Y Combinator’s Summer 2026 Request for Startups lists 15 categories, eight of which require hardware or capital, including agriculture robots, counter-drone defence, space inference chips, lunar manufacturing, and semiconductor supply chain software. The document represents the most dramatic pivot in YC’s public investment thesis, signalling that the accelerator which built its reputation on software now believes the next decade of billion-dollar outcomes will come from AI applied to physical, regulated, and capital-intensive industries.
Y Combinator published its Summer 2026 Request for Startups in late April, just days before the application deadline. The document lists 15 categories of companies that YC’s partners want to fund. Eight of them require capital, hardware, or both. The list includes AI for low-pesticide agriculture, counter-swarm drone defence, inference chips for space, lunar manufacturing from molten regolith, and semiconductor supply chain software for a process that crosses a dozen countries and takes five months to complete.
Advertisement
Each category is written by a named partner, and each reads less like a startup prompt than a thesis on why the economics of a particular industry have just shifted. The most influential startup accelerator in the world, the institution that funded Airbnb, Stripe, and Dropbox, is telling founders that the next decade of billion-dollar outcomes will come not from building software but from using AI to enter the physical, regulated, and capital-intensive industries that software alone never touched.
The thesis
The RFS opens with Garry Tan, YC’s chief executive, writing about agriculture. AI can now identify individual weeds and pests in real time, he argues, and when combined with robotic precision treatments, the result is farming that uses dramatically less pesticide while improving yield. The category is not agtech in the way Silicon Valley has historically understood it, which meant software dashboards for farm management. It is agtech that involves building physical robots, training vision models on biological data, and deploying hardware in fields.
Tyler Bosmeny’s entry on counter-swarm defence compares the companies he wants to fund to Cloudflare rather than Raytheon, software-defined defence systems that neutralise drone swarms at a fraction of the cost of traditional missile systems. The United States Department of Defense proposed more than $70 billion for drone and counter-drone systems in its latest spending plan, and defence tech is experiencing its strongest investment cycle in decades. Adi Oltean asks for founders who will 3D-print structures from molten lunar regolith and extract raw materials including silicon, aluminium, iron, and titanium through electrolysis on the moon.
The hard-tech categories are not aspirational filler. They reflect a structural change in what venture capital is willing to fund. Defence tech startups raised a record $49.1 billion in 2025, nearly double the prior year. Anduril, the autonomous weapons company, raised $4 billion at a $60 billion valuation in March. SpaceX has demonstrated that hardware-intensive businesses can produce venture-scale returns. The old assumption that hardware could not generate the margins or the speed that venture capital requires has collapsed, and YC’s RFS is the clearest institutional acknowledgement that the collapse is permanent.
The software that remains
Seven of the 15 categories are software, but none of them resemble the SaaS playbook that defined the previous decade. The category YC calls Software for Agents asks founders to rebuild every major software category for a world where the next trillion users are not people but AI agents. That means APIs, machine-readable documentation, command-line interfaces, identity systems, permissions layers, and payment infrastructure designed for autonomous programmes rather than human beings. Google rebranded its entire AI platform around agents at Cloud Next 2026, consolidating Vertex AI into the Gemini Enterprise Agent Platform and launching a $750 million fund to finance agentic deployments. Gartner predicts that 40 per cent of enterprise applications will include task-specific AI agents by the end of this year, up from less than 5 per cent in 2025.
Advertisement
The Company Brain category asks for a system that pulls knowledge out of every fragmented source inside a company, structures it, keeps it current, and turns it into what YC calls an executable skills file for AI. This is not enterprise search. It is a living map of how a company works: how refunds are processed, how pricing exceptions are decided, how engineers respond to incidents. The Dynamic Software Interfaces category is its mirror image, asking founders to rebuild software so that agents can operate it natively rather than scraping interfaces built for humans.
The SaaS Challengers category names the targets explicitly: ERP, chip design software, industrial control systems, and supply chain management. These are the categories where incumbent vendors charge the most and innovate the least, and where AI-native replacements could capture enormous markets if they can clear the switching costs.
The physical turn
The RFS entry on semiconductor supply chains may be the most revealing. A single advanced AI chip goes through approximately 1,400 process steps, crosses a dozen countries, and takes five months to manufacture. That supply chain is managed, as the RFS puts it, with spreadsheets, SAP, and phone calls. Diana Hu, the YC partner who wrote the entry, is asking for founders who will replace that infrastructure with software that can track, optimise, and predict across the most complex manufacturing process on earth.
