AI Industry Daily Radar · June 25, 2026
Executive Summary
- OpenAI unveiled Jalapeño, its first custom AI inference chip built with Broadcom, showing 50% cost savings over typical GPUs — a direct challenge to Nvidia's inference monopoly.
- Micron and Qualcomm earnings ignited a $400 billion AI chip stock rally on June 24, lifting the entire semiconductor sector including Nvidia.
- Salesforce launched Agentforce Commerce with three autonomous AI agents for shopping, buying, and merchandising — marking the arrival of agentic commerce at enterprise scale.
- GPT-5.6 odds hit 83% on Polymarket for a release before June 30, with a redesigned reward model pipeline and a rumored 1.5M token context window.
- New hiring data from SignalFire reveals engineering jobs are the most AI-resilient — engineers now make up 55% of new tech hires (up from 46% in 2019), contradicting layoff narratives.
- MoEngage acquired Aampe to deploy per-customer autonomous AI agents for B2C marketing, processing hundreds of millions of decisions weekly.
Top Stories
1. OpenAI Unveils "Jalapeño" — First Custom Chip Built with Broadcom
Summary
OpenAI unveiled its first custom AI chip on June 24, developed in partnership with Broadcom. Codenamed Jalapeño, the processor is purpose-built for AI inference — running pre-trained models in response to user queries — rather than the more compute-intensive pre-training workloads.
Early testing shows roughly 50% cost savings compared with typical AI GPUs, according to Broadcom CEO Hock Tan. OpenAI President Greg Brockman framed the chip as targeting "specific workloads that are underserved," emphasizing that the company now operates across the full stack: chip architecture, kernels, memory systems, networking, scheduling, and deployment.
The chip was designed with assistance from OpenAI's own AI models and is optimized for real-time coding workloads — directly benefiting products like Codex. While heavy training workloads will remain on Nvidia hardware for now, Jalapeño represents OpenAI's most aggressive move yet toward vertical integration and reduced infrastructure dependence.
Source
https://techcrunch.com/2026/06/24/openai-unveils-its-first-custom-chip-built-by-broadcom/
2. Micron and Qualcomm Earnings Ignite $400 Billion AI Chip Stock Rally
Summary
Semiconductor stocks surged on June 24 after Micron Technology and Qualcomm reported earnings that exceeded Wall Street expectations, driven by surging AI demand. The combined rally added approximately $400 billion in market value across the AI chip sector.
Micron's results highlighted insatiable demand for high-bandwidth memory (HBM) used in AI accelerators, while Qualcomm's forecast underscored the growing AI inference market at the edge. The rally lifted Nvidia, AMD, Broadcom, and Arm Holdings in a broad-based semiconductor upswing.
The timing is notable: the rally came on the same day OpenAI unveiled its custom Broadcom chip, and Barron's reported on June 25 that Nvidia shares continued to gain "even as AI chip competition hots up." The market appears to be pricing in that the AI chip opportunity is expanding faster than competitive threats can erode incumbents.
Source
https://finance.yahoo.com/technology/ai/articles/micron-qualcomm-forecasts-ignite-400-230933990.html
3. Salesforce Launches Agentforce Commerce — AI Agents Go Shopping
Summary
Salesforce released Agentforce Commerce on June 24, deploying three autonomous AI agents — Shopper Agent, Buyer Agent (B2B), and Merchant Agent — across its commerce platform. The launch represents the most significant agentic commerce deployment at enterprise scale to date.
Shopper Agent handles conversational shopping from discovery through checkout with full catalog and customer context. B2B Buyer Agent enables procurement via WhatsApp and SMS. Merchant Agent allows natural-language catalog management, with early customers reporting an 88% reduction in time-to-task.
Salesforce cited striking data: AI influenced 20% of global online sales ($262 billion) during the last holiday season, and retailers with shopper agents grew sales 59% faster. The platform integrates natively with ChatGPT, Google Search AI Mode, and Gemini. Additional capabilities including Agentic Commerce Search are slated for July 2026.
Source
https://www.salesforce.com/news/press-releases/2026/06/24/agentforce-commerce/
4. GPT-5.6 Looms: 83% Odds Before June 30
Summary
Polymarket odds for a GPT-5.6 release before June 30 sit at 83%, with evidence mounting that OpenAI is preparing a quiet technical update during its IPO quiet period. An internal memo from Chief Scientist Jakub Pachocki described GPT-5.6 as "a meaningful improvement" over GPT-5.5.
The release is driven in part by the now-infamous "Goblin Incident" — a post-mortem revealed GPT-5.5 exhibited a 175% increase in goblin, gremlin, and troll metaphors due to a miscalibrated reward model. GPT-5.6's core structural change is a redesigned reward audit pipeline to catch such biases before production. Capability improvements (rumored 1.5M token context window, sharper long-horizon coding, faster Codex) are downstream of this safety fix.
An enterprise Codex API response header briefly exposed the internal codename kindle-alpha on June 12. Expected rollout: ChatGPT and Codex first, with API access days or weeks later.
Source
https://www.buildfastwithai.com/blogs/ai-news-today-june-25-2026
5. Engineering Jobs Prove Most Resilient to AI Displacement
Summary
Despite widespread fears that AI coding tools would decimate software engineering jobs, new hiring data from venture firm SignalFire paints the opposite picture. Published on June 24, the analysis tracked millions of employees across 80+ million companies and found:
- Total tech hiring dropped 25% compared to 2019 levels, but engineering hiring declined only 11%.
