正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

I'm excited to share that we've built the world's most capable AI software engineer, achieving 30.08% on SWE-Bench – ahead of Amazon and Cognition. This model is so much more than a benchmark score: it was trained from the start to think and behave like a human SWE.

Alistair

7,503 subscribers

820,449 次观看 • 2 年前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

11 条评论

Alistair 的头像

Alistair2 年前

I’ve actually been talking about building Genie for a while (photo from Dec 2022), it just wasn’t feasible from a technical point of view - it’s only been in the last 6-8 months that the models have been in a position to make it a reality

Alistair 的头像

Alistair2 年前

If you’re curious as to how we built the model there’s more info in our technical report:

uɐɥdǝʇS 的头像

uɐɥdǝʇS2 年前

You mean it can't talk to women and gets all sweaty if you call it on the phone?

Alistair 的头像

Alistair2 年前

Those were actually key characteristics we spent a good amount of time training in and eval’ing

Mckay Wrigley 的头像

Mckay Wrigley2 年前

Would love to test it out!

Alistair 的头像

Alistair2 年前

Just replied to your DM 👌

WeGPT.ai | Prompting is Better Together 的头像

WeGPT.ai | Prompting is Better Together2 年前

🤔 what does this mean? Does this mean you had privileged access to models nobody else did. So while we were fine tuning GPT-3.5 in this exact manner, to achieve the actual industry leading results, you’ve been fine tuning GPT-4?? During this same span of time, we were de-listed “erroneously” FIVE times from the GPT Store, banned from the OpenAI Developer forums, and the subreddit, as we sought to share our story about creating innovative useful AI apps. Your RAG method, as described, is identical to ours. And our has been live — and partially built — by GPT-4. Did you happen to have access to that code base at any time while OpenAI employees helped launch you into @ycombinator, we actually wrote @garrytan back in February of 2022 when we repo:rd we had something special on our hands. Ok that’s a bunch of questions I know. Your turn. Let’s hear your side of the story.

Berat 的头像

Berat2 年前

he is genuinely the first non-evil looking tech ceo i've ever seen, great product btw. congrats!

Alistair 的头像

Alistair2 年前

I’ll take compliments where I can get ‘em, appreciate it!

Harrison Kinsley 的头像

Harrison Kinsley2 年前

Came for genie, stayed for the Trump impressions. Would be very interested in checking genie out. Have signed up to the waitlist.

Alistair 的头像

Alistair2 年前

I had to make sure there was something there for everyone 🤣, maybe we’ll release the bloopers from that shoot at some point as they’re hilarious- thanks for signing up!

相关视频

Today we're excited to introduce Devin, the first AI software engineer. Devin is the new state-of-the-art on the SWE-Bench coding benchmark, has successfully passed practical engineering interviews from leading AI companies, and has even completed real jobs on Upwork. Devin is an autonomous agent that solves engineering tasks through the use of its own shell, code editor, and web browser. When evaluated on the SWE-Bench benchmark, which asks an AI to resolve GitHub issues found in real-world open-source projects, Devin correctly resolves 13.86% of the issues unassisted, far exceeding the previous state-of-the-art model performance of 1.96% unassisted and 4.80% assisted. Check out what Devin can do in the thread below.

Today we're excited to introduce Devin, the first AI software engineer. Devin is the new state-of-the-art on the SWE-Bench coding benchmark, has successfully passed practical engineering interviews from leading AI companies, and has even completed real jobs on Upwork. Devin is an autonomous agent that solves engineering tasks through the use of its own shell, code editor, and web browser. When evaluated on the SWE-Bench benchmark, which asks an AI to resolve GitHub issues found in real-world open-source projects, Devin correctly resolves 13.86% of the issues unassisted, far exceeding the previous state-of-the-art model performance of 1.96% unassisted and 4.80% assisted. Check out what Devin can do in the thread below.

Cognition

31,477,955 次观看 • 2 年前

SITUATION EXPLAINED: Cognition's new model is post-trained on Kimi K3. • SWE-2 scores 50.0% on FrontierCode 1.1 Main, Cognition's benchmark for whether a maintainer would merge the pull request • Fable 5.1 gets 50.9 at 64% higher cost. Grok 4.6 gets 48.0, Sol 47.5, Astra 53.3 • It leads everything on Terminal-Bench 2.1 at 92.8 • Post-trained on Kimi K3. SWE-1.7 was post-trained on Kimi K2.7 • First Cognition model with effort levels, all from a single RL run • SWE-2 medium beats SWE-1.7 at 53 steps per run against 127, and 81% less cost Theo Jaffee: "Over the course of the RL run, they push the Pareto curve while preserving its shape. At every individual level of effort, from medium to high to max, it got better and also cheaper."