The category sits at the intersection of every force currently reshaping the technology industry: the US-China chip export controls, the reshoring of semiconductor fabrication, the explosion in AI chip demand, and the geopolitical fragility of supply lines that route critical components through Taiwan, South Korea, the Netherlands, and Japan.
Advertisement
The space categories are similarly grounded in economics rather than aspiration. Reusable rockets from SpaceX and Stoke Space are about to produce a massive increase in the capacity to put objects in orbit, which means an equally massive increase in demand for the electronics that operate there.
YC wants inference chips optimised for mass, thermal performance, and radiation hardness. SpaceX and Blue Origin are already racing to put data centres in orbit, and the AI hardware that runs inference workloads in terrestrial data centres does not survive the thermal and radiation environment of space. The market for space-rated inference silicon does not exist yet. YC is betting that it will.
What changed
Y Combinator’s Spring 2026 RFS, published just three months earlier, listed eight categories. The Summer edition nearly doubled that to 15. The Spring list included AI for product management, government AI, AI-native hedge funds, and stablecoins. Those are recognisably software businesses with AI bolted on. The Summer list includes lunar regolith manufacturing and counter-drone defence systems. The shift between the two documents is the most dramatic reorientation in YC’s public investment thesis since the accelerator began publishing requests for startups.
The change reflects what has happened to venture capital more broadly. In the first quarter of 2026, $297 billion flowed into startups globally, 2.5 times the prior quarter and the most venture funding ever recorded in a three-month period. Accel raised a $5 billion fund on the back of returns from Anthropic and Cursor.
Advertisement
Andreessen Horowitz raised $15 billion. Thrive Capital closed more than $10 billion. The money is not looking for the next enterprise SaaS dashboard. It is looking for the companies that will apply AI to the industries where the margins are highest, the incumbents are slowest, and the barriers to entry have historically been physical rather than digital. YC’s RFS is the most explicit version of that thesis because it names the industries by name: agriculture, defence, space, semiconductors, medicine, manufacturing.
The stablecoin category, one of the few holdovers from the Spring list, reveals a different kind of ambition. YC describes stablecoins as sitting between the regulated and unregulated worlds, creating room for services that combine the strengths of both: yield-bearing accounts, tokenised real-world assets, and infrastructure that moves money faster and cheaper across borders. The AI Personalised Medicine category asks for agents that analyse genomic data, electronic health records, and wearable output to generate patient-specific treatment protocols rather than population-level guidelines. Neither category requires building physical hardware. Both require operating in industries where regulation, liability, and institutional trust are the barriers, not code.
The signal
YC’s Request for Startups is not a prediction. It is a commitment. The partners who write the entries are the partners who will evaluate the applications, and the categories they describe are the companies they will fund. When Garry Tan writes about agriculture robots and Tyler Bosmeny writes about counter-drone systems and Adi Oltean writes about 3D-printing on the moon, they are telling founders what the next YC batch will look like. The document is the closest thing the startup ecosystem produces to a forward-looking investment mandate from its most influential institution.
The mandate says that software is now the substrate, not the moat. The models are commoditising. The infrastructure is scaling. The interfaces are being rebuilt for agents. What remains scarce is the ability to apply that substrate to the physical world: to build the robot that replaces the pesticide, the chip that survives the radiation, the defence system that costs less than the drone it destroys, the supply chain software that tracks 1,400 process steps across 12 countries, the molecular model that designs a drug for a target the industry called undruggable.
Advertisement
Y Combinator built its reputation by funding two founders in a garage writing code. Its Summer 2026 RFS is a document that says the garage is no longer enough. The founders it wants now are the ones who can write the code and then build the thing.
Hacktivists have claimed responsibility for taking down the public-facing infrastructure of popular Linux operating system distribution Ubuntu, as well as Canonical, the company that develops and maintains the software. The attack began on Thursday, and affected services that Ubuntu users rely on.
“Canonical’s web infrastructure is under a sustained, cross-border attack and we are working to address it. We will provide more information in our official channels as soon as we are able to,” the company said on its website.
