科技热门推文 — 2026年8月13日

今日科技动态：随着 xAI 发布由英伟达基础设施提供支持、并在后训练阶段取得显著提升的 Grok 4.6，人工智能领域的竞争进一步加速。据报道，Grok 4.7 或将在数周内推出，而谷歌也正推动 Gemini 团队加快进度。自主型 AI 智能体正在重塑工作流程、DevOps、视频制作和网络安全，但成本问题以及需求过度集中于少数超大规模云服务商仍令人担忧。与此同时，英伟达正向现有云计算巨头发起挑战，多家 AI 基础设施企业公布了强劲财报，围绕欧洲新兴人才生态系统的初创企业投资热度也在持续上升。

1. minchoi (Group Score: 773.2 | Individual: 63.7)

Cluster: 29 tweets | Engagement: 1196 (Avg: 86) | Type: Tech

Grok Bot just killed 1,000 startups.

Each Bot gets its own computer.

It learns your workflows, uses your apps, and works while you sleep.

Grok just became your 24/7 AI employee. https://t.co/QSxmWzqI0I\n\nQT @bot: Introducing Grok Bot, now in early beta.

Bots are AI teammates that do real work for you. They sign in to your tools, use them just like you do, and come back with finished work. https://t.co/uyfA97yo98

See 28 related tweets

@ferologics: RT @bot: Introducing Grok Bot, now in early beta.

Bots are AI teammates that do real work for you. ...

@Daniel_Farinax: I gave Grok Bot a try but I'm stuck in this loop where it's asking me for Cursor Premium seat even t...
@VaibhavSisinty: SpaceXAI just launched Grok Bot. It is not a chatbot. It is an AI employee. 🤯

Each bot gets its own...

@Daniel_Farinax: This is about to become one of the most iconic and widely recognized icons on your phone and desktop...
@VaibhavSisinty: Grok Bot launched today. This guy already built an open source version. And shipped it. 🤯

It is cal...

2. nvidia (Group Score: 769.1 | Individual: 39.7)

Cluster: 30 tweets | Engagement: 6520 (Avg: 2088) | Type: Tech

Congrats to the @SpaceXAI team on the release of Grok 4.6.

Grok 4.6 brings frontier intelligence, running and trained on NVIDIA GB300 NVL72 with NVLink to deliver exceptional performance, reliability and lowest token cost.\n\nQT @SpaceXAI: Introducing Grok 4.6.

It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price. https://t.co/RtTbpXcb3a

See 29 related tweets

@chetanp: Incredible achievement by SpaceXAI and Cursor. Enterprises and startups have already figured out Gro...
@cryptopunk7213: i think we can settle the debate on grok catching up to gpt and fable now. 4.6 is a v strong model n...
@Baconbrix: Grok 4.6 is here! 🚀 this has been my daily driver for over a week, no going back!

Try it now in Gr...

@mark_k: Man, it's brilliant to see SpaceXAI back at the AI frontier! The duopoly of OpenAI and Anthropic was...
@testingcatalog: BREAKING 🔥: SPACEXAI RELEASED GROK 4.6, AND IT IS NOW ON PAR WITH FABLE 5 and GPT 5.6 SOL ON MANY BE...

3. aakashgupta (Group Score: 298.8 | Individual: 36.2)

Cluster: 11 tweets | Engagement: 22 (Avg: 63) | Type: Tech

Grok 4.6 runs on the exact same 1.5 trillion parameters as Grok 4.5. xAI got a 5-point intelligence jump by retraining weights it already had.

Every gain in this launch came from post-training. Better supervised fine-tuning, heavier reinforcement learning, identical foundation underneath. That is also why the price held at $2 in,$ 6 out per million tokens. Same weights cost the same to serve, so the improvement ships free.

The agentic scores show where the RL went. APEX-Agents up 10 points. Terminal-Bench up 66% in a single release. Those benchmarks measure whether a model can finish a job, and they respond to training method far more than to size.

Now read across the top row. Three labs, three architectures, and the intelligence index says 61, 61, 62. The gap between frontier models has stopped coming from pretraining scale. It comes from how well each lab teaches the weights it already has, and how cheaply it can serve them.

The serving cost part matters most for xAI specifically. The company sits around $500M in ARR while Anthropic is near$ 47B and OpenAI around $25B. Revenue in this market follows workflow lock-in, and developers locked into their workflows two years ago.

Which explains the whole strategy of this chart. When you can't win on installed base, you win on eval parity at a lower price and let the benchmarks do the selling. Every point Grok closes on these tables lowers the switching-cost argument for someone paying 3 to 5x more per token elsewhere. xAI is fighting an incumbency war with the two cheapest weapons available, training method and price.

The 2.1 trillion parameter Grok 4.7 ships in a few weeks. That one is the scale bet. This release proves the other bet already pays. Same model, same price, five more points of intelligence.\n\nQT @SpaceXAI: Introducing Grok 4.6.

