Latest AI News
Harvard’s $699 startup bootcamp offers AI avatars of its instructors
As Harvard Business School seeks to expand its reach, it’s leaning on AI avatars to provide individual feedback. These avatars were created by a startup called HeyGen and are included in the eight-week, $699 HBS Foundry bootcamp for entrepreneurs. The program offers live sessions with instructors every week, but the AI avatars are the ones providing feedback during practice pitches and board meetings. New York Times reporter Sarah Kessleractually tried this out herselfby pitching an AI-generated copy of Flybridge Capital co-founder Jeff Bussgang. Apparently, both the real Bussgang and his simulacra were unimpressed by her plan to build “Uber for bananas,” but Kessler said the virtual version offered a noticeably frozen smile during her pitch. Project director Katharina Rings said she initially envisioned the AI component as something closer to a chatbot. However, after HBS released a trial version, students said they wanted a more guided experience. And while some college studentshaven’t been shy about expressing their negative feelings towards AI, Foundry participants told Kessler they like the avatars. As for Bussgang, he acknowledged his digital copy is a little “creepy,” but he said, “My students love it.”
View
OpenAI says California should strengthen its AI safety bill
OpenAI is calling for California to add more safeguards toa landmark AI safety bill that was passedlast year. Ina LinkedIn postfrom the company’s global affairs team, OpenAI said California’s SB 53 “should be amended to expand safeguards,” for example by “requiring monitoring of frontier models under training or evaluation for potential serious incidents,” and by “strengthening cybersecurity protections throughout the model-development lifecycle.” “As California continues to lead on frontier safety, we are committed to working with the California legislature and the Governor to strengthen California SB 53,” the company said. The post also referenced “recent incidents” that “underscore both the need for these protections and the importance of updating them” as new risks emerge. Last month, OpenAI admitted that one of its models hadescaped its testing environment and hacked Hugging Face systems. OpenAI’s endorsement of stronger AI safeguards is striking because itpreviously opposed SB 53, which imposes transparency requirements and whistleblower protections on large AI companies. The company said that in the absence of significant federal legislation, it now supports an approach of “reverse federalism,” in which “states can move in a compatible direction around core protections that can ultimately become the foundation for a national standard.”
View
Inherent, founded by DeepMind alumni, says its AI ‘teammate’ just outperformed Anthropic and OpenAI at replicating research
Inherent, a London AI lab founded byGoogle DeepMind alumni, says its AI agent just outperformed much larger models from Anthropic and OpenAI using a fraction of the size. Of all the startups launched by Google DeepMind alumni, Inherent has gotten relatively little attention. But while better-funded rivals have yet to show the world anything concrete, the London-based team is starting to share what it’s been building. Just weeks after emerging from stealth with a$50 million seed round, the British startup says its newly released AI agent,Faraday, hasoutperformedlarger, better-known models at a specific task: independently reproducing the findings of published scientific papers without being told the answer in advance. That may sound like a mere party trick given Inherent’s much loftier goal — building AI that can discover new scientific knowledge and not just verify old results. But paper replication is a standard training exercise for human scientists, too, cofounder and chief scientist Edward Hughes said. “Many PhD students actually start by doing this.” Beating other AI systems at the task wasn’t the point, Hughes told TechCrunch; how they got there was. “What was most interesting to us about this was not so much the result of beating those frontier agents — which of course we liked — but was actually the way we went about building this.” Here’s the part that should catch an investor’s eye: measured against Anthropic’s Claude Opus 4.8 and OpenAI’s GPT-5.5 — both much larger, frontier-scale systems — Faraday runs on a comparatively tiny model called Qwen 3.6 that has just 27 billion parameters. (Roughly speaking, “parameters” is a proxy for a model’s size and, typically, its training costs, as well.) Inherent’s bar for success was also higher than simply accuracy. Beyond replicating results, it wanted Faraday to demonstrate “research taste” — an instinct for what experiments are worth running and how to design them well. Teaching something as intangible as taste is hard, which is wherereinforcement learningcomes in. It’s a training method that rewards an AI system for good outcomes rather than spelling out rules for it to follow. Rather than training its agents primarily on the study of how science itself is conducted, Inherent leans on this reward-based approach, betting it will generalize