what I want to write about: [describe your idea, experience, lesson, or observation — rough is fine].

My audience: [who follows you or who you want to reach]. My tone: [direct / warm / analytical / contrarian / conversational].

Write me three versions — each a completely different structure:

1. The one-idea post — one insight, stripped to its sharpest form. No backstory. No buildup. Under 150 words. The whole post should be something a reader could screenshot and send to a colleague.

2. The story post — open with a specific moment or decision that actually happened. Let the lesson come out of the story, not before it. The reader should feel like they were in the room. Under 250 words.

3. The list post — each item must be a complete, standalone thought. No filler. Save the strongest item for last. Under 200 words.

Hard rules for all three: No "I am excited to share." No "this is a reminder that." No corporate language. The first sentence must stop a scroll on its own. End with something that invites a response — a question, a provocation, or a statement people will want to agree or push back on.

OpenAI just built its own chip. It cuts inference costs in half. And it changes the entire economics of AI.

OpenAI and Broadcom unveiled Jalapeño — OpenAI's first custom-designed AI inference chip — at OpenAI's San Francisco headquarters on June 24. Engineering samples were delivered in person by Broadcom CEO Hock Tan to Sam Altman and Greg Brockman. The chip is designed specifically for LLM inference and targets a 50% reduction in per-token serving costs.

The details:

• OpenAI spent approximately $14 billion serving ChatGPT on third-party GPUs in 2025. A 50% cost reduction at that scale is not a marginal engineering win — it is the difference between profitable and unprofitable at current pricing levels.

• Jalapeño is not commercially available. Broadcom expects a small prototype data center deployment by end of 2026, full production ramp in 2027, and scale in the first half of 2028. OpenAI and Broadcom have committed to deploying OpenAI-designed accelerators at 10 gigawatt scale with Microsoft and other partners through 2029.

• The chip arrives at the same moment Anthropic is in early talks with Microsoft to run Claude inference on Microsoft's custom Maia 200 chips via Azure. Both companies are simultaneously trying to reduce their dependency on NVIDIA's GPU pricing.

• The broader context: every major AI company is now building or procuring custom silicon. Google has TPUs. Amazon has Trainium and Graviton. Microsoft has Maia. Apple has its Neural Engine. NVIDIA's position as the only viable AI accelerator is being challenged from every direction simultaneously.

Why it matters: OpenAI's profitability timeline — currently expected no earlier than 2029-2030 — depends heavily on closing the gap between what it costs to serve a model and what it can charge for it. Jalapeño is the most direct move OpenAI has ever made on that gap. If it performs at the claimed 50% cost reduction in production, it is not just a chip story. It is an IPO story. The S-1 risk factors about infrastructure costs look different with Jalapeño in the roadmap than without it.

AI news highlights

• OpenAI and Broadcom unveiled Jalapeño — the first custom OpenAI inference chip — 50% cheaper LLM serving. OpenAI spent $14B on third-party GPUs in 2025. This is a profitability-defining move, not a marginal win.
• Anthropic formally accused Alibaba of running 28.8 million fraudulent exchanges against Claude — the largest known distillation attack on Anthropic to date. Senate Banking Committee letter dated June 10.
• Four senior Google Gemini researchers have now left for Anthropic in six days — Jonas Adler, Alexander Pritzel, and two more confirmed. Google expanded its AI coding strike team in response.
• GPT-5.6 is days away — Polymarket prices June 28 at 83% probability — developer tracking of Codex backend logs and OpenAI's historical shipping pattern point to an imminent release.
• Claude Code holds 40% of the generative AI coding market — Codex holds 21% — Mordor Intelligence June 2026 forecast. $9.3B market today, $30B by 2031. Anthropic leads.
• Anthropic privacy policy requires government ID from July 8 — the Fable 5 restoration mechanism — verified US citizens get access back. International users remain on Claude Opus 4.8 under this scenario.
• Two former xAI employees: porn accounts for well over half of all Grok traffic — xAI is leaning into it. OpenAI, Anthropic, and Google will not touch adult content. A real market split is forming.
• Sail emerges from stealth — $80M, $450M valuation — optimises AI models on existing chips — led by Kleiner. If it works at scale, inference costs drop without new hardware. Every AI company is a customer.

Trending AI tools

• Viktor.com — AI employee that does the work inside Slack and Teams — #1 of June 2026, 523 upvotes
• minimi — Ambient memory for Claude — #15 of June 2026, 553 upvotes
• Framer Agents — Design, write, and organize your site with AI agents on the canvas
• Wispr Flow — Speak naturally in any app — writes in your style with auto-edits
• Backgrind — Run AI agents over any app with sandboxed parallel sessions
• Granola — AI meeting notes on top of any call tool — no bot, no permissions
• Vapi — Build, test, and deploy voice AI phone agents
• bolt.new — Prompt, run, edit, and deploy full-stack apps from your browser

That’s a Wrap

SPONSOR US

Get your business in front of over 90k+ AI professionals

8020AI is the world’s #1 AI Newsletter, Read by 90k+ professionals from leading companies such as Google, OpenAI, Meta, and Microsoft.

We've assisted in promoting Over 500 AI-Related Products. Will yours be the next?

What We Can Offer:

• Launch an Advertising Campaign

• Introduce New Product or Features

• Other Business Cooperation

Get in touch today.

Or Email our founder Alamin at [email protected]

FEEDBACK

How was your experience with 8020AI today?

Login or Subscribe to participate in polls.

If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.