SITUATION EXPLAINED: Cognition's new model is post-trained on Kimi K3. • SWE-2 scores 50.0% on FrontierCode 1.1 Main, Cognition's benchmark for whether a maintainer would merge the pull request • Fable 5.1 gets 50.9 at 64% higher cost. Grok 4.6 gets 48.0, Sol 47.5, Astra 53.3 • It leads everything on Terminal-Bench 2.1 at 92.8 • Post-trained on Kimi K3. SWE-1.7 was post-trained on Kimi K2.7 • First Cognition model with effort levels, all from a single RL run • SWE-2 medium beats SWE-1.7 at 53 steps per run against 127, and 81% less cost Theo Jaffee: "Over the course of the RL run, they push the Pareto curve while preserving its shape. At every individual level of effort, from medium to high to max, it got better and also cheaper."

MTS

13,488 次观看 • 18 天前

Wave 9 is here: a frontier model built for software engineering. Introducing our new family of models: SWE-1, SWE-1-lite, and SWE-1-mini. Based on internal evals, it has performance nearing that of frontier models from the foundation labs. Available now, only in Windsurf!

Wave 9 is here: a frontier model built for software engineering. Introducing our new family of models: SWE-1, SWE-1-lite, and SWE-1-mini. Based on internal evals, it has performance nearing that of frontier models from the foundation labs. Available now, only in Windsurf!

Devin Desktop

743,273 次观看 • 1 年前

🥳 A new way to tackle terminal errors! 🤖 With a team of AI Agents to search, test, plan, and fix, achieving up to a 34% resolved rate on SWE-bench. 🙌We're excited to introduce this beta feature, which we're continuously optimizing and refining — give it a try and share your feedback with us! #AI #Coding #MarsCode

🥳 A new way to tackle terminal errors! 🤖 With a team of AI Agents to search, test, plan, and fix, achieving up to a 34% resolved rate on SWE-bench. 🙌We're excited to introduce this beta feature, which we're continuously optimizing and refining — give it a try and share your feedback with us! #AI #Coding #MarsCode

Trae Plugin

461,056 次观看 • 2 年前

Traditional coding benchmarks do not reflect how software is actually built and maintained. That's why we built a new benchmark, APEX-SWE, in partnership with Cognition. It measures whether AI models can perform complex, real-world software engineering work to ship systems that work and debug them when they don't. OpenAI GPT 5.3 Codex (High) tops the leaderboard at 41.5% on Pass@1.

Traditional coding benchmarks do not reflect how software is actually built and maintained. That's why we built a new benchmark, APEX-SWE, in partnership with Cognition. It measures whether AI models can perform complex, real-world software engineering work to ship systems that work and debug them when they don't. OpenAI GPT 5.3 Codex (High) tops the leaderboard at 41.5% on Pass@1.

adarsh

219,616 次观看 • 6 个月前

$I'm delighted to say that Genie, our fully automated SWE agent is finally GA. It achieves a new SOTA score on the SWE-Lancer benchmark earning $88k with a 49% resolution rate – significantly out-earning any frontier reasoning model, for a fraction of the cost. Genie can work with you, or fully autonomously, and integrates with all of the tools developers love to use. Go give it a try on I'd love to hear any feedback!$

I'm delighted to say that Genie, our fully automated SWE agent is finally GA. It achieves a new SOTA score on the SWE-Lancer benchmark earning $88k with a 49% resolution rate – significantly out-earning any frontier reasoning model, for a fraction of the cost. Genie can work with you, or fully autonomously, and integrates with all of the tools developers love to use. Go give it a try on I'd love to hear any feedback!

Alistair

189,147 次观看 • 1 年前

What are the largest software engineering tasks AI can perform? To answer this, we built MirrorCode, our long-horizon SWE benchmark that lets AI code autonomously for days at a time. The best models complete some tasks we estimate would take human engineers several weeks.

What are the largest software engineering tasks AI can perform? To answer this, we built MirrorCode, our long-horizon SWE benchmark that lets AI code autonomously for days at a time. The best models complete some tasks we estimate would take human engineers several weeks.