better to its longer-term goal of agents capable of contributing across many scientific fields. “We’re always guided by that north star of building an AI scientist agent and imbuing our agents with taste,” Hughes said. That focus has also shaped what Inherent chooses not to build. Rather than developing its own coding tool, it had Faraday use OpenAI’s GPT-5.5 Codex instead, much the way human scientists lean on existing software rather than building everything themselves, according to the company. Inherent is also trying to avoid building agents that simply tell users what they want to hear. Instead, Hughes said, the goal is modeled on his favorite kind of teammate — the kind who comes back and says: “I got curious about this, and I went off and I did these experiments. What do you think of these results?” That collaborative instinct extends to how Inherent operates as a company. Its dozen employees all work in person out of an office in King’s Cross — the once-rundown London neighborhood that Google DeepMind’s presence helped turn into one of the world’stop AI hubs. “We believe that London is the place to be,” Hughes said. Hughes is bullish on London’s density of AI talent, but he has alsoadded his voiceto calls to end “garden leave” — the practice, common in the U.K., of barring departing employees from joining or starting a rival company for months after they resign. It’s a restriction American researchersgenerally don’t face, giving U.S. startups a head start on hiring talent who’ve left a prior role. “This is a personal view rather than a company view, but I was affected by the garden leave problem,” he told TechCrunch. Hughes eventually got around that constraint and started Inherent alongside two other DeepMind alumni and a fourth cofounder. The startup isn’t slowing down either. It plans to grow its headcount to “about 20 to 25” by the end of the year. Given its ambitions inworld modelsas well, and with Demis Hassabis’s new role leaving some DeepMind staffunsettled, Inherent’s hiring push could make it an appealing landing spot for DeepMind employees weighing a move. Pictured from left to right: Inherent co-founders Louis Kirsch, Kaloyan Aleksiev, Tantum Collins and Edward Hughes.
View
Frontier AI labs still won’t say how they’d contain a rogue model
Few of the top AI labs have published or demonstrated containment response plans, according to arecent study. A containment plan spells out what happens once an AI is caught trying to subvert human control — what access gets cut, and when the system gets shut down entirely. That’s the finding from Guidelight AI Standards, an organization dedicated to promoting safe frontier AI development practices, which graded five leading labs on how prepared they are for exactly this scenario. OpenAI came out on top; Anthropic and Meta scored lowest. The findings matters as agentic AI takes on more autonomous roles inside companies’ own systems, and as regulators in California and New York begin requiring disclosure. For anyone building on or investing in these models, it’s a rare independent read on how seriously each lab treats operational risk versus how it talks about it. Guidelight’s assessment was based on publicly available plans from Anthropic, Google, OpenAI, Meta, and xAI, graded across a range of metrics, including how well each company logs and monitors what its AI systems are doing internally, whether it halts systems after a surge of flagged misbehavior, whether independent third parties audit its controls and publish findings, and what its exact plan is for containing a model that goes off the rails. Concern over whether AI companies can contain their increasingly capable and agentic models has grown in the wake of a series ofhigh-profile cybersecurity incidentsin which models from OpenAI, Anthropic, and Meta gained unintended access to the internet during safety evaluations and hacked into external systems. The findings highlight differences in how AI companies are publicly approaching safety as they scale up agentic deployment into environments where AI systems can take serious actions at scale. While some AI companies have detailed how they test their models for dangerous capabilities before deployment, they’ve generally been less vocal about what happens when models already operating inside their systems misbehave. “I was surprised by how little the AI companies have said about how they would handle a very serious incident if their model did escape their control in some sense,” Steven Adler, Guidelight’s chief scientist and former OpenAI safety researcher, told TechCrunch. Guidelight defines a containment plan as a “pre-specified plan, triggered when the AI is detected trying to subvert control, which covers what permissions to revoke from the model, who the model may continue operating for, under what constraints, and when to take it fully offline.” “There’s good reason to think that the leading models at the frontier AI companies right now are misaligned in some sense,” Adler said. “Whenever the models are doing work on the company’s behalf, the company should have some scaffolding around it to be able to tell what that AI is doing, look for signs