What OpenAI Is Betting On Now

Voice, founders, and where the money's moving

January 05, 2026

Reading time: 5 minutes

🗞️In this edition

  • OpenAI's Strategy to Shape AI Startups From Day Zero

  • Sponsored: Meet KIN, Your Personal AI Advisory Board

  • OpenAI Bets on Voice to Replace Screens

  • What 17 AI Experts Actually Expect to Happen in 2026

  • In other AI news –

    • Neuralink Plans Mass Production and Automated Brain Chip Surgeries for 2026

    • New State AI Rules Start Today, But Trump May Block Them

    • Why Apple's Slow AI Approach Might Win in 2026

  • 4 must-try AI tools

Something changed quietly.

Not with a launch. Not with a headline. But with a series of decisions that point in the same direction.

OpenAI is not just improving models. It is shaping how AI grows, how it interacts with us, and how it fits into everyday life.

This edition looks at those moves. From infrastructure to voice, from short-term releases to long-term systems.

If you want to understand where AI is actually heading, this is a good place to start.

What's happening:

OpenAI just opened applications for the second cohort of Grove, a program designed for people at the very earliest stage of building AI companies.

This isn't a traditional startup accelerator. It's for people who don't even have a concrete idea yet. Just curiosity about building something with AI.

The program runs for five weeks at OpenAI's San Francisco headquarters. Around 15 people will be selected for this cohort.

Participants get mentorship from OpenAI researchers and technical leaders. They get early access to new models and tools before they're released to the public. They work alongside other people exploring AI startups.

The program includes in person workshops during the first and last weeks. The middle weeks require about four to six hours of work per week.

OpenAI provides 50,000 dollars in API credits so participants can experiment with their technology. This removes the cost barrier for trying things out.

After the program ends, participants can pursue raising money for their startup, join OpenAI internally, or continue building independently.

Applications close on January 12, 2026. The program runs from January 22 to February 27, 2026.

Why this is important:

This matters because it shows OpenAI's strategy for building its ecosystem at the earliest possible point.

Most accelerators require you to already have a company and a product. Grove catches people before they've even decided what to build.

This gives OpenAI influence over what gets created. By mentoring people before they form ideas, OpenAI shapes what kinds of AI applications get developed.

It's also a talent pipeline. Some participants will build successful companies. Others might end up working at OpenAI. Either way, OpenAI builds relationships early.

The 50,000 dollars in API credits is strategic. Once people build their prototypes using OpenAI's technology, they're more likely to keep using it as they grow.

Early access to unreleased models is a major advantage. Participants can experiment with technology that competitors don't have yet. This creates a head start.

The focus on pre-idea founders is interesting. These are people who want to build something but don't know what yet. OpenAI is essentially helping them figure out what problems are worth solving with AI.

This is different from how most companies think about developer relations. Instead of supporting existing projects, OpenAI is creating new ones from scratch.

The program also addresses a real gap. Early stage founders have the least resources but often need the most support. Most programs ignore this stage entirely.

By gathering 15 people in one place for five weeks, OpenAI creates a dense network. These people will likely stay connected and help each other long after the program ends.

Comments from the editor:

What stands out is how early OpenAI is engaging with potential founders.

This isn't about funding promising startups. It's about creating promising startups in the first place.

The model is smart. Traditional accelerators compete for startups that already exist. Grove creates startups that might not have existed otherwise.

It's worth noting that other companies are doing similar things. Google and Microsoft both have AI accelerator programs. But Grove's focus on pre-idea founders seems unique.

The tight integration with OpenAI's research team is a major draw. Participants aren't just getting generic business advice. They're getting technical guidance from people building the underlying technology.

The requirement for in person participation matters. Being physically present at OpenAI headquarters creates stronger relationships and faster learning than remote programs.

The 50,000 dollar API credit is generous but also strategic. It removes financial barriers to experimentation while ensuring participants build on OpenAI's platform.

The relatively small cohort size means more attention per participant. With only 15 people, everyone gets meaningful access to mentors and resources.

The fact that OpenAI launched this in September 2025 and is already running a second cohort shows they see value in the model. They're not testing it cautiously. They're scaling it.

There's a clear competitive element here. By nurturing founders early, OpenAI ensures they use OpenAI technology rather than competitors like Anthropic or Google.

The program serves multiple purposes. It identifies talent for potential hiring. It creates companies that will become future customers. It generates goodwill in the developer community.

It's also a form of market research. By watching what early founders try to build, OpenAI learns what problems people want to solve with AI. That can inform their own product development.

The question is sustainability. Can OpenAI run meaningful cohorts multiple times per year? Will the quality remain high as they scale?

But for now, Grove represents an interesting experiment in shaping the AI startup ecosystem from its earliest stages. Not by funding companies, but by helping create them in the first place.

Meet KIN, Your Personal AI Advisory Board - MyKin.ai

w

Turn founder chaos into clear next steps.

Kin gives you a personal AI Advisory Board so you stay clear and steady across the five parts of your life: work, relationships, energy, communication, and personal growth.

  ✔ Stop rehearsing hard conversations in your head

  ✔ Say the thing you've been sitting on for weeks

  ✔ Make the call instead of spiralling on it

  ✔ Get home and actually be present

Feel the difference in 5 minutes

Talk to Kin like a real mentor and see how much lighter your evenings feel.

Try Kin for FREE on iOS or Android.