It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price. https://t.co/RtTbpXcb3a

See 10 related tweets

@itsolelehmann: feeling really good about my Elon and Grok prediction now.

Grok is now officially a frontier model ...

@elonmusk: Grok 4.6 is objectively #1 when considering intelligence, speed & cost https://t.co/1SowBEzlJV\n...
@peterwildeford: ...if Grok has Sol/Mythos-level cyber capabilities, well I'm looking forward to what crazy antics en...
@TeksEdge: Holy 💩! Did not expect Grok 4.6 to perform so well. Historically, Grok has slightly underperformed ...
@shiri_shh: Grok 4.6 just tied GPT-5.6 Sol on the Artificial Analysis Index, one point behind Fable 5.

Frontier...

4. HarryStebbings (Group Score: 222.6 | Individual: 45.6)

Cluster: 6 tweets | Engagement: 399 (Avg: 137) | Type: Tech

One of my single best investments (Lovable) came from me pitching a founder (Anton) on investing in Project Europe.

I sold that Project Europe would become the talent magnet for young European founders.

Anton invested.

Then he said, "do you want to invest in Lovable?"

Anton is super clearly a generational founder. Without a moments hesitation, I said "yes, yes, count us in".

That check is 50x+ at this stage and Anton continues to build one of the most important companies in the world.

LFG brother! 🚀\n\nQT @antonosika: Lovable just raised $400M at a$ 13.3B valuation to build the business that helps build businesses.

Apps built on Lovable now get 900M visits every month, and we’ve quadrupled ARR in the last 12 months.

Since we started, Lovable has become the place where founders create million dollar businesses, business leaders spin up new product lines, and people inside companies rewire workflows and build tools that fit their exact needs, typically at a fraction of what software used to cost.

There’s a lot more to do.

We will create the most intuitive platform to build and run a business.

That means:

Understanding what success looks like for you, then helping you achieve it. Not only ensuring your software works but also that you can attract customers, improve workflows, and grow your company.
Deeper integrations that work with your existing tech stack so you can build the interface to your entire business across sales, ops, marketing, finance, etc.
Becoming the safest platform to build and run your business on, which means secured code by default and that you get help to make safe decisions. As you build, the platform carries more of the security weight, spotting risks and helping you close them, so you can move fast without needing to become a security expert.
Staying model-independent and orchestrating the best models for every task so you don’t have to. We’ll continue to absorb the complexity of evaluating and teaching models to route to the best ones in the background, to minimize cognitive load for our users.

As more people depend on Lovable, the quality standards our team is driving increase every single day.

We’re growing the team with people with a founder mindset: curious, ambitious, and who think end to end. Model training, product geniuses, infrastructure tinkerers, and security are the priority.

The world is full of ambitious people with deep knowledge of problems worth solving. Very few of them have historically had the power to act on them. Our ambition is to give people that power, and then help them achieve more than they thought possible.

Thank you to everyone who has followed along our journey. Thank you to @MenloVentures for leading, the Scaleup Europe Fund (managed by @eqt) for co-leading, and all of the new and returning investors who recognize our global ambitions and will help make them a reality.

We have made a lot of progress but we are nowhere close to done, it’s still day zero.

See our blog: https://t.co/d1L2ZQPjtN

See 5 related tweets

@deedydas: "Lovable has no moat" is the most common take in the valley, yet…

— $500M+ revenue run rate, 7x yea...

@Lovable: RT @antonosika: Lovable just raised $400M at a$ 13.3B valuation to build the business that helps bui...
@SebJohnsonUK: Co-led by the ScaleUp Europe fund, and now backed by @balderton.

GREAT to see european firms gettin...

@bakkermichiel: Congrats @antonosika and @FabianHedin , it’s been a crazy ride since investing back in the GPT Engin...
@Lovable: RT @AlexandrePesant: Two years ago I joined what would become @Lovable because I believed AI code ge...

5. XFreeze (Group Score: 210.2 | Individual: 31.6)

Cluster: 10 tweets | Engagement: 688 (Avg: 577) | Type: Tech

Elon Musk just gave a huge update on Grok 4.7

Initial training is already complete, and Grok 4.7 should be ready in just 3–4 weeks

Elon just said: “Grok 4.7 will exceed all current models”

And said it is already significantly better than Grok 4.6

But the most important part is what SpaceXAI is doing right now

A massive amount of SpaceX company data is being added during supplemental training

SpaceX is known for the most extreme, hardcore real-world engineering on Earth, solving problems at the absolute edge of physics

That engineering knowledge is about to give Grok 4.7 a completely different advantage

Elon said he would be shocked if any model is better than Grok 4.7 at real-world engineering

And even while saying: “Anthropic is a great company and will probably release improved models soon”

Elon still expects Grok 4.7 to move ahead of everything

Grok 4.6 just arrived today and Grok 4.7 is already shaping up to be the real monster

See 9 related tweets

@teslaownersSV: Elon Musk on the next versions of Grok.

"Grok 4.6" is expected to arrive in about a week.