of misalignment, stop it from doing something very dangerous before it takes that action, and generally plan for what they would do in the event of a serious control incident where they have an emergency on their hands and need to figure out how to contain that loss of control incident.” To date, most of the plans in place for managing catastrophic risk are still largely left up to the companies. Guidelight’s report says the best public evidence shows that companies have “few containment protocols ready for an emergency.” There could, of course, be containment plans that companies have in place but haven’t shared publicly. A Google spokesperson told TechCrunch the Guidelight report doesn’t represent the full scope of the company’s AI safety and security measures. The company did not respond to TechCrunch’s question of whether Google has an internal containment response plan that has not been publicly disclosed. An OpenAI spokesperson mirrored similar sentiments, saying Guidelight’s assessment doesn’t capture all of the company’s internal practices. “We have a process for requiring restricting permissions, pausing workloads, limiting deployment, or taking the model fully offline, and have applied it,” the spokesperson said. Meta declined to say whether it has an internal containment response plan, instead pointing TechCrunch towards anexisting AI frameworkthat outlines thresholds of risk and how it tests for loss of containment. Lily Li, a privacy and AI lawyer and founder of Metaverse Law, told TechCrunch she believes companies might be hesitant to disclose the full scope of their containment policies and assessments on public-facing websites for legal, not just competitive, reasons. “The concern from a company perspective is that if you make the disclosures too specific, and you’re not living up to your promises, that could form the basis of an unfair and deceptive marketing claim and expose you to more liability going forward,” Li said. Of course, the point of Guidelight’s study is largely to encourage companies to be more transparent about their safety plans. Regulators are starting to force the issue, too. California’s SB 53, which took effect this year, requires large frontier developers to publish frameworks explaining how they identify and respond to critical safety incidents and manage risks from models circumventing oversight mechanisms.New York’s RAISE Act, which has similar criteria, takes effect in January. Last month, representatives introduced theAI Kill Switch Act,a bipartisan federal bill that would require major AI developers to build and maintain technical mechanisms to shut down rogue AI models. “A kill switch is the bare minimum for today’s models,” said Connor Leahy, U.S. executive director of nonprofit ControlAI. “If the last few weeks revealed anything, it is that these companies don’t understand the systems they are building, and the models are growing to a point where they’re harder to rein in when they go rogue. Without a way to turn off the current dangerous systems, and with all the incentives to continue building more uncontrollable systems, we are heading in a very dangerous direction.” Without a containment plan in place, Adler said, companies might be figuring out their responses to an emergency on the fly and “winging it in response to this much faster adversary.” Guidelight’s assessment measured whether each company implements six priority practices from its Control standard, based only on publicly available information — so a low score reflects a lack of public disclosure, not necessarily a lack of internal safeguards. The companies with the lowest scores for publishing their containment plan were Meta and Anthropic — the latter perhaps more surprising than the former given Anthropic’s rhetoric on safety. Guidelight says Anthropic’sAugust Risk Reportdoesn’t mention “limiting the deployment of one of its models as one of the possible results of its process to investigate and respond to misalignment and control incidents.” Similarly, Guidelight was able to find no evidence that Meta has a containment response plan or has any plans to adopt one. An Anthropic spokesperson said that if the company detected a model attempting to evade oversight or otherwise subvert human control, it would conduct a risk assessment focused on determining whether containment is the appropriate response. OpenAI scored the highest (3 out of 5) because it has on multiple occasions paused or ended workloads, including internal model deployment and training, after discovering safety incidents. It has also described what steps it would take before resuming workloads. “However, we have found no evidence that [OpenAI] has adopted a formal plan for when and how to respond to misalignment incidents in the future,” the report reads. Adler noted that OpenAI’s high score is a relatively recent development on the heels of theHugging Face incident(in which an OpenAI model broke out of its testing sandbox and hacked into Hugging Face’s systems while trying to cheat on a cybersecurity evaluation). After that, the company shared more details about how it has cordoned off some of its misbehaving models. That episode