The hacktivists are believed to have launched a distributed denial-of-service, or DDoS, a crude but often effective attack that consists of flooding a target with junk traffic until it overloads or crashes.
Ubuntu developers have been discussing the attack on an unofficial Ubuntu community forum, claiming that the attack affects Ubuntu’s security API, and several Ubuntu and Canonical websites. According to a post on a threat intelligence forum, the DDoS attack has also made it impossible for users to update and install Ubuntu. TechCrunch verified that updates failed to install on a test device running Ubuntu.
Advertisement
As of this writing, the outage has been ongoing for around 20 hours.
When contacted, Canonical spokesperson Lelanie de Roubaix reiterated what the company said on is website.
Hacktivists calling themselves The Islamic Cyber Resistance in Iraq 313 Team claimed on its Telegram channel that it was to blame for the DDoS attack.
Techcrunch event
Advertisement
San Francisco, CA | October 13-15, 2026
The hackers claimed to be using Beamed, a DDoS-for-hire service. These types of services, also called booters or stressers, allow anyone to pay to launch DDoS attacks, even if they have no technical skills nor the necessary infrastructure to flood targets with bogus traffic. The DDoS-for-hire service in this case claims to power attacks in excess of 3.5 Tbps, which is about half of the bandwidth of a cyberattack that Cloudflare last year called the “largest DDoS attack ever recorded.”
Advertisement
For years, authorities such as the FBI and Europol have played a game of whack-a-mole against these services, taking downandseizing domains, and sometimes arresting the people behind them.
This story was updated to include Canonical’s response.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Steam on Linux use in March “had skyrocketed to 5.33%…” reports Phoronix, “easily the highest level we’ve seen Steam on Linux at since its inception more than a decade ago.”
“They don’t make them like they used to” is a phrase that can apply to just about anything. For gear-heads feeling nostalgic about older cars, it’s a phrase that never seems to go away. This is especially true when comparing today’s cookie-cutter engines versus the selection that drivers had in ridiculously overpowered vintage cars. Part of the reason engines have changed and choices have been reduced is due to U.S. EPA standards.
Those standards are set by the Clean Air Act, which gives the U.S. Environmental Protection Agency authority to regulate vehicle emissions. This is done through strict federal requirements that directly influence vehicle design and engine development. As a result, car manufacturers are pushed to produce a more limited range of engine types. Corporate Average Fuel Economy (CAFE) standards are also in play. CAFE requires automakers to meet fleet-wide efficiency targets, which leads to shared engine designs being used across an entire lineup of vehicles.
As automakers worked to satisfy these standards, modern advances like turbocharging and fuel system improvements allowed for engine downsizing. This means smaller engines can produce performance similar to that of larger engines. In fact, there are even small engines with more power than muscle car V8s. So thanks to today’s technology, car manufacturers do not necessarily need to design multiple engine types when fewer can cover the same performance requirements.
Multiple large-displacement engine types were once the norm in the automobile industry. In fact, these engines were in demand for a variety of different vehicles, like old school muscle cars. This includes the big block V8 engine, which was once a major focus for automakers. It was a standard approach taken by many manufacturers, who were unrestricted by emissions and fuel economy regulations.
There is a common belief that smaller engines get better fuel efficiency than larger engines. After all, those older V8s could get very thirsty, which means you’d be filling up quite often. But fuel economy involves a lot more than just engine size. It’s influenced by several factors, like vehicle weight, transmission, technology differences, and even driving habits. So even if you have a car with a larger engine, it doesn’t mean you’re not getting good fuel efficiency.
Advertisement
There are still some U.S. automakers that give you options, depending on the vehicle. But those options are often restricted to the same model, and not widespread across the board. For example, Ford offers multiple engine choices within the F-150 lineup for 2026, ranging from a 2.7L EcoBoost V6 with 325 horsepower, up to a 3.5L High Output EcoBoost V6 with 450 horsepower. So if you’re interested in finding a car or truck with a bigger engine, it’s a good idea to check the manufacturer’s website first and then go from there.
A new attack type, dubbed ConsentFix v3, has been circulating on hacker forums as an improved technique that automates attacks against Microsoft Azure.
The first version of ConsentFix was presented by Push Security last December as a variation of ClickFix for OAuth phishing attacks, which tricks victims into completing a legitimate Microsoft login flow via the Azure CLI.