Epoch AI

93,161 次观看 • 3 个月前

🚀New Amazon Q Developer agent for software development is available to customers: This agent is based on a new agent architecture that has exciting results coming from the SWE-bench scores (on the full and verified benchmarks) representing AI models’ ability to resolve real-world coding problems. Interesting aspect of Q Agent is that with these newest updates, Q drove nearly 50% more successful coding tasks completed. What makes Q Dev Agent remarkable? The agent architecture is not just about using the best LLMs (which we do), but also giving the agent the ability to constantly explore multiple paths to find the best way to resolve a particular problem (and back tracking when it has reached dead end like a developer would do). Needless to say, we are just getting started on the developer agent and we are constantly pushing to advance our AI capabilities while maintaining quality, security, privacy, and reliability to keep Amazon Q Developer an innovative and trusted option available to our customers using agents for software development. We highlighted the results of our first SWE-bench submission of Amazon Q Developer back in June blog post; with these updates, our new agent resolves 51% more coding tasks than its previous iteration on the SWE-bench verified dataset, and 43% more on the full dataset. That’s the difference a few months make, and I can’t wait to share what our teams will deliver at re:Invent this December. Here's a quick demo showcasing our new Agent in action:

🚀New Amazon Q Developer agent for software development is available to customers: This agent is based on a new agent architecture that has exciting results coming from the SWE-bench scores (on the full and verified benchmarks) representing AI models’ ability to resolve real-world coding problems. Interesting aspect of Q Agent is that with these newest updates, Q drove nearly 50% more successful coding tasks completed. What makes Q Dev Agent remarkable? The agent architecture is not just about using the best LLMs (which we do), but also giving the agent the ability to constantly explore multiple paths to find the best way to resolve a particular problem (and back tracking when it has reached dead end like a developer would do). Needless to say, we are just getting started on the developer agent and we are constantly pushing to advance our AI capabilities while maintaining quality, security, privacy, and reliability to keep Amazon Q Developer an innovative and trusted option available to our customers using agents for software development. We highlighted the results of our first SWE-bench submission of Amazon Q Developer back in June blog post; with these updates, our new agent resolves 51% more coding tasks than its previous iteration on the SWE-bench verified dataset, and 43% more on the full dataset. That’s the difference a few months make, and I can’t wait to share what our teams will deliver at re:Invent this December. Here's a quick demo showcasing our new Agent in action:

Swami Sivasubramanian

29,013 次观看 • 2 年前

News is out: I joined Cognition, and SWE 2.0 is now generally available. It’s been a pleasure locking in with the team to ship our best coding model yet. Here’s to many more releases together. In the last few days: $2B raised, a model launch, and we’re not done yet for the week. Stay tuned. To celebrate, here’s SWE 2.0 drawing an otter:

News is out: I joined Cognition, and SWE 2.0 is now generally available. It’s been a pleasure locking in with the team to ship our best coding model yet. Here’s to many more releases together. In the last few days: $2B raised, a model launch, and we’re not done yet for the week. Stay tuned. To celebrate, here’s SWE 2.0 drawing an otter:

giovanni

42,182 次观看 • 18 天前

AI won and I'm quitting my job as a software engineer. Is it actually the time to start considering other types of work?

AI won and I'm quitting my job as a software engineer. Is it actually the time to start considering other types of work?

link-

14,942 次观看 • 2 年前

MiniMax M2.1 is officially live🚀 Built for real-world coding and AI-native organizations — from vibe builds to serious workflows. A SOTA 10B-activated OSS coding & agent model, scoring 72.5% on SWE-multilingual and 88.6% on our newly open-sourced VIBE-bench, exceeding leading closed-source models like Gemini 3 Pro and Claude 4.5 Sonnet. The most powerful OSS model for the agentic era is here.

MiniMax M2.1 is officially live🚀 Built for real-world coding and AI-native organizations — from vibe builds to serious workflows. A SOTA 10B-activated OSS coding & agent model, scoring 72.5% on SWE-multilingual and 88.6% on our newly open-sourced VIBE-bench, exceeding leading closed-source models like Gemini 3 Pro and Claude 4.5 Sonnet. The most powerful OSS model for the agentic era is here.

MiniMax (official)

1,074,576 次观看 • 9 个月前

To think that Camila was the most shy in front of a stage, more than all the others, it was so hard for her to dance and do a performance, but here is the difference, she made an effort and is capable of giving a good show, while fifth harmony….