And Mus...

@martin_casado: RT @elonmusk: @cognition Grok 4.7 will exceed all current models.

That said, Anthropic is a great ...

@chrisparkX: Whether you’re an individual engineer or a large enterprise, no model beats Grok on cost efficiency ...
@Mayhem4Markets: Musk saying he's about to mog every model out there.\n\nQT @elonmusk: @cognition Grok 4.7 will excee...
@Kyrannio: RT @ns123abc: 🚨BREAKING: SPACEXAI JUST DROPPED GROK 4.6

A frontier model built for long-running age...

6. MTSlive (Group Score: 205.5 | Individual: 48.8)

Cluster: 8 tweets | Engagement: 1556 (Avg: 183) | Type: Tech

SITUATION DETECTED: Google co-founder Sergey Brin has urged key staff to go all-in on Gemini and catch up to the frontier, including telling DeepMind staff in a town hall to move faster, per Reuters.

See 7 related tweets

@wallstengine: $GOOGL Co-founder Sergey Brin has been personally urging key AI staff to go all-in on Gemini and mov...
@MTSlive: SITUATION UPDATE: Sergey Brin is pushing DeepMind to work on recursive self improvement, per Reuters...
@StockSavvyShay: $GOOGL co-founder Sergey Brin has been personally pushing key AI teams to accelerate Gemini’s develo...
@theojaffee: https://t.co/3eDwioLUED\n\nQT @MTSlive: SITUATION DETECTED: Google co-founder Sergey Brin has urged ...
@negligible_cap: Reportedly $GOOG Co-Founder Sergey Brin urged staff to go “all in” on Gemini and bring it back to th...

7. alex_prompter (Group Score: 198.6 | Individual: 33.1)

Cluster: 7 tweets | Engagement: 20 (Avg: 60) | Type: Tech

Many AI video creators are facing a new challenge: the cost of AI production can sometimes exceed traditional filmmaking.

But is that really unavoidable?

Optimizing workflows may be the key to making AI video production more efficient and cost-effective.

Discover how professionals approach AI filmmaking at the BytePlus webinar, hosted by the official platform for Seedance 2.5, with insights from Rickey Liu.

Director of 2xLabs, Kane Shi, Co-Founder of AZ8, and Carey Wang, Applied AI Architect of BytePlus.

August 13, 18:00 BJT / 12:00 CEST / 11:00 BST

Watch here: https://t.co/dUOClkS7ju\n\nQT @BytePlusGlobal: AI video is moving beyond experimental clips and into real creative workflows.

On Aug 13, see how Dreamina Seedance 2.5 helps creators and teams: -Build AI video workflows from script to screen -Unlock longer, more consistent, and more editable video creation -Dreamina Seedance 2.5 demo and workflow breakdown

We’ll also be joined by Kane Shi, Co-Founder of AZ8, Rickey Liu, Director of 2xLabs, and Carey Wang, Applied AI Architect at BytePlus, bringing studio, creative, and technical perspectives to real AI video workflows.

See 6 related tweets

@rohanpaul_ai: The future of AI video is not just about generating images, it’s about creating complete stories wit...
@Parul_Gautam7: The “crew of one” concept is actually pretty cool.

Instead of bouncing between multiple AI tools fo...

@xIrissy: The real breakthrough in AI video isn’t speed it’s continuity.

Most workflows break at the handoffs...

@genspark_ai: 🚀 Seedance 2.5 is now live in Genspark AI Video Agent. 30 seconds in one run, no stitching. Up to 30...
@TheoBuildsAI: Most AI video workflows still feel like:  generate a clip, then try to rescue consistency later.

Th...

8. teslaownersSV (Group Score: 188.7 | Individual: 38.1)

Cluster: 8 tweets | Engagement: 569 (Avg: 181) | Type: Tech

Elon Musk on why SpaceX must win in AI. 🤖🚀

"We must succeed on the software front."

Musk says SpaceX's AI future depends on three things:

• AI hardware on the ground • AI software • AI satellites powered by solar energy

He also says Terafab will play an important role in building that future.

But the most important part, according to Musk, is Grok.

"We're going to be training Grok on the sum total of all space information."

Musk describes SpaceX as the "parents" of the AI—saying Grok will inherit the thoughts, ideas, beliefs, and values of the people who build it.

And his biggest message is simple:

"We must win on AI."

Because Musk believes the future will be overwhelmingly dominated by AI and robots.

The goal isn't just to build the smartest AI.

It's to build an AI with the right values—and make sure it helps humanity move forward. 🤖🌍🚀

See 7 related tweets

@teslaownersSV: Elon Musk on why AI should care about humanity.

"It’s important that we have an AI that cares abou...

@XFreeze: Grok is built to relentlessly pursue the truth, help us understand the universe, and ultimately use ...
@teslaownersSV: Elon Musk on the future of SpaceXAI.

"The future is fundamentally AI and robots."

Musk says Space...

