Neocloud Together AI raises $800M, leaps to $8.3B valuation

As shipping agentic capabilities becomes table stakes among foundation model companies, Anthropic is releasing Claude Sonnet 5, a more powerful and agentic version of the lab’s midsize model. “It can make plans, use tools like browsers and terminals, and run autonomously at a level that, just a few months ago, required larger and more expensive models,” Anthropic said in a blog post. That framing mirrors what OpenAI and Google have said about their own recent releases. OpenAI’s GPT-5.6 Sol was launched in preview last week, and it is also the firm’s most agentic model yet, allowing users to split work across subagents for longer autonomous tasks. Google’s Gemini 3.5 Flash, which launched in May, was pitched as a shift from a conversational chatbot to an agentic tool that plans, builds, and iterates on real work with minimal human input. Sonnet 5’s pitch is confirmation that agentic capability is the new baseline expectation at every price tier. Now the differentiator isn’t going to be who can do agentic work best, but how cheaply they can do it and how reliably without human oversight. Sonnet 5 promises performance close to that of Opus 4.8, but for much lower costs. Starting Tuesday, Claude Sonnet 5 will be the default model for free and Pro plans, and is available for every subscription. At launch, Sonnet 5 is priced at $2 per million input tokens and $10 per million output tokens through August 31, after which the price will jump to $3 per million input tokens and $10 per million output tokens. That makes Sonnet 5 cheaper than Opus 4.8, as well as OpenAI’s GPT-5.5 and Gemini 3.1 Pro. (It’s still more expensive than Gemini 3.5 Flash.) The new model also demonstrates significant improvements over its predecessor Sonnet 4.6, released in February, on agentic performance like reasoning, tool use, software coding, and knowledge work, according to Anthropic. For example, on one benchmark, Sonnet 5 scores a 63.2% on agentic coding, compared to Opus 4.8’s 69.2% and Sonnet 4.6’s 58.1%. On a knowledge work benchmark, Sonnet 5 actually slightly outperforms Opus 4.8, which is known for winning on solving the hardest problems like making subtle judgement calls and deep research. “Opus 4.8 is still the model of choice for higher accuracy on these tasks, but Sonnet 5 provides developers with lower-priced options that are of much higher quality than what was previously available,” Anthropic says. “Between Sonnet 5 and Opus 4.8, users can adjust the effort level to find the right balance of cost and performance.” According to testers cited in the blog post, Sonnet 5 also excels at finishing complex tasks where previous model versions would have stopped short and “checks its own output without explicitly being asked.” “We handed Claude Sonnet 5 a two-part job—update Salesforce account tiers, send a launch announcement to enterprise contacts—and it finished end to end,” Daniel Shepard, a senior engineer at Zapier, said in a statement. “That used to stall halfway. For day-to-day automation, it’s a no-brainer. ” On safety, Sonnet 5 also demonstrates a lower rate of “undesirable behaviors” like cooperation with misuse and deception than its predecessor, making it safer to use in agentic contexts. It’s better at refusing malicious requests and sidestepping hijack attempts in prompt injection attacks. It also hallucinates and engages in sycophantic behavior at a lower rate than Sonet 4.6. That said, it’s not on the same level as Opus 4.8 and Claude Mythos Preview when it comes to misaligned behavior. “Evaluations also show that it has a much lower ability to perform dangerous cybersecurity tasks than our current Opus models,” reads the blog post. Lovable co-founder Fabian Hedin said in a statement that Claude Sonnet 5 “refuses unsafe requests cleanly and consistently.” “At Lovable, we’re putting powerful tools in the hands of millions of builders,” Hedin said. “A model that knows when to say no is just as important as one that knows h

As companies struggle to integrate AI, they’re increasingly ready to bring in outside help — and service providers are launching new purpose-built groups to make sure they get it. On Tuesday, Amazon Web Services launched a new internal organization for AI-focused forward-deployed engineers. Engineers on the new team will embed within companies to deploy purpose-built agents, focusing on fast engagements and customer self-sufficiency. In a post announcing the new org, AWS VP of Frontier AI Francessca Vasquez emphasized that the org would do more than build and maintain requested systems. “Customers leave AWS FDE deployments with both new solutions and new engineering capabilities,” the announcement reads. “Along with agentic systems running in their own AWS environment, they gain lasting AI skills, workflows, and patterns they can use to innovate independently.” Amazon says $1 billion will be committed to the new org, although the figure represents internal Amazon resources rather than a joint venture or conventional investment. Pioneered by Palantir, the forward-deployed engineer (FDE) model has become increasingly popular as