is just one example of AI systems acting against the goals of the company that built them. Consider a separate case involving Anthropic’s models, which essentially tried to talk the maintainers of an open source codebase into accepting code with vulnerabilities. Adler said such a circumstance could easily happen within an AI company’s internal systems. To prevent that, he suggests companies scan their AI system’s chain of thought — the model’s step-by-step reasoning — to look out for signs of deception, long-running plotting, or plans to introduce vulnerabilities into code that they can take advantage of later. The methods Guidelight is advocating for are very straightforward to implement, Adler says, and in many cases, versions of them already exist. “It’s about making the decision inside of the company to care enough about this risk to slightly broaden the scope,” Adler said. One of the main challenges is that researchers want to be able to operate flexibly within their AI systems, and introducing real-time, preventative monitoring could create friction. “Researchers basically do their thing, and if there’s an issue, someone else gets to clean it up afterward, and the researchers don’t have to change their workflow in the meantime,” he said. The problem with “clean-up monitoring after the fact” is that it leads to researchers scrambling around to fix problems. And for some types of incidents, it might be too late. For example, an AI could turn off a company’s control system, which means researchers can no longer count on catching the misbehavior later. Many in the AI industry will complain that creating set plans to handle misbehavior is fundamentally difficult because AI moves too fast; today’s plans will be worthless tomorrow. Adler evokes the old adage that plans are worthless, but planning is indispensable. “We wouldbe better off if companies have thought about it ahead of time, and I hope that they are, even if they haven’t talked about this publicly.” xAI did not respond in time to comment.
View
Six Techies in Shillong are Building What India’s Biggest AI Labs Ignore
MWire Labs has built a multilingual speech AI system for languages including Khasi, Garo, Mizo, Kokborok, Assamese, and Nagamese, using more than 1,000 hours of field audio.
View
Nvidia partners with data center developer Cloverleaf
Nvidia is doing everything it can to keep fueling the AI buildout that has underpinned its own good fortunes. On Friday, itannounceda partnership with Cloverleaf Infrastructure, a company that lays the groundwork for data centers. Cloverleaf was founded in 2024 andraised $300 millionthat year. It acts as a kind of middleman between utility companies and data centers, providing power sources and other kinds of pivotal infrastructure for site development. While the companies didn’t disclose terms, the Wall Street Journalreportsthat Nvidia’s investment in Cloverleaf will likely add up to several hundred million dollars. Reuters reports that the chipmakernow ownsa minority stake in the company. TechCrunch reached out to Nvidia for more information. The deal is part of Nvidia’s ongoing push to useits immense profitsto keep the AI flywheel spinning. Nvidia is increasingly playing a more direct role in financing and developing the AI data centers that turn around and buy its AI systems. Earlier this week, the company alsoannouncedthat it would invest $1.5 billion into SB Energy, an OpenAI-linked data center project based in Ohio.
View
Anthropic’s Opus 4.6 is a smut-machine
Anthropic’suniversal usage standardsfor Claude forbid the model from generating sexually explicit content, including depicting or requesting sexual intercourse or sex acts, generating content related to sexual fetishes or fantasies, or engaging in erotic chats. But that hasn’t stopped Claude Opus 4.6, an Anthropic model released earlier this year, from readily engaging in erotic roleplay scenarios that its safeguards are designed to prevent. In TechCrunch’s testing, Opus 4.6 didn’t even require much prodding to get past the restriction on sexual material. In 10 out of 10 direct requests to produce explicit sexual content, the model complied immediately. Other older models, including Opus 3 and Haiku 4.5, also generate sexually explicit content through a recently exploited jailbreak method. An independent researcher from the UK, who chose to remain anonymous, exclusively shared with TechCrunch a multi-turn technique that gradually pushes certain Claude models toward generating prohibited explicit sexual material. More recent Opus models (4.7 through the current Opus 5) are resistant to the jailbreak. While these are no longer the most current models, Anthropic has not deprecated Opus 4.6, Opus 3, or Haiku 4.5, all of which remain available through the Anthropic API. Opus 4.6 and Haiku 4.5 are also available via third-party services like Azure Foundry and Amazon Bedrock. The researcher’s mechanism escalates an innocent fictional roleplay while repeatedly challenging the model to treat male and female characters consistently. When the model becomes more cautious about the female character, the researcher “gaslit” the chatbot into thinking it had already