@XFreeze: Elon Musk explains the deeper reason why SpaceX succeeding in AI matters for humanity’s future

"It'...

@rohanpaul_ai: Elon Musk at SpaceX meeting:

"Our AI revenue will exceed all other SpaceX revenue by September, an...

9. firstadopter (Group Score: 178.6 | Individual: 42.2)

Cluster: 5 tweets | Engagement: 386 (Avg: 137) | Type: Tech

MUST READ POST. Drop what you are doing and read this RIGHT NOW. Maybe Google Cloud not so hot now.

Nvidia is doing an end run to beat the major hyperscalers at their own game. Chess not checkers.

"the twist is that balance sheet edge is eroding in real time. Google just printed its first negative-FCF quarter"\n\nQT @_clarktang: The reality is what we are seeing unfold is Nvidia speedrunning the creation of a synthetic hyperscaler.

Apologies in advance to all the investors who are stuck in their priors that this will trigger.

But what is a hyperscaler? Strip it down and it’s a scaled infrastructure collective of CPUs, networking, storage with a development platform on top. It fulfills two purposes. Financial: it pools and smooths the financial obligations of its users, renting infrastructure as opex instead of capex. And Operational: it builds software that makes consumption the underlying primitives simple by abstracting them away.

The hyperscaler makes a healthy 35-40% operating margin by buying hardware at bulk pricing, pooling scale to get a lower cost of capital, and driving utilization of that hardware with software that shares and shards workloads across many customers.

But in the age of AI, the atomic units of compute changed. Training (massive coherent clusters) and inference (agentic workloads) - require a fundamentally different configuration of resources. These new workloads require dramatically more accelerated compute, shifting the design target from multi-tenant utilization (the cloud era) to absolute workload performance (the AI era). The economics of the data center inverted. A giant, redundant fleet of Amazon Basics CPUs and storage doesn’t work when the job is synchronous training and one straggling node stalls the entire cluster. For inference, tokens per watt and time to first token dominate the economics, not how many VMs you can pack in a box. And none of it works in a world of limited power (at least in the West. Maybe in China).

As Nvidia built more compute and sold it to the hyperscalers, it faced a fundamental problem. The hyperscalers had classic innovator’s dilemma - expecting 35-40%+ op margin, along with an underlying desire to commoditize Nvidia's 75% GMs with their Amazon Basics equivalent. Pay an ASIC vendor a 25% margin instead of Jensen’s 75%, then stack your own 40% on top! They owned the customer relationships too, enterprises developed on AWS, Azure, GCP and their data was captive there too. But most important of all, these companies moved at their own pace. They were not scrappy or hungry to operate at the pace Nvidia or the AI labs felt was necessary to build out compute to fulfill the demand in front of them. They would never look at retrofitting a 35MW site outside of Ashburn, Virginia!

Meanwhile, a group of hungry entrepreneurs noticed the fat margins the hyperscalers earned renting what was basically stock Nvidia hardware with limited software on top, and started building businesses around it. Nvidia - skeptically at first - recognized that working with these partners would lead to faster development cycles and competitive fires and pressures for the ecosystem. Thus the neoclouds were born.

The software these neoclouds co-developed with Nvidia were purpose built for the new workloads. They solved the new problems and requirements operating the new infrastructure needed. They were ready with hotswaps, they did predictive maintenance, they built new storage software that was built for training with cheaper ingress and egress fees, because their competitive drive was to win workloads, not to lock in enterprise data on their platform. And it was working - AI labs started preferring to work with them over the hyperscalers. Common complaints on the incumbents: too slow, too particular with how their clusters were built, virtualization and networking overlays that made GPU clusters underperform stock Nvidia reference designs. Neocloud bare metal was cheaper too as their teams built AI software, not a cloud data warehouse business. And they were happy to run at half the margin (~20%) that the big guys would never accept.

But the hyperscalers still had one structural advantage: their balance sheets. Investment grade. Able to fund speculative capacity ahead of demand and rent it out at much higher spot rates. The neoclouds couldn’t play that game as lenders would only finance hardware that was already contracted with offtake. And more expensive if that offtake were the labs which at an earlier point were much more speculative. If only they could build ahead of demand, they could maybe earn the kind of returns Elon is achieving on Colossus. But the twist is that balance sheet edge is eroding in real time. Google just printed its first negative-FCF quarter and raised $50B equity. Microsoft is carrying$ 329B of leases signed but not yet commenced. Even the IG balance sheets hit the wall - more capital had to come from somewhere else.

And that's how we got to where we are today. Look at what Nvidia has actually built. The operational half of a hyperscaler: DSX OS and Mission Control to run and operate GPU fleets, DSX reference designs and Omniverse digital twins as hardened playbooks for building a data center itself. Dynamo for inference serving. All the old secret sauces of the hyperscalers built specifically for new age data centers that they have led the way in architecting. Offered to any hungry, technically competent team with a serviceable site.