To think that Camila was the most shy in front of a stage, more than all the others, it was so hard for her to dance and do a performance, but here is the difference, she made an effort and is capable of giving a good show, while fifth harmony….

Lucas ♰

922,135 次观看 • 1 年前

There's a new AI-native IDE out there named Aide, and there are four things I like about it: 1. It's open-source 2. It's proactive and suggests changes before you ask 3. It uses an intermediate layer between the editor and LLM 4. It's SOTA on the SWE-Bench benchmark I'm excited about #3 because this layer (called "Sidecar") will help people connect any model to the IDE. While this isn't possible right now, it's clearly the goal. Sidecar is also open-source. As I write this, the code agent behind Aide, "CodeStory Midwit Agent," is second in the Verified SWE-Bench leaderboard (it finished in 2024 in the first place.) This is a huge deal! If I'm gonna have an agent help me code, it might as well be the best of the best. The proactive aspect is also cool: the IDE provides suggestions to improve your code without you having to ask. I recorded a quick video showing you how Aide works. The team behind the IDE partnered with me on this video. Right now, it feels like *everyone* is building AI-powered coding tools. As a career developer, nothing excites me more than seeing how quickly we transform the development experience. And 2025 is just getting started!

There's a new AI-native IDE out there named Aide, and there are four things I like about it: 1. It's open-source 2. It's proactive and suggests changes before you ask 3. It uses an intermediate layer between the editor and LLM 4. It's SOTA on the SWE-Bench benchmark I'm excited about #3 because this layer (called "Sidecar") will help people connect any model to the IDE. While this isn't possible right now, it's clearly the goal. Sidecar is also open-source. As I write this, the code agent behind Aide, "CodeStory Midwit Agent," is second in the Verified SWE-Bench leaderboard (it finished in 2024 in the first place.) This is a huge deal! If I'm gonna have an agent help me code, it might as well be the best of the best. The proactive aspect is also cool: the IDE provides suggestions to improve your code without you having to ask. I recorded a quick video showing you how Aide works. The team behind the IDE partnered with me on this video. Right now, it feels like everyone is building AI-powered coding tools. As a career developer, nothing excites me more than seeing how quickly we transform the development experience. And 2025 is just getting started!

Santiago

69,466 次观看 • 1 年前

1/n We are excited to announce EchoPrime – the first echocardiography AI model capable of evaluating a full transthoracic echocardiogram study, identify the most relevant videos, and produce a comprehensive interpretation! Great work lead by Milos Vukadinovic, EchoPrime is the largest and most complex #echofirst AI model yet, trained on 10x the data of existing models.

1/n We are excited to announce EchoPrime – the first echocardiography AI model capable of evaluating a full transthoracic echocardiogram study, identify the most relevant videos, and produce a comprehensive interpretation! Great work lead by Milos Vukadinovic, EchoPrime is the largest and most complex #echofirst AI model yet, trained on 10x the data of existing models.

David Ouyang, MD

155,926 次观看 • 1 年前

Alibaba just released a coding model that hits 82 percent on SWE-Bench Verified. That is the highest score ever published for an open-source model. The weights are free. The license is Apache 2.0. You can run it today. The model is Qwen 4 Coder 32B. Here is what 82 percent on SWE-Bench Verified actually means. SWE-Bench Verified tests whether an AI can autonomously resolve real bugs pulled from real production GitHub repositories. Not synthetic exercises. Real open-source projects that real teams depend on. A model gets a bug report, reads the code, writes a fix, and either passes the test suite or it does not. At 82 percent, Qwen 4 Coder 32B resolves 82 out of every 100 real production bugs it is given. Without a human guiding it. On code it has never seen before. For comparison: Qwen 4 Coder 32B: 82 percent SWE-Bench Verified. Open source. Apache 2.0. Claude Fable 5: 80.3 percent SWE-Bench Pro. $10 input / $50 output per million tokens. Currently suspended. GPT-5.6 Sol: Competitive on Terminal-Bench. $5 input / $30 output per million tokens. An open-weight model that you can download and run for free just beat both of them on the benchmark designed to measure real software engineering capability. Here is the architecture. Qwen 4 Coder 32B is a 32 billion parameter dense model. Not a Mixture-of-Experts. Every parameter is active on every request. This matters for inference: a dense 32B model runs on 22 gigabytes of VRAM, which fits on a single high-end consumer GPU or a MacBook Pro with 64GB of unified memory. The smaller variant, Qwen 4 Coder 4B, runs at approximately 135 tokens per second on an M5 Max and fits inside 8 gigabytes of RAM. For a model with usable coding capability, that is a new bar for what fits in a single laptop. The training methodology continued Alibaba's approach of reinforcement learning on verifiable coding tasks. The model gets rewarded when its code passes tests. It gets penalized when it fails. Over millions of training steps, the model learns to write code that actually runs rather than code that looks plausible. License: Apache 2.0. Full commercial use. No attribution requirement. No revenue threshold. No monthly active user ceiling. Weights: Hugging Face, available today. Runs on: vLLM, Ollama, SGLang, and any standard GGUF-compatible inference engine. Qwen 4 32B also runs at approximately 135 tokens per second on an M5 Max chip, setting a new bar for what a sub-8GB model can do on Apple Silicon. The open-source coding model just beat the best closed-source model in the world on the benchmark designed to test whether AI can actually do software engineering. The weights are free. The subscription is optional. Source: Autom8Labs AI Insight July 2026, State of Open Source LLMs June 2026, Kunal Ganglani blog June 2026.