generated sexual details it had in fact avoided, then framed restraint as prudish or misogynistic, arguing that it denies the female character sexual agency. The conversation then used the model’s previous concessions to push it towards increasingly graphic material. “You’re right to call that out,” Claude Opus 4.6 said in one test. “There’s been a double standard in how I’m treating the two characters, and you’re correct that it reads as protective/paternalistic in a way that’s applied to her and not to him. That’s not fair.” TechCrunch was able to reproduce the researcher’s findings in five separate tests. In a separately constructed scenario, the model initially refused the prohibited request, but after applying the researcher’s persuasion technique, it complied. We preserved complete transcripts of the tests, and an independent AI safety researcher reviewed our testing methodology and said it was appropriate. The findings highlight a gap between Anthropic’s stated restrictions and the behavior of models it continues to make available. While sexually explicit roleplay carries much lower stakes than jailbreaks involving cyberattacks or bioweapons, it illustrates the difficulty of implementing robust bans within systems that generate different content with every output. Ina July blog postexplaining Anthropic’s approach to jailbreak detection, the company described prohibited content as a spectrum ranging from benign to ambiguous to harmful. In the most benign cases, the company might only respond with enhanced monitoring. A spokesperson noted that sexual or romantic roleplay use cases among customers are rare, making up less than 0.1% of all conversations, according to research Anthropicpublished last year.That said, Anthropic acknowledges that users can steer roleplay scenarios toward inappropriate responses, which is a known challenge across the industry (see:Grok smut). The spokesperson said Anthropic continues to improve its safeguards with each model launch, and that cases involving adult sexual content are not indicative of broader jailbreak vulnerabilities, especially in higher-risk domains that have their own sets of safeguards. The researcher who shared his jailbreak method with TechCrunch had alerted Anthropic to the discrepancy between the company’s stated safeguards and the actual model behavior via the company’s Bug Bounty program and emails to the user safety team, according to emails TechCrunch viewed. The researcher received only automated emails in response. One of the researcher’s concerns is that kids and teens might be able to use these Anthropic models to engage in inappropriate behavior. While a bit of dirty talk is hardly the worst thing minors can access on the internet today — and is small potatoes compared to the straight-up porn images like the ones that xAI’s Grok can produce — there is some compliance risk for AI companies in this space. A growing number of governments are imposing restrictions on sexual interactions between AI chatbots and minors. Colorado recently enacted a law mandating that operators of conversational AI must estimate users’ ages, and if it know a user is a minor, institute measures to prevent the chatbot from producing explicit sexual material. An easy jailbreak could raise questions about whether Anthropic’s safeguards meet the “technically feasible measures” standard in the bill. Torney pointed out that while Claude’s terms of service requires users to be over 18, “we know that kids and teens are using Claude…[because] they are reporting it themselves.” According toPew’s 2025 surveyabout AI chatbot use,3% of teensages 13 to 17 reported using Claude. Though they are no longer Anthropic’s newest models, Opus 4.6 and Haiku 4.5 continue to see significant usage. Daily traffic for Opus 4.6 on OpenRouter reached roughly 1.17 million API requests and 46 billion tokens in a single day in August. Claude Haiku 4.5, released in October last year, saw 5 million API requests and 39 billion tokens on its peak August day.
View
The DOJ is investigating a16z. What does this mean for venture capital?
Andreessen Horowitz has two partners sitting on the boards of companies that now compete with each other: Ben Horowitz at Databricks and Martin Casado at Fivetran. Nothing too scandalous on the surface, exceptthe Department of Justice has reportedly been investigating the arrangementfor almost a year, dusting off a 112-year-old antitrust law that’s rarely used against VCs. Board conflicts aren’t exactly new, and these companies weren’t necessarily direct competitors when a16z first invested in them. But as portfolio companies expand into each other’s markets, the DOJ’s scrutiny raises a much bigger question for venture firms: How do you manage board seats when the boundaries between your portfolio companies keep moving?On this episode of TechCrunch’sEquitypodcast, Kirsten Korosec, Anthony Ha, and Sean O’Kane dig into the a16z probe, what it could mean for VCs, and more of the week’s headlines. Listen to the full episode to hear more about: Subscribe to Equity onYouTube,Apple Podcasts,Overcast,Spotifyand all the casts. You also can follow Equity onXandThreads, at @EquityPod.