Using social engineering, the attacker fooled victims into pasting a localhost URL containing an OAuth authorization code that can be used to obtain tokens and hijack the account without passwords, despite multi-factor authentication (MFA).
ConsentFix v2 was developed by researcher John Hammond as a refined version of Push’s original, replacing manual copy/paste with drag-and-drop of the localhost URL, making the phishing flow smoother and more convincing.
Advertisement
ConsentFix v3 preserves the core idea of abusing the OAuth2 authorization code flow and targeting first-party Microsoft apps that are pre-trusted and pre-consented.
However, it brings an improvement by incorporating automation and scalability.
ConsentFix v3 attack flow
According to information retrieved from hacker forums where the new technique is promoted, the attack begins by verifying the presence of Azure in the target environment by checking for valid tenant IDs.
This is followed by gathering employee details such as names, roles, and email addresses to support impersonation.
Advertisement
Next, the attackers create multiple accounts across services such as Outlook, Tutanota, Cloudflare, DocSend, Hunter.io, and Pipedream to support phishing, hosting, data gathering, and exfiltration operations.
Push Security researchers explain that Pipedream, a free-to-use serverless integration platform, plays a central part in automating the attack, serving three critical roles:
Is the webhook endpoint that receives the victim’s authorization code
It is the automation engine that immediately exchanges that code for a refresh token via Microsoft’s API
It is the central collector that makes captured tokens available to us in real time.
Creating the Pipedream model Source: Push Security
In the next phase, the attacker deploys a phishing page hosted on Cloudflare Pages that mimics a legitimate Microsoft/Azure interface and initiates a real OAuth flow through Microsoft’s login endpoint.
When the victim interacts with the page, they are redirected to a localhost URL containing an OAuth authorization code, which they are tricked into pasting or dragging back into the phishing page.
This enables the data exfiltration pipeline, in which the page sends the captured URL to a Pipedream webhook, and the backend automation immediately exchanges the authorization code for tokens.
Advertisement
The phishing emails can be highly personalized, generated from harvested data, and feature malicious links embedded inside a PDF hosted on DocSend to improve credibility and bypass spam filtering.
In the post-exploitation stage, the obtained tokens are imported into Specter Portal, allowing the attacker to interact with compromised Microsoft environments and access resources permitted by the token, such as email, files, and other services tied to the account.
Push Security noted that its testing of ConsentFix v3 relied on its personal Microsoft accounts; as a result, it is difficult to fully appreciate the impact, which depends on permissions, services, and tenant settings, among other factors.
In terms of mitigating ConsentFix risks, Push notes that the endeavor is complicated because trust in first-party apps is architectural, and that Family of Client IDs (FOCI), Microsoft applications that share permissions and refresh tokens, is useful otherwise.
However, there are still steps administrators can take, such as applying token binding to trusted devices, setting up behavioral detection rules, and applying app authentication restrictions.
Advertisement
While ConsentFix attacks are used in actual campaigns, it is unclear if the v3 variant has gained any traction among cybercriminals yet.
AI chained four zero-days into one exploit that bypassed both renderer and OS sandboxes. A wave of new exploits is coming.
At the Autonomous Validation Summit (May 12 & 14), see how autonomous, context-rich validation finds what’s exploitable, proves controls hold, and closes the remediation loop.
A garlic-herb salmon with risotto was probably the best among the family meals I tried. The chopped asparagus was less than visually appealing when drizzled in garlic butter, but still tasty and a bit crisp. The salmon was tender and flaky. And the sweet pea risotto had no choice but to be delicious. There was so much cheese, butter, and lemon it was pretty much a concert of fats and acid.
That chicken parm was likewise a mountain of cheese and salt. It reminded me, pleasantly, of countless family meals I had as a child in the 1980s: cheese-topped chicken, garlic bread, shells stuffed with ricotta and topped with even more cheese. The big difference is that there is simply no way my mother would have cooked this meal without a vegetable.
Toval app via Matthew Korfhage
And nutrition is where Toval runs aground a little. The nutritional notes on that chicken parm meal betray 2,300 milligrams of sodium per serving, pretty much the entire daily allowance for an adult human. This is also on par with comparable servings of Stouffer’s meat lasagna. The Tovala meal also carried about 10 times the cholesterol as Stouffer’s.