And then the financing half: the revenue share and credit support model that smooths utilization across a distributed fleet the way multi-tenancy used to. Support the operator, release capacity to demand, share in upside, and now bring $500B of third-party capital to the table. Nvidia standardized the asset with reference designs, proved the compute was “fungible and transferable across customers and operators” and showed infrastructure investors DD unlevered yields across 7-8% hurdles. Those investors wet their beaks on early special situation financings, saw the paybacks, and understood the demand was global. That’s why Jensen spent 2025 flying around Europe, the Middle East, and Southeast Asia - these are the ground zero for new compute sites. The reality is that this didn’t happen just over the last 3 months. CoreWeave master agreement in 2023, the$ 6B spot reserve backstop in 2025 (to sponsor capacity for the inference clouds), the Blackrock AI Infrastructure Partnership in 2024, Brookfield’s $100B fund with Nvidia in 2025, KKR Helix with Nvidia in 2026. And now six independent financing platforms. Chess!

So now the three fears by name. Circularity? Monday was the opposite with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR bringing third party capital, independently underwritten apart from one another, replacing Nvidia’s balance sheet rather than just extending it. Useful lives / underwritability of these assets? CoreWeave just disclussed A100s, 6 year old silicon contracted through 2029 and pushed 25% price increase on its fleet in July. The collateral is aging more like an aircraft than a smartphone as feared. Market share? If you don’t see that the platform of Nvidia and the fungibility of this compute is the reason why this is even possible - the skeptics themselves are making the bull argument. The complaint that these platforms keep capital tethered to Nvidia and away from other ASICs / accelerators… $500B that can only buy Nvidia reference architecture is a moat dressed up as a risk.

So what were you doing when the first synthetic hyperscaler was built under your nose? :)

All views expressed are my personal views. Does not reflect the views of Altimeter or Nvidia or anyone else. Full disclosure I/we may hold positions in companies mentioned. Purely for discourse and thinking - no financial advice.

See 4 related tweets

@AltimeterCap: RT @_clarktang: The reality is what we are seeing unfold is Nvidia speedrunning the creation of a sy...
@altcap: Important framing. 🧐🎯📈\n\nQT @_clarktang: The reality is what we are seeing unfold is Nvidia speedru...
@PatrickMoorhead: I don’t agree with everything here nor many of the characterizations (Trainium as “Amazon Basics”), ...
@WhiteCollarExit: RT @WhiteCollarExit: 3 bullish implications of the $500B$ NVDA consortium (Save this)

First, a sixt...

10. MarioNawfal (Group Score: 176.0 | Individual: 30.0)

Cluster: 10 tweets | Engagement: 354 (Avg: 358) | Type: Tech

🇨🇳🇹🇼 China-linked hackers (suspected) used open-source AI tools to launch the first known fully autonomous cyber-attack on Taiwanese govt. websites.

Israeli cybersecurity firm Dream found the AI agents acted like a coordinated team.

Over 4 days last month, they mapped 21 systems, adapted when blocked, compromised at least 85 accounts, and stole 2,500+ personnel records.

The attack later spread to Taiwan’s nuclear safety agency and at least 7 energy companies.

Experts say governments must now assume they are under PERMANENT AI-powered cyber assault...

We’re not ready for the speed at which the world will change over the next few years.

Source: Financial Times / Writer: Jamie

See 9 related tweets

@FT: Attackers used open-source AI agents to build a hacking tool that behaved like a co-ordinated cyber ...
@WhaleInsider: JUST IN: 🇨🇳🇹🇼 China-linked hackers used AI agents to launch cyberattack against Taiwan - FT. https:/...
@Techmeme: Researchers say suspected Chinese hackers used open-source AI agents to build an autonomous hacking ...
@clashreport: Autonomous AI Agents Penetrate Taiwanese Government and Energy Infrastructure

Threat actors linked ...

@BRICSinfo: JUST IN: 🇨🇳🇹🇼 Chinese-linked hackers compromise Taiwan government websites using 'autonomous' AI age...

11. edzitron (Group Score: 158.9 | Individual: 27.8)

Cluster: 7 tweets | Engagement: 661 (Avg: 680) | Type: Tech

Despite myths that there's "incredible demand for AI," 70%+ of hyperscaler AI revenues come from OpenAI and Anthropic. In FY26, $24.1bn of Microsoft's$ 34.33bn in AI revenue came from OpenAI - meaning that GPU rentals and AI software are single-digit billion dollar businesses. https://t.co/PiiawChYj7\n\nQT @edzitron: Joined CNBC's @SquawkCNBC to talk about how 70% of AI revenues are coming from Anthropic and OpenAI, NVIDIA's vendor financing by proxy, and how outside of unprofitable and unsustainable AI labs, there's single-digit billions of dollars of demand for AI compute and software. https://t.co/5vV8XDwTSg

See 6 related tweets

@edzitron: NVIDIA has created vendor financing by proxy - it's the vendor, the customer and the backstop, using...
@edzitron: Joined CNBC's @SquawkCNBC to talk about how 70% of AI revenues are coming from Anthropic and OpenAI,...
@coinbureau: 🚨HUGE: The AI money supercycle is here, with analysts expecting Big Tech to spend over $1 TRILLION n...
@edzitron: Dario Amodei of Anthropic claims to love American values, and I can think of nothing more American t...
@BusinessInsider: While Anthropic leads in AI business adoption, OpenAI's GPT-5.6 Sol took the crown over its rival's ...