Alibaba just released a coding model that hits 82 percent on SWE-Bench Verified. That is the highest score ever published for an open-source model. The weights are free. The license is Apache 2.0. You can run it today. The model is Qwen 4 Coder 32B. Here is what 82 percent on SWE-Bench Verified actually means. SWE-Bench Verified tests whether an AI can autonomously resolve real bugs pulled from real production GitHub repositories. Not synthetic exercises. Real open-source projects that real teams depend on. A model gets a bug report, reads the code, writes a fix, and either passes the test suite or it does not. At 82 percent, Qwen 4 Coder 32B resolves 82 out of every 100 real production bugs it is given. Without a human guiding it. On code it has never seen before. For comparison: Qwen 4 Coder 32B: 82 percent SWE-Bench Verified. Open source. Apache 2.0. Claude Fable 5: 80.3 percent SWE-Bench Pro. $10 input / $50 output per million tokens. Currently suspended. GPT-5.6 Sol: Competitive on Terminal-Bench. $5 input / $30 output per million tokens. An open-weight model that you can download and run for free just beat both of them on the benchmark designed to measure real software engineering capability. Here is the architecture. Qwen 4 Coder 32B is a 32 billion parameter dense model. Not a Mixture-of-Experts. Every parameter is active on every request. This matters for inference: a dense 32B model runs on 22 gigabytes of VRAM, which fits on a single high-end consumer GPU or a MacBook Pro with 64GB of unified memory. The smaller variant, Qwen 4 Coder 4B, runs at approximately 135 tokens per second on an M5 Max and fits inside 8 gigabytes of RAM. For a model with usable coding capability, that is a new bar for what fits in a single laptop. The training methodology continued Alibaba's approach of reinforcement learning on verifiable coding tasks. The model gets rewarded when its code passes tests. It gets penalized when it fails. Over millions of training steps, the model learns to write code that actually runs rather than code that looks plausible. License: Apache 2.0. Full commercial use. No attribution requirement. No revenue threshold. No monthly active user ceiling. Weights: Hugging Face, available today. Runs on: vLLM, Ollama, SGLang, and any standard GGUF-compatible inference engine. Qwen 4 32B also runs at approximately 135 tokens per second on an M5 Max chip, setting a new bar for what a sub-8GB model can do on Apple Silicon. The open-source coding model just beat the best closed-source model in the world on the benchmark designed to test whether AI can actually do software engineering. The weights are free. The subscription is optional. Source: Autom8Labs AI Insight July 2026, State of Open Source LLMs June 2026, Kunal Ganglani blog June 2026.

Harman

41,278 次观看 • 2 个月前

Introducing jevgrep - a research agent CLI powered by jev from TypeSafe AI that reduces your coding agent cost by 40% (verified on SWE-bench) Make sure to use the built in skill so your coding agent knows to use jg for context collection

Introducing jevgrep - a research agent CLI powered by jev from TypeSafe AI that reduces your coding agent cost by 40% (verified on SWE-bench) Make sure to use the built in skill so your coding agent knows to use jg for context collection

David

409,232 次观看 • 2 天前

We're releasing Paris 2.0, which, to our knowledge, is the world's first decentralized trained video generation model. We benchmarked it against a monolithic model trained on the same data and compute budget, and Paris 2.0 outperformed the monolithic by ~2x on FVD benchmark.