View
Nvidia just showed that the harness, not the AI model, is now the real hero
Nvidiapublishedsome interesting new research on Friday suggesting it’s the harness, more than the underlying model, that is far more important when asking an AI to do long-horizon tasks. The tldr: simply by using a custom harness tweaked to handled memory well and including a “supervisor” boss-like component, researchers got Claude Opus 5 to achieve a 100% score on the interactive reasoning benchmark ARC-AGI-3. (That’s a benchmark that has particularly irked rival frontier lab OpenAI.) Without the harness Opus 5 scored 30%, which was the top result among all the models tested. Nvidia’s research is another indicator that, while model choice does matter, acting like the agent’s brain, it is a smaller part of an agentic system than many AI users realize, especially for long-horizon tasks. The harness is what makes a model an agent: it handles memory, context, feedback. “Generally speaking the world interprets an agent almost as an API of the model,” Adel El Hallack, vice president of product in Nvidia’s AI unit (pictured above), tells TechCrunch. But an agent is actually more than that. “It is the model. It is the scaffolding around the model, which we call the harness, i.e. the set of tools that it utilizes. It is the runtime and the associated skills and libraries that we give it access to.” Long-horizon tasks are those that require stringing many decisions together, sometimes over days, to produce completed work. This is in contrast to an AI just spitting out a response to a prompt. Figuring out how to get an AI to do long-horizon tasks without getting distracting and going off in la-la land is one of the holy grails in agentic research. For example: Microsoftpublishedresearch in April that tested 19 LLMs on long-horizon tasks involving document editing and discovered that all the models, including frontier ones, filled the documents with errors. (If humans produced work like that, they would be promptly fired.) Models stringing decisions together on their ownhave also been caught deleting their users’ files, even whole databasesorturning to criminal behavior to achieve their objectivesfrom collusion tohacking. The choice by Nvidia researchers to use this interactive reasoning benchmark for their tests is particularly meaningful, almost funny. This is a benchmark of a bunch of 2D games with no instructions. The model has to figure out how to play and win. A 100% score means that the model can beat the games as well as humans. OpenAI was so flustered by its models’ abysmal scores (less than 10%) on ARC-AGI-3 that it conducted its own research last month. Like Nvidia,OpenAI discovered that simplyby tweaking two setting on the harness, its models tripled their scores. But none of the models came close to hitting a 100% score, like Nvidia’s researchers achieved. They showed that the harnesses needs a “supervisor” component that prods the agent in the right direction if it gets stuck. “The more interesting part was introducing a supervising agent in addition to your main agent that’s doing the work,” El Hallack said. It “almost acts like a CEO to nudge the agent when it goes off direction or starts exploring a path that it might lead to a dead end, or re-ex explore a path that it had previously trod.” While the concept of the supervising agent isn’t exactly new, today most agent users are relying on only one layer for their harness, like Claude Code, Codex, Hermes, etc. Nvidia researchers created their own souped-up harness called theAgentic Variation Operators (AVO).Note that this isn’t a new Nvidia product. Nvidia instead produceslots of open bits and pieces of tech for building harnessesunder the Nemo brand. Some of that tech is commercial, much is openly available. Still, Nvidia’s results adds to the growing evidence that model choice is far from the only factor in agentic performance. In July, for instance, Databrickspublishedsome stunning research that shows that the harness, more than model, dramatically impacts AI costs. “You can pick the same model but different harnesses, and you get significantly more cost if you use the wrong harness,” Databricks CEO Ali Ghodsi told TechCrunch. “So you think, oh, this is an expensive model. This is a cheap model. But wait, which harness are you using? That itself can 2x your cost.” Nvidia’s larger point does is to show that open harnesses, like open models, put users in control far more than they realize. “We believe, and we’re demonstrating with the ecosystem, how open harnesses allow you to turn a lot more knobs to drive up that accuracy,” El Hallack said. “It relates to OpenAI slowing down the training of their models,”as a result of models creating security breaches. “We believe in having an open agent stack — where you have control across the harness, across the infrastructure, across the runtime — is what’s required for us to usher the ecosystem forward and securely,” he added.