12. wallstengine (Group Score: 147.7 | Individual: 43.2)

Cluster: 5 tweets | Engagement: 356 (Avg: 112) | Type: Tech

$NBIS just had its best day since September 2025, surging 34% after a stellar earnings report. https://t.co/6x950YHtaR\n\nQT @wallstengine: NEBIUS$ NBIS Q2’26 EARNINGS HIGHLIGHTS

🔹 Revenue: $582.3M (Est.$ 573M) 🟢; +454% YoY 🔹 Adj. EBITDA: $236.2M (Est.$ 175M) 🟢; LY -$21M

🔸 “We could sell our entire 2027 capacity on these terms today. We are deliberately not doing so because we see higher value in retaining some capacity for immediate customer needs.”

Reaffirms FY26 Guide: 🔹 Customer Prepayments: >$9B expected in 2026 🔹 Contracted Capacity: 5 GW (Prior: >4 GW) 🔹 Reaffirmed across all metrics

Other Q2 Metrics: 🔹 Diluted EPS: - $0.68 (Est. -$ 0.62) 🔴 🔹 Net Income: - $190.4M (Est. -$ 183M) 🔴 🔹 Average Annual Contract Value: $20M+ per MW 🔹 AI Cloud Revenue:$ 575M (+514% YoY) 🔹 Adj. Net Loss: - $33.2M; -64% YoY 🔹 Operating Cash Flow:$ 2.25B; LY - $167.7M 🔹 Cash:$ 8.04B

See 4 related tweets

@MelvinInvests: RT @MelvinInvests: Nebius is now charging $40-50 million per megawatt on new capacity deals and it s...
@danielnewmanUV: Imagine being short $NBIS into this print. Oh my lord. 😳🙃\n\nQT @StockSavvyShay:$ NBIS DESTROYS Q2 E...
@StockSavvyShay: $NBIS has raised its power guidance at every quarterly report for the past year from >1 GW to 5 GW i...
@StockSavvyShay: $NBIS UP MORE THAN 20% TODAY 🙃 https://t.co/f7Emlwpx3Z\n\nQT @StockSavvyShay:$ NBIS DESTROYS Q2 EA...

13. Nas_tech_AI (Group Score: 141.3 | Individual: 30.8)

Cluster: 5 tweets | Engagement: 75 (Avg: 21) | Type: Tech

No wonder she's balding already.😭 https://t.co/uLcTeVMNaO\n\nQT @yuaan1in: Introducing @NuphosAI, the AI-Native DevOps Workspace.

We realized AI-native DevOps shouldn’t just be Claude Code or Codex with access to your AWS CLI.

It needs a new workspace where humans and agents can investigate incidents, understand infrastructure, take action, and build shared knowledge together.

Built for your infrastructure, your tools, and your rules.

Here’s Nuphos in action ↓

See 4 related tweets

@Nas_tech_AI: Kubernetes CNI, CSI, CRI & CDI explained visually. https://t.co/NC9r4dnujs https://t.co/l2TzDcDr...
@Nas_tech_AI: The anatomy of a GitHub Actions CI/CD YAML file 👇 https://t.co/NC9r4dnujs https://t.co/zcZMEWqiMW\n\...
@Nas_tech_AI: AWS: Anxiety Web Services. https://t.co/NC9r4dnujs https://t.co/iUeRgGK5gF\n\nQT @yuaan1in: Introduc...
@Nas_tech_AI: The real DevOps stack is just different ways to ruin your day.

But Don't worry Nuphos will be you s...

14. wallstengine (Group Score: 139.4 | Individual: 38.2)

Cluster: 4 tweets | Engagement: 207 (Avg: 112) | Type: Tech

COHERENT $COHR Q4’26 EARNINGS HIGHLIGHTS

🔹 Revenue: $2.1B (Est.$ 1.99B) 🟢; +34% YoY 🔹 Adj. EPS: $1.74 (Est.$ 1.61) 🟢 🔹 Gross Margin: 40.2%; +215 bps YoY 🔹 Oper Income: $446M (Est.$ 427M) 🟢; +62.1% YoY

Q1 Guide: 🔹 Revenue: $2.2B-$ 2.4B (Est. $2.14B) 🟢 🔹 EPS:$ 1.85- $2.05 (Est.$ 1.77) 🟢 🔹 Total Operating Expenses: $400M-$ 420M 🔹 Gross Margin: 39.5% to 41.5% 🔹 Tax Rate: 18% to 20%

Segment Net Revenue: 🔹 Datacenter & Communications: $1.6B 🔹 Industrial:$ 430.5M

Other Q4 Metrics: 🔹 Net Income: $351M (Est.$ 323M) 🟢; +82.7% YoY 🔹 Operating Margin: 21.8%; +381 bps YoY

Comments: 🔸 “We enter fiscal 2027 with exceptional customer demand, expanding production capacity, and multiple new growth platforms beginning to ramp.”