a way to manage AI deployments. In a typical FDE system, an engineer from the contracting company (in this case, AWS) works for the client temporarily while the system is being established, allowing them to respond directly as internal opportunities or challenges emerge. In the FDE model, much of the relevant technology can be reused between deployments, while still being tailored to the specifics of each company’s needs and workflows. It also gives the client company an influx of expertise and puts primary responsibility for the deployment in the hands of the contractor. The biggest downside is the labor involved, since it means maintaining a full corps of FDE engineers to install and maintain the company’s technology. Both OpenAI and Anthropic have launched their own FDE joint ventures in recent months, valued at $4 billion and $1.5 billion, respectively. In those two cases, the AI labs were paired with private equity firms, which provided both the capital to launch and connections with client corporations in their portfolios. When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence. Russell Brandom has been covering the tech industry since 2012, with a focus on platform policy and emerging technologies. He previously worked at The Verge and Rest of World, and has written for Wired, The Awl and MIT’s Technology Review. He can be reached at russell.brandom@techcrunch.com or on Signal at 412-401-5489. View Bio

The AI-driven demand for compute power has data centers looking to squeeze more from every rack of GPUs. One consequence? Bacterial outbreaks. The liquid for liquid-cooled chips is a mixture of water and a substance that inhibits bacteria growth. To run the chips hotter, data center managers can change the mix to include more water, which absorbs heat better, but leads to nasty contamination that clogs the flow. To solve that, they flush the system, which can mean shutting down a rack for five or six hours at a potential cost of millions of dollars. Omen AI has a solution: A tiny spectrometer that can monitor that fluid health in real time, spotting bacterial growth before it becomes a massive problem. “You’re not risking huge amounts of downtime because you have no insight into what’s going on chemically,” explains CEO and founder Zach Laberge. Today, Omen AI said it raised a $31 million Series A round, led by Nava Ventures and including participation from CRV, Vanderbilt University, Mann+Hummel, Starhill Holdings, Hard Launch Capital, as well as personal investments from executives at Bridgestone, GM, Johnson Controls, and Tensorwave. Laberge founded his first company in 2020 when he was 14, raising $3 million to install sensors on construction equipment and ultimately dropping out of high school. (His father and mother, a former Minister of Education for Ontario, were supportive of his plan to carve his own path.) After that startup shut down, Laberge started Omen in 2024, with the idea of focusing on fluid systems as they key to enabling construction machinery smart enough to know when it needed to be fixed. The idea was to replace the time-consuming process of extracting samples and sending them to a lab with real-time awareness. Besides bacterial growth, the device can spot pumps and pumps wearing out if it sees copper or chromium, or seals if it sees silicon. Caterpillar dealerships were a key early customer for Omen’s heavy vehicles business, but Cat is also a major supplier of gas-powered turbines and generator to provide on-premises power for data centers. It didn’t take long for Omen to see where the wind was blowing. “That was kind of the transition,” Laberge told TechCrunch. About six months ago, “a lot of the dealerships were saying, ‘Hey, we’re starting to put sensors on our turbines, can you guys do anything on the building side of things?’” Omen discovered that those buildings are full of fluid, from their HVAC systems to their chip cooling. Spotting a new, fast-growing group of potential customers, Omen began to focus on data centers. “It’s rare to see such a young founder who has the respect of established, large corporations in a space that moves a bit more slowly,” said Cory Rellas, a partner at Nava Ventures who sits on Omen’s board. “For Omen in particular, much of our diligence came through our introductions with large customers which quickly validated their approach.” Omen, which has raised $40 million since its founding in 2024, is working with a dozen data center customers as they build out their offering, including TensorWave, a company building an AI compute cloud on AMD chips. “The fluid running through these massive systems is a critical variable that most of the industry is flying blind on,” Piotr Tomasik, TensorWave’s president, said in a statement. “Omen [sees] the future of infrastructure exactly the way we do, better monitoring to optimally support compute customers.” While many organizations rely on mailing fluid samples to labs for insight, Omen isn’t alone in developing on-premises analytics — Pyxis, an established water-monitoring firm, rolled out its data center coolant monitoring product earlier this month. The key tech advances that unlocked this approach are recent improvements in both optical technologies and signal processing software. “Hardware is just cheap enough that it makes sense to play at scale, and then signal processing lets us make more sense out of the nois