What's happening:

OpenAI is reorganizing its teams around audio. Over the past two months, they've unified engineering, product, and research groups to focus on building better voice technology.

This isn't just about making ChatGPT sound nicer. It's about preparing for a completely new type of device that could launch in 2026.

The device will be audio first. Possibly screenless. Users would interact with it entirely through voice.

OpenAI is building a new audio model set to release in early 2026. It will handle interruptions better, sound more natural, and can even speak while you're talking.

The company is working with Jony Ive, former Apple design chief. His firm was acquired by OpenAI for 6.5 billion dollars earlier this year.

Ive has said publicly he wants to use audio first design to reduce screen addiction and fix mistakes from past consumer devices.

The hardware plans include a family of devices. Smart glasses, screenless speakers, and possibly even a pen like device have all been discussed.

This fits into a bigger industry trend. Meta added a five microphone array to Ray Ban smart glasses. Google is testing audio summaries of search results. Tesla is putting conversational AI in cars.

Why this is important:

For years, screens have been the default. Every device had one. Every interaction needed one. Now major companies are betting that might change.

The push toward audio isn't just about convenience. It's about attention. Screens demand you look at them. Audio lets you stay present in the world while still accessing technology.

When OpenAI reorganizes entire teams around audio, it means they believe this is the future. Not a side project. The main direction.

The timing matters too. OpenAI's current audio models aren't as good as their text models. Researchers inside the company admit this. The reorganization is about closing that gap.

If audio becomes the primary interface, it changes everything. How software gets designed. How people expect to interact with AI. What counts as a computing device.

The involvement of Jony Ive adds weight. He designed the iPhone and other products that defined how we use technology. His focus on reducing screen addiction suggests this isn't just about novelty. It's about solving real problems screens created.

The competition is moving in the same direction. When Meta, Google, Tesla, and OpenAI all bet on audio, it suggests this isn't a fad. It's a genuine platform shift.

Smart speakers are already in over a third of US homes. Voice assistants aren't new. But combining them with advanced AI that can actually hold conversations changes what's possible.

The shift also matters for accessibility. Voice interfaces work for people who can't easily use screens. Making audio the primary interface could make technology more inclusive.

What's happening:

The overall theme is clear. AI will keep improving rapidly. But the real world economic impact will remain modest.

They predict big tech companies will spend over 500 billion dollars on AI infrastructure. OpenAI and Anthropic will both hit their ambitious revenue targets.

But GDP growth will stay normal. No economic takeoff. No dramatic productivity boom. Just steady progress within typical ranges.

On the technical side, they expect AI to handle longer software tasks. Models might be able to complete work that takes human engineers 20 hours.

Context windows will likely stay around one million tokens instead of growing much larger. Companies are hitting limits with current approaches.

The legal environment will tighten. Courts won't shut down AI companies, but they will impose real consequences for harmful outputs and copyright violations.

No AI catastrophes are expected. No major bioweapon attacks enabled by AI. No devastating cyberattacks. The technology isn't ready for that yet, and companies are adding safeguards.

Chinese companies might surpass Waymo in total robotaxi fleet size. Tesla will probably launch truly driverless service in at least one city, but expansion will be slow.

On the consumer side, the first fully autonomous vehicle might be sold to regular people. But it probably won't come from Tesla.

Why this is important:

This matters because it represents a sober, realistic view of where AI is actually heading.

These aren't hype merchants trying to sell products. They're researchers and analysts who study AI carefully. And they're saying progress will be real but measured.

The prediction about GDP is particularly important. Some people believe AI will cause explosive economic growth starting in 2027. These experts are saying that's unlikely. If we were heading for an economic takeoff, we'd see signs of it in 2026. They don't expect those signs.

The massive spending predictions matter too. Over 500 billion dollars in one year is enormous. But the experts don't think this is a bubble about to pop. They believe the spending is justified by actual customer demand.

The legal tightening is significant. The wild west period for AI is ending. Companies will face real financial consequences if their systems cause harm. This will change how they operate.

The robotaxi predictions show autonomous vehicles are becoming real. Not just prototypes. Actual commercial services expanding to multiple cities. This is a concrete example of AI technology reaching consumers.

The fact that no catastrophes are expected is reassuring. Despite fears about AI enabling bioweapons or cyberattacks, experts think the risk in 2026 is low. The technology isn't there yet, and safety measures are improving.

The software engineering prediction is notable. If AI can handle 20 hour tasks reliably, that changes how software gets built. Not immediately. But it's a clear step toward AI doing substantial coding work.

  1. Glasp YouTube Summarizer - A chrome extension that runs YouTube videos through ChatGPT/ Claude and summarizes them

  2. RoomGPT - Allows users to take a picture of their room and generate a new version of their room in different themes

  3. Interactive Mathematics - A platform that provides math lessons and an AI-powered math problem solver to help students improve their math skills

  4. WellyBox - A tool that helps users track and manage their receipts and invoices

AI is moving from performance to presence.

What matters now is not how impressive a system looks, but how reliably it shows up, how naturally it fits, and how well it earns trust over time.

OpenAI’s recent moves suggest a shift toward patience, foundations, and interaction that feels less like software and more like environment.

That is the lens we bring to OpenTools.

If something in this edition helped you see that shift more clearly, reply and tell us. We read every message.

The OpenTools Team

Interested in featuring your services with us? Email us at [email protected]