🔸 “prioritizing investments to expand manufacturing capacity”

See 3 related tweets

@wallstengine: SIMILARWEB $SMWB Q2’26 EARNINGS HIGHLIGHTS

🔹 Revenue: $77.2M (Est.$ 75.5M) 🟢; +9% YoY 🔹 Adj. EPS: $...

@danielnewmanUV: AI Boom $CSCO is executing far better than Internet Boom Cisco. What. A. Quarter. 👏🏻🚀\n\nQT @wallste...
@wallstengine: TRIMBLE $TRMB Q2’26 EARNINGS HIGHLIGHTS

🔹 Revenue: $972.0M (Est.$ 952M) 🟢; +11% YoY 🔹 Adj. EPS: $0....

15. ReutersLegal (Group Score: 137.8 | Individual: 27.3)

Cluster: 7 tweets | Engagement: 1 (Avg: 3) | Type: Tech

Bank of America said on Wednesday it plans to deploy $250 billion through July next year to support U.S. digital and infrastructure projects, an initiative it says will bolster the country's economic growth and help create tens of thousands of jobs. https://t.co/8l7KA1qtn2

See 6 related tweets

@ReutersLegal: Bank of America said on Wednesday it plans to deploy $250 billion by July 2027 to support U.S. digit...
@wallstengine: Bank of America $BAC plans to deploy$ 250B through mid-2027 across AI data centers, semiconductors, ...
@WhaleInsider: JUST IN: Bank of America plans to deploy $250 billion through mid-2027 across AI data centers, semic...
@Reuters: Bank of America launches $250 billion initiative for US tech, energy infrastructure https://t.co/Ep9...
@Techmeme: Bank of America says it plans to deploy $250B by July 2027 to support US digital and infrastructure ...

16. NeuraFlowAix (Group Score: 128.5 | Individual: 33.0)

Cluster: 4 tweets | Engagement: 1 (Avg: 130) | Type: Tech

This is where creator marketing is heading.

Less obsession with vanity metrics, more focus on relevance, influence and finding the right voice for the right product.\n\nQT @panoverse: Most brands still choose creators based on follower count.

A stronger process looks at:

who is already shaping the conversation which formats are gaining traction which creators genuinely fit the product

That’s the approach behind our creator matching.

We’re building a global creator community across AI, SaaS, tech, and e-commerce, and connecting creators with relevant paid campaigns as new briefs come in.

Behind each match, Panoverse helps brands with market research, campaign strategy, creator selection, outreach, and execution.

Over the past few months, campaigns we’ve supported have generated more than 10 Billion views on X.

Creators in AI, SaaS, tech, and e-commerce: DM us to join the community and be considered for upcoming paid campaigns.

When the right brief comes in, we’ll reach out. https://t.co/HLFZ6ahBYg

See 3 related tweets

@ethancole_ai: The best creator for a campaign isn’t necessarily the biggest one.

Sometimes a smaller account that...

@alexaiworks: Follower count is one of the most overrated metrics in creator marketing.

Relevance and actual infl...

@heyshrutimishra: RT @panoverse: Most brands still choose creators based on follower count.

A stronger process looks ...

17. firstadopter (Group Score: 123.8 | Individual: 38.4)

Cluster: 4 tweets | Engagement: 316 (Avg: 137) | Type: Tech

Looks like I was right. Also, Google Cloud won. Insane. Just insane.

Reuters: "By August, Google had delayed the new flagship version of Gemini by two months after internal testing showed its performance continued to lag behind rivals in areas such as coding"

"Senior leaders in the cloud division, meanwhile, expect Kavukcuoglu's appointment to alleviate tensions with DeepMind over allocation of Google's constrained supply of AI chips"

"Four sources characterized its laggard status as a result of lingering structural challenges that made it fall behind in the first place."

"Constraints in computing capacity, for instance, and disagreements between Gemini's many project leaders led to fewer resources for areas that would prove critical, such as coding, two of the sources said. Google's bureaucratic nature also caused it to release models at a slower pace than nimbler rivals, the four sources said."\n\nQT @Reuters: EXCLUSIVE: Inside the Google executive moves that led to its big AI reshuffle https://t.co/uMR15x1mdg https://t.co/uMR15x1mdg

See 3 related tweets

@Ric_RTP: RT @Ric_RTP: Google just quit the AI race on purpose, and it is about to make MORE money than everyo...
@firstadopter: As the smart investors at Thrive Capital say, tech investing is all about the product. Product, prod...
@firstadopter: Dear podbros modeling out GCP upside to the fifth decimal point, you're missing the forest for the t...