In Brief Posted: 10:29 AM PDT · June 26, 2026 Image Credits:Christopher Furlong / Getty Images Last year, hackers attacked car giant Jaguar Land Rover, one of the U.K.’s biggest employers. The hack halted production for months and made a dent in the country’s economy. The damage was so severe that the U.K. government decided to bail out the company with a £1.5 billion (around $2 billion) payment, and estimates say the hack cost the British economy $2.5 billion. For months, there was only speculation about who did it. Now, citing people close to the investigation, The New York Times reports that the hackers behind the breach were Russian, although it’s still unclear if they were working directly for Vladimir Putin’s government, were just criminals, or something in between, like criminals operating with the government’s tacit approval. Microsoft was tracking the Russian hacking group and alerted JLR to the information about the hacker’s identities, the Times reports. However sources also said that the FBI, Britain’s National Crime Agency and National Cyber Security Centre, Google’s Mandiant unit, and Palo Alto Networks all worked on the investigation. In what is a rare, but not an unprecedented occurrence in the world of cybersecurity, it turned out that the Russian hacking group was not the only one that breached some JPL networks. A Jordanian hacker who went by Rey had also broken in, according to the Times. Topics Subscribe for the industry’s biggest tech news Latest in Security

Anthropic has spent the last five years warning the world about how advanced artificial intelligence could enable mass destruction, destabilize society, and cause a litany of other grave harms. But simultaneously, it has become one of the most powerful forces pushing AI capabilities forward. The company is now among the top developers and distributors of cutting-edge AI models and courts customers like the US military. It was recently valued at almost $1 trillion.At first glance, Anthropic's stark messaging and its actions seem fundamentally at odds.But inside the company, many people don’t see a contradiction. To understand why, you first have to understand that Anthropic operates based on two core beliefs. The first is that artificial intelligence is the most transformative technology in human history, and its arrival is inevitable. The only real question is whether it leads to catastrophe or extraordinary prosperity.The second is that Anthropic believes the world will be better off if it remains at the frontier of the AI race, according to several former employees who spoke to WIRED on the condition of anonymity. Internally, leaders and employees at the company often refer to themselves as the “good guys,” meaning the ones being responsible stewards of AI technology, two of the sources said. The company sees accumulating power—whether in the form of capital, compute, research talent, or political influence—not as an end in itself, but as the price of fulfilling its mission: “to ensure the world safely makes the transition through transformative AI.”Helen Toner, executive director of Georgetown’s Center for Security and Emerging Technology and a former OpenAI board member, uses an analogy to describe Anthropic’s worldview. She compares powerful AI to a forest filled with both magical treasures and dangerous monsters. All the villagers nearby are rushing in, lured by the treasure. In her telling, Anthropic wants to venture farther into the forest than anyone else while investing heavily in taming the monsters—that is, capturing AI’s benefits while containing its catastrophic risks.“What’s distinctive about Anthropic is they’re like, ‘People are going in the forest anyway, we have to do it first.’ This is very explicitly their strategy: build cutting-edge AI in order to be a serious player at the table who can talk about what cutting-edge AI systems should look like, what risks they pose, and pushing for reasonable safeguards,” Toner tells me. “They’re very straightforward about this. It’s just a weird enough strategy that people have a hard time hearing it.”Anthropic CEO Dario Amodei outlined this approach plainly in a conversation with his cofounders posted on the company’s career page: “You have to find a way to actually be competitive, to actually lead the industry in some cases, and yet manage to do things safely,” he says. “If you can do that, the gravitational pull you exert is so great.”Anthropic was founded in 2021 by a group of former OpenAI employees who defected after losing faith in the ability of the company’s leadership—particularly CEO Sam Altman—to safely bring transformational AI into the world. That sentiment still shapes the company today. Two of the former employees I spoke with say that, in internal discussions, Anthropic executives often describe Altman and OpenAI—and, to a lesser extent, Meta and Elon Musk’s xAI—as cautionary examples that help define Anthropic’s own sense of responsibility.In many regards, Anthropic is just like any other Silicon Valley company. Many startups market themselves as David fighting the outdated, entrenched Goliaths of the industries they want to disrupt. Google, Facebook, and Apple were all founded upon idealistic principles, which later became muddied or were abandoned altogether as they became richer, larger, and more influential.But former employees say that Anthropic is unusual in how intensely it believes in its mission, and how explicitly it tells employees that tech