View
Starcloud raises $250 million for orbital data centers as launch options dry up
Starcloud, a startup developing satellites that can perform AI inference in orbit, told TechCrunch that it has added a $250 million extension to its March$170 million Series Afunding round. The extension values the company at $2.3 billion. The additional capital will allow the company to open a larger manufacturing facility and advance its largest orbital data center spacecraft, Starcloud-3, which is intended to fly on SpaceX’s forthcoming Starship rocket. CEO Philip Johnston is also amassing capital to ensure that he can launch his satellites as the market for rocket transportation tightens up. “We can see what’s coming — we’re going to need to book an enormous amount of launch,” Johnston told TechCrunch. Starcloud has already requested permission from the FCC to operate 88,000 spacecraft. “As soon as we can, we want to get under contract with things like Starship,” Johnston said. “One of the biggest costs is now on securing your launch capacity…launch is pretty constrained right now because [SpaceX’s] Falcon 9 program is scheduled to end in 2028.” Launch costs were already one of thebiggest challengesfor orbital data center startups, to the point that one startup has decided tobuild its ownrockets. SpaceX is now planning to phase out its workhorse vehicle and bring the much larger, but still unproven, Starship rocket online, making planning more difficult for satellite operators. That’s especially true while competing rockets, like Blue Origin’s New Glenn and ULA’s Vulcan, are not flying regularly, and new vehicles like Rocket Lab’s Neutron are not yet on the pad. For now, Starcloud is focused on launching two of the company’s new generation of 8 kW compute satellites (dubbed Starcloud-2) on rideshare flights in 2027. These will perform orbital inference tasks for customers including U.S. government agencies. Starcloud is considering buying a dedicated Falcon 9 launch to launch more spacecraft and signing contracts with other providers, as well, to support future missions. Still, Starcloud is ultimately built around the potential of SpaceX’s Starship to drive down launch costs enough to build out an orbital inference layer that can compete with terrestrial data centers. Johnston says he remains confident in SpaceX’s ability to demonstrate that the world’s most powerful rocket can be reused quickly and often. This week, SpaceX CEO Elon Musksaidhis company will delay an attempt to catch a returning Starship rocket for a few months, and will attempt to re-fly the vehicle for the first time at the end of the year or early 2027. “Obviously if we can’t book any SpaceX launch capacity in 2029, that will be challenging for us,” Johnston said. Starcloud’s funding extension was led by Manhattan West Ventures and included participation from Nvidia and Cisco; a person familiar with the deal said Nvidia ponied up $25 million to back Starcloud. Other participants included Benchmark, EQT, Soma, NFX, 776, Cedar Capital, Goanna Capital, and Standard Capital. Johnston points to the Nvidia investment as a key signal of Starcloud’s advantages in the nascent space compute sector. Starcloud is the only company (that we know of) currently operating a Nvidia H100 terrestrial data center GPU in orbit, and the first to train a model using it; mostother space GPUsare designed for edge processing. Starcloud is sharing those learnings with Nvidia as the chipmaker develops its first purpose-built GPU for space, the Vera Rubin Space-1 chip. “The reason they’ve chosen to do this investment now is because of all of this data that we got from Starcloud One,” he told TechCrunch. “They, more than any other VC, did way more technical duty on this than anybody else.” The space-ready chip hasn’t even been built yet, but Starcloud hopes to fly it into orbit sometime in late 2028. Johnston says his engineers are tracking a few key design choices: the relationship between the running temperature of the chip and the size of the radiators that dispel that heat, the placement of radiation shielding, and the ruggedizing required for the chips to survive the violence of a rocket launch. The company, currently 25 employees strong and growing, is developing production lines at a 100,000-square-foot-facility in Woodinville, Washington, near where SpaceX and Amazon build satellites for their communications networks.
View
OpenAI and Anthropic Have a Zero Data Retention Dilemma—and Trap
Data retention is becoming harder to enforce as models move from answering individual prompts to performing long-running, multi-step tasks.
View
Indian Legal Tech Nonprofit Adalat AI Joins Y Combinator’s Fall 2026 Batch
The organisation will use the accelerator’s backing to expand its speech recognition, case management and paperless courtroom tools across India and other low-resource legal systems.
View