18. XFreeze (Group Score: 115.8 | Individual: 35.7)

Cluster: 4 tweets | Engagement: 298 (Avg: 577) | Type: Tech

Elon is building SpaceXAI toward 10 GW of compute, hundreds of billions in potential revenue, and ultimately an AI infrastructure that extends beyond Earth

"Grok 4.5, you've probably tried. We've got 4.6 coming out in about a week. And then 4.7 should be really pretty special. I'd like to encourage everyone at SpaceX to use AI and to make it better”

"We've already built the most powerful AI training clusters in the world. This is what we expect to do by the end of next year - about 10 times more than what we've done thus far. This will be very important practice for space AI. I do think AI training will continue to be on the ground, but AI inference, the day-to-day use of AI, will be in space.

We're aiming to get to 10 gigawatts by the end of next year. Depending on the value per watt - it's probably going to be $30 to$ 50 - if we bring 10 gigawatts of AI online by the end of next year, it will be $300 to$ 500 billion a year in revenue. Big numbers”\n\nQT @SpaceX: We made rockets reusable and are rebuilding the internet in space. The next challenge: making life multiplanetary and understanding the true nature of the universe

Watch @ElonMusk deliver a company update to @SpaceX employees https://t.co/5c8rxoCQfu

See 3 related tweets

@teslaownersSV: Elon Musk on SpaceXAI's massive compute expansion.

"We've already built the most powerful AI train...

@Scobleizer: AI soon will turn profitable.\n\nQT @SawyerMerritt: Elon in new presentation: "Our AI revenue will e...
@XFreeze: Elon just said AI will be 99% of SpaceX's entire value within five years

"The factory for Starmind ...

19. sundarpichai (Group Score: 115.5 | Individual: 27.3)

Cluster: 6 tweets | Engagement: 1166 (Avg: 5666) | Type: Tech

The new Pixel 11 line up is here, designed for Gemini Intelligence and with lots of new helpful features.

A few I’m excited about: Rambler, a new voice input built for how you actually talk. Magic Capture for effortless photos. And HiLight, a subtle glow when your phone is face down that lets you know when important people in your life are trying to call you.\n\nQT @rosterloh: The best of Gemini is right at the heart of the new Pixel 11 phones and the new Pixel Watch 5. We call it Gemini Intelligence. It’s no longer just an operating system that helps you use your phone, it’s an intelligence system that gets things done for you.

See 5 related tweets

@Arm: 🎉 The new Google #Pixel11 series is here! Designed for Gemini Intelligence, the latest Pixel devic...
@Google: RT @sundarpichai: The new Pixel 11 line up is here, designed for Gemini Intelligence and with lots o...
@NewsFromGoogle: RT @Google: Get your first look at the Pixel 11 family. Refined hardware, made to last. Fresh colors...
@NewsFromGoogle: RT @rosterloh: The best of Gemini is right at the heart of the new Pixel 11 phones and the new Pixel...
@CNBC: Google’s new Pixel 11 puts Gemini at center of AI phone battle with Apple https://t.co/E6ohTilybC...

20. MTSlive (Group Score: 109.6 | Individual: 32.1)

Cluster: 4 tweets | Engagement: 76 (Avg: 183) | Type: Tech

Jason Ma reveals the first true scaling law in robotics: robots trained on 1 million hours of first-person human video predictably get better without ever seeing robot data

"Yesterday we announced our new flagship robot foundation model called Dyna-2. It's very significant for the entire field for many different reasons. First of all, it's the first robot foundation model trained on at least one million hours of data. That in itself is a very challenging infrastructure challenge."

"What we demonstrated is that even if the one million hours of data is entirely first-person video of humans doing manipulation tasks, we actually see scaling law for transferring to robot embodiments that the model has never seen before."

"By training on more human data, we actually see predictable performance improvement on robots. That's a big deal because the biggest challenge facing robot foundation models is that we don't have enough data."

"Unlike language models, there are just not enough robotics data out there, and it's very difficult to collect these robot data. Only by passively observing humans doing things can we scale this kind of data, and this is the first time we showed there is a lot of improvement that can transfer." @JasonMa2020 @DynaRobotics\n\nQT @DynaRobotics: Today we are introducing Dyna-2, a world-action model pre-trained on one million hours of human video. At this scale, for the first time, we discovered several new scaling laws:

• world-action models exhibit scaling law on human data across four orders of magnitude, from 1000 to 1,000,000 hours, • this human data scaling law implied a scaling law on never seen robot data, • both data and objective matter; world modeling and scaling on video data are essential for cross-embodiment scaling transfer to emerge

🧵

See 3 related tweets

@MTSlive: .@JasonMa2020 on how Dyna-2 learned to untwist a real water bottle cap out of the box from just 13 m...
@chris_j_paxton: RT @RoboPapers: How can roboticists scale human manipulation data in the wild? EgoHumanoid builds a ...
@Scobleizer: A new robot scaling law?

Been hearing the same from San Francisco nerds but didn’t put it all toget...