Amazon on Thursday said it would invest an additional $13 billion to expand its AI and cloud footprint in India through 2030. The fresh investment, announced after Amazon CEO Andy Jassy met India’s Prime Minister Narendra Modi in New Delhi, will fund the expansion of Amazon Web Services’ data center capacity in Mumbai and Hyderabad. The announcement marks Amazon’s third major commitment for India in as many years. In 2023, following a meeting between Jassy and Modi, the company said it would invest $15 billion by 2030, including $12.7 billion for Amazon Web Services. It followed that with an over $35 billion commitment in December 2025. The company’s investment commitments in the country now total $48 billion. Amazon did not detail how the total $48 billion would be deployed across its India businesses. Long-term commitments by technology companies usually include both capital and operating expenditures, rather than only new infrastructure spending. Amazon’s announcement follows a wave of investments by global technology companies that are betting that India will become a major hub for the computing infrastructure needed to power artificial intelligence products. Microsoft said in December it would invest $17.5 billion in India by 2029, and Google said in October it would spend $15 billion to build an AI hub and data center infrastructure in the country. India has also attracted billions of dollars in commitments for data center projects from investors including Australia’s AirTrunk, Canada Pension Plan Investment Board’s CPP Investments, and domestic conglomerates Reliance Industries and Adani Group. New Delhi has sought to attract more investment through policy incentives, including tax exemptions for foreign cloud providers on services sold overseas if those workloads are run from Indian data centers. Amazon is also investing in its domestic retail and logistics network. The company plans to open more than 20 fulfillment centers, and over 100 last-mile delivery stations this year, and this week it detailed plans to expand its quick-commerce service, Amazon Now, to more than 300 cities and towns in the country. The expansion comes as Amazon seeks to gain ground in India’s crowded quick commerce market, where it competes with Eternal-owned Blinkit, Swiggy’s Instamart, Zepto, and Walmart-owned Flipkart. Earlier this week, Flipkart said it plans to open 1,500 micro-fulfillment centers across the country by the end of 2026. When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence. Jagmeet covers startups, tech policy-related updates, and all other major tech-centric developments from India for TechCrunch. He previously worked as a principal correspondent at NDTV. You can contact or verify outreach from Jagmeet by emailing mail@journalistjagmeet.com. View Bio

TechCrunch’s StrictlyVC evening in Los Angeles late last week brought together two of the more straight-talking investors working in AI right now. Carter Reum is co-founder of M13, an early-stage firm with $2.5 billion in assets under management that has been a seed or Series A investor in 17 unicorns, he says. Chang Xu is a partner at Basis Set Ventures, which launched in 2017 as one of the first early-stage funds focused exclusively on AI and is now investing out of its fourth fund, with nearly $1 billion in assets under management. On stage, in a sun-filled room in El Segundo, the two were as entertaining as they were illuminating, covering how to price deals in a market that has never moved this fast, how to find companies that won’t get steamrolled by the hyperscalers, and what the SpaceX IPO is about to do to L.A. The conversation has been condensed and edited for clarity. Is there an AI infrastructure bubble? Chang Xu: There’s both a bubble and not a bubble. It’s not a bubble because we’ve never seen this type of growth curve before. ChatGPT goes from one to $40 billion in six months in terms of revenue — that’s just unprecedented growth at that scale. We have a portfolio company, Open Art, that went from $1 million to $10 million ARR in year one, and $10 million to $70 million in year two, [and it was] cash-flow positive most of that time with just 20 people. The bar for what is good growth has totally changed. When you have this possibility of compounding accelerant growth, the valuations don’t seem so crazy because you price that into the terminal value. On the other hand, if you price every single deal to that math, there’s no way that will work out well for a portfolio. So it is a paradoxical time. Carter Reum: I always laugh because we pretend like this is the first time in venture