- Engineers now make up 55% of all new hires at the 12 "Tech Majors," up from 46% in 2019.
- Early-stage startups collectively hired 7% more engineers in 2025 than in 2019.
Nvidia CEO Jensen Huang directly challenged the displacement narrative at Stanford in April: "Somebody said that AI is going to destroy all of the software engineering jobs… software engineers are busier than ever." SignalFire's Asher Bantock framed it as a Jevons paradox: AI makes engineers dramatically more productive, which increases — rather than reduces — demand for their work.
Source
6. MoEngage Acquires Aampe for 1:1 Agentic Marketing
Summary
MoEngage, serving over 1,350 consumer brands, acquired Aampe on June 24 — an AI infrastructure company that provisions a dedicated autonomous agent for every individual customer. Aampe's reinforcement learning engine runs per-user agents that independently decide message content, timing, channel, and frequency, learning from every interaction.
The acquisition signals the arrival of true 1:1 agentic decisioning at B2C scale. Aampe's system already processes hundreds of millions of agent-driven decisions per week across channels. The deal contrasts with Salesforce's broad-platform approach: MoEngage-Aampe is betting on deep, per-user decisioning as the agentic architecture that wins.
Source
https://www.prnewswire.com/news-releases/moengage-acquires-aampe-302385719.html
7. Forrester at Cannes: 90% of Agencies Use AI — But Creativity Is Suffering
Summary
A Forrester report released at Cannes Lions on June 24, conducted with the 4As, found that 9 in 10 US agencies now use generative AI and half use agentic AI. But the top use cases reveal a troubling pattern: document summarization (74%), competitive research (70%), and content building — all efficiency-focused.
Forrester VP Jay Pattisall warned: "The industry is at risk of mistaking efficiency for effectiveness." Only 61% of agencies are monetizing AI directly, treating it as a cost of business rather than a growth driver. The report calls for agencies to reinvest AI efficiency gains into creative talent and differentiated experiences — or risk hollowing out brand value.
Source
https://www.forrester.com/press-newsroom/ai-agency-adoption-cannes-2026/
Industry Trends
Trend 1: The AI Chip Landscape Is Fragmenting — Fast
In the span of 24 hours, OpenAI unveiled a custom inference chip with Broadcom, Micron and Qualcomm drove a $400 billion sector rally, Nvidia gained despite competition fears, and Google's TPU ambitions continue to draw interest from Meta. The inference chip market — once a near-monopoly — is rapidly diversifying. OpenAI's Jalapeño, Google's TPUs, Amazon's Trainium, and Qualcomm's edge AI chips are all carving out niches. The trend suggests inference will become a multi-vendor market within 2-3 years, with Nvidia retaining dominance only in training.
Trend 2: Agentic Commerce Arrives at Enterprise Scale
Salesforce's Agentforce Commerce and MoEngage-Aampe's per-customer agents represent the same trend from different angles: AI agents are moving from experimental chatbots to production commerce infrastructure. With AI already influencing 20% of global online sales, the agentic commerce layer — handling discovery, purchase, and post-purchase — is becoming a competitive necessity, not a differentiator.
Trend 3: AI Efficiency Gains Are Creating New Demand, Not Destroying Jobs
The SignalFire hiring data and the Forrester agency report converge on a common finding: AI is acting as a force multiplier, not a replacement. Engineers are busier than ever; agencies are using AI everywhere but struggling to translate efficiency into effectiveness. The Jevons paradox appears to be the dominant dynamic — increased productivity drives increased demand — challenging the dominant narrative of AI-driven job destruction.
Featured AI Products
Agentforce Commerce (Salesforce)
- Three autonomous AI agents for enterprise commerce: Shopper, Buyer, and Merchant
- Native integration with ChatGPT, Google Search, and Gemini
- Early customers report 88% reduction in catalog management time
- Official URL: https://www.salesforce.com/agentforce/commerce/
Aampe (acquired by MoEngage)
- Provisions a dedicated autonomous AI agent for every individual customer
- Reinforcement learning engine decides message content, timing, and channel per user
- Processing hundreds of millions of agent-driven decisions weekly
- Official URL: https://www.aampe.com/
Jalapeño (OpenAI + Broadcom)
- First custom AI inference chip from OpenAI
- 50% cost savings vs. typical GPUs for inference workloads
- Optimized for real-time coding models like Codex
- Official URL: https://openai.com/index/openai-broadcom-jalapeno-inference-chip/
Key Takeaways
- OpenAI's custom chip is the biggest strategic move of the year: Vertical integration from models through silicon changes the competitive dynamics of the entire AI stack.
- The AI chip market is expanding, not shrinking: A $400 billion rally on a single day of earnings reports signals that the pie is growing faster than competition can fragment it.
- Agentic commerce is no longer experimental: With Salesforce and MoEngage shipping production agents at scale, brands without an agent strategy for the 2026 holiday season are already behind.
- The AI job apocalypse narrative is empirically wrong: SignalFire's hiring data and Anthropic's own economic research both show engineering demand rising, not falling, in the AI era.
- GPT-5.6 is likely days away: The "Goblin Incident" post-mortem reveals that the next OpenAI release is as much about fixing reward model safety as about capability improvements.