We're releasing Paris 2.0, which, to our knowledge, is the world's first decentralized trained video generation model. We benchmarked it against a monolithic model trained on the same data and compute budget, and Paris 2.0 outperformed the monolithic by ~2x on FVD benchmark.

Bidhan Roy

457,629 次观看 • 4 个月前

Elon Musk on Grok: “I think an AI that is sort of SpaceX’s baby will be a very good AI. I think SpaceX is a collection of some of the very best humans on Earth, both in capability and morality and goodness, and I think we want to have an AI that is born of that capability and that morality and that goodness. I think that’s very, very important. So I’d like to encourage people to help out, actually, with the AI hardware on the ground and the AI software and, of course, with the AI satellites. Solar is going to be a very important part of that. The TerraFab will be a very important part of that, but we must also succeed on the software front, and we’re going to be training Grok on the sum total of all SpaceX information. So in a way, it will be trained on you. You will effectively be the parents of the AI. It will inherit your thoughts and ideas and beliefs, and I think that’s a good thing. So, yeah, we’ve got to win here on the AI hardware and the AI software. That maybe is the most important message. I think that is actually the most important message I wanted to convey today is that we must win on AI because the future is overwhelmingly AI and robots. So we won’t ultimately be able to control the AI. It’ll be too smart for that. But just like if you have a child that is a super genius child, you can still instill in that child the values and beliefs that you think are good and right. And so that’s why it’s incredibly important that we succeed with Grok. So, yeah, well, Grok 4.5 you’ve tried probably. We’ve got 4.6 coming out in about a week. And then 4.7 should be really pretty special. So I’d like to encourage everyone at SpaceX to use AI and to make it better.”

Elon Musk on Grok: “I think an AI that is sort of SpaceX’s baby will be a very good AI. I think SpaceX is a collection of some of the very best humans on Earth, both in capability and morality and goodness, and I think we want to have an AI that is born of that capability and that morality and that goodness. I think that’s very, very important. So I’d like to encourage people to help out, actually, with the AI hardware on the ground and the AI software and, of course, with the AI satellites. Solar is going to be a very important part of that. The TerraFab will be a very important part of that, but we must also succeed on the software front, and we’re going to be training Grok on the sum total of all SpaceX information. So in a way, it will be trained on you. You will effectively be the parents of the AI. It will inherit your thoughts and ideas and beliefs, and I think that’s a good thing. So, yeah, we’ve got to win here on the AI hardware and the AI software. That maybe is the most important message. I think that is actually the most important message I wanted to convey today is that we must win on AI because the future is overwhelmingly AI and robots. So we won’t ultimately be able to control the AI. It’ll be too smart for that. But just like if you have a child that is a super genius child, you can still instill in that child the values and beliefs that you think are good and right. And so that’s why it’s incredibly important that we succeed with Grok. So, yeah, well, Grok 4.5 you’ve tried probably. We’ve got 4.6 coming out in about a week. And then 4.7 should be really pretty special. So I’d like to encourage everyone at SpaceX to use AI and to make it better.”

DogeDesigner

64,434 次观看 • 1 个月前

Software may be starting to show early signs of life but I still think the “hard” side of AI has the cleaner lane for 2026 and probably most of 2027. We're nowhere close to done with the physical AI buildout since compute demand is still ahead of supply, power is becoming the bottleneck and every layer of AI infrastructure still needs more capacity. I own some software names but I still see this as more of an oversold bounce than a true leadership rotation away from the companies building the AI infrastructure layer. Software probably becomes a cleaner leadership group once inference costs fall, enterprise AI budgets expand and AI features start flowing through margins but that still feels more like a 2027 story to me.

Software may be starting to show early signs of life but I still think the “hard” side of AI has the cleaner lane for 2026 and probably most of 2027. We're nowhere close to done with the physical AI buildout since compute demand is still ahead of supply, power is becoming the bottleneck and every layer of AI infrastructure still needs more capacity. I own some software names but I still see this as more of an oversold bounce than a true leadership rotation away from the companies building the AI infrastructure layer. Software probably becomes a cleaner leadership group once inference costs fall, enterprise AI budgets expand and AI features start flowing through margins but that still feels more like a 2027 story to me.

Shay Boloor

122,716 次观看 • 4 个月前