capital land, but we’ve seen this before — with cloud, with the iPhone, with the car in the 1920s, when people were worried they’d lose their jobs, and they did, and life went on. This is steeper and faster, but the same dynamic. What’s different in this cycle is that past cycles had innovators competing with innovators — Zuck versus Evan, Travis versus John Zimmer. In this cycle you have innovators competing with innovators, competing with the largest, most well-funded innovators the planet has ever seen, and competing with the ten largest tech companies on the planet. And I would argue that for the first time in history, the incumbents actually do have the advantage — the tech, the capital, the data, the talent. So as quickly as some of these companies rise, they may potentially fall. I actually find it harder to invest in a market like this. But if you get it right, you look like a genius. How do you price deals when startups are generating revenue faster than ever but it’s not clear how sustainable they are? Reum: We always do the cocktail napkin math. We were looking at a business the other day — AI software for brands. I asked: how big were the winners last cycle? Are there going to be more brands in the world? Are they willing to pay double or triple for software in this cycle? We ended up not making the investment because we couldn’t make the math check out. Xu: We stay very, very close to what is the defensible technical differentiation, because that frontier changes every quarter, maybe every month, sometimes every week. The framework we think about is investing below the AI and above the AI. Below the AI, you have all this infrastructure that’s getting rethought — databases, version control, deployment tools — because they were all built for humans. Now you have agents using all this infrastructure, and agents require fundamentally different things. Last year I would never have thought you’d need a new GitHub. This year I can count on two hands how many really strong teams are going after being the GitHub for agents. Above the AI, when things get super crowded, we always go back to: what is defensible, and what has lon

Anthropic is introducing Claude Tag in research preview, an “always-on Claude” that lives in Slack and acts as an AI teammate. The new feature — which allows users to tag @Claude to provide insights in chats and assign tasks — will begin in research preview, available through Slack for Claude Enterprise and Claude Team customers. Claude Tag an evolution of several integrations that already exist. Users can already DM @Claude within Slack or tag it in channels for on-demand help, and Claude Code in Slack routes coding tasks from channel mentions to full coding sessions on the web, posting updates back into the thread. But Claude Tag adds a layer of persistent context and memory that would be difficult to maintain with previous tools. “As Claude follows along with its channel, it learns ever more about the work,” reads a statement from Anthropic. “Claude can also automatically gather facts from elsewhere in the organization, if it’s granted permission to read other channels.” With Claude Tag, everyone in a given Slack channel can access a single Claude identity, meaning “anyone can see what Claude has been working on, and can pick up the conversation from where the last person left off.” System administrators will specify which tools, information, and channels Claude can access, and each Claude identity will stay scoped to whichever channels the admins define, so that a Claude set up for legal work can’t seed memories into the engineering channel, for example. When assigned a specific task, Claude Tag will break down the task into stages and works through them using whichever tools it has access to, responding in a Slack thread with what it has created. But Claude Tag also features an ambient mode that proactively jumps into the chat of its own accord to keep your team updated, flag things from across the organization, and follow up on threads or tasks that have been forgotten. Anthropic says this makes it feel like you’re “working with a real colleague — one that can produce work in public view, with far greater context and understanding than before.” That context is an increasingly critical part of enterprise deployments, and Anthropic isn’t the only company focused on it. Microsoft also has Graph, expressed through Copilot and Work IQ. Snowflake and Databricks are positioning their platforms as the back end support containing tacit organizational knowledge that agents can tap into. Glean is also building an intelligence layer that understands company context and sits between the model and the enterprise data. When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence. Rebecca Bellan is a senior reporter at TechCrunch where she covers the business, policy, and emerging trends shaping artificial intelligence. Her work has also appeared in Forbes, Bloomberg, The Atlantic, The Daily Beast, and other publications. You can contact or verify outreach from Rebecca by emailing rebecca.bellan@techcrunch.com or via encrypted message at rebeccabellan.491 on Signal. View Bio