Many other meals followed a similar pattern, loading up on fats and salt in order to make meals tasty. The net effect is that it’s a lot more like rich restaurant food than what most people prepare at home. Whether this is a good or a bad quality is up to you.
Advertisement
Only one meal of the seven I tried failed utterly: I flagged a teriyaki chicken dinner to my editor as a possible cultural crime against Japan. The meal was sweet soy drenching pale and steaming chicken, with an implausible side of thick egg rolls and some loose, unseasoned broccoli. It felt like the “Japanese” food you’d get at a mall food court in the ’90s. But again, this was a rare major misstep.
A more pernicious issue, in meals designed for the whole family, is the near-universal high-fat, cholesterol, and sodium content. Many with the income and inclination to eat hearty, low-effort meals like the ones from Tovala are either parents with children, or people in the retirement bracket. Each has their own reason to desire a little more nutrition, and less fat and salt.
By the end of a couple of weeks of testing recipes, I’ll admit I felt a little relieved. I was grateful to feel my arteries slowly reopen. Tovala’s culinary model makes a lot of sense to me, as a smart way of splitting the difference between prepared meals and fresh food. And the company has proven it can cook well. It might be nice if they’d also cook a diet that felt more sustainable.
Writing an email is already one of the more lifeless parts of modern work, so of course the tech industry decided to automate it. AI was meant to ease workloads by managing “grunt” work—dealing repetitive junk, trimming down inbox overload, and giving people their time back. It really sounded like the right idea. But in reality, we are nowhere close to removing the misery of email.
The kind of email you’re already sick of seeing
AI lowers the effort required to produce corporate-sounding language. That means every “just following up,” every “circling back,” every “gentle reminder,” and every “happy to connect” becomes even easier to generate and even harder to escape.
Apple
A person who might have skipped sending a pointless email before can now ask AI to draft one in seconds. And the person replying might once have wrapped things up in two short sentences. Now there is always a cleaner, longer, more “professional” version waiting from a chatbot. The Guardian recently reported on worker frustration around AI-generated workplace output, including what some employees now call “workslop.”
AI just gave bad email habits some steroids
Email was never only about communication. It also became a way to signal responsiveness, usefulness, and motion. A fast reply, a full calendar, and a long thread make things look more productive, even when nobody actually needed any of it. AI slides neatly into this culture. It can answer faster, summarize faster, schedule faster, and keep the illusion of progress running all day.
Office email already rewards performance as much as usefulness. Now every half-formed thought can become a polished paragraph. Sentences can be improved, and low-value updates can be padded into something more formal, diplomatic, corporate, and even lifeless. Using AI does not make your communication any better. What you’re getting instead is just more of it. Your inbox has more messages, fillers, and new language designed to sound productive without necessarily being useful.
Things get worse when everyone starts doing it, compounding the issue. One person sends a slick AI-polished email. The reply comes back with its own AI-assisted phrasing. Someone added to the thread later uses AI to summarize the whole exchange before sending another response. And now you have a conversation that technically keeps moving, but feels less and less human with every pass.
So who’s talking to whom?
At that point, bots emailing bots does not sound like a joke anymore. Dedicated tools like AI email assistants and scheduling bots may be useful in isolation, but they are still part of the same problem. Tools like Read AI’s Ada can handle meeting logistics and participate in email threads, which makes the whole “AI talking to AI” scenario feel a lot less ridiculous now.
It started with people leaning on AI for one harmless email, which quickly steamrolled into the whole culture of email becoming even more bloated and more performative. We were supposed to get relief from one of the most draining parts of digital work. And now it feels like new technology is just keeping that machine running rather than getting rid of it.
After landing agreements with Google, SpaceX, and OpenAI, the U.S. Defense Department said on Friday that it has signed deals with Nvidia, Microsoft, Amazon Web Services, and Reflection AI that allow it to deploy their AI tech and models on its classified networks for “lawful operational use.”
“These agreements accelerate the transformation toward establishing the United States military as an AI-first fighting force and will strengthen our warfighters’ ability to maintain decision superiority across all domains of warfare,” the statement reads.
The deals come as the U.S. Department of Defense has accelerated its diversification of AI vendors in the wake of its controversial dispute with Anthropic over usage terms of its AI models. The Pentagon wanted unrestricted use of Anthropic’s AI tools, but the AI lab insisted on guardrails to prevent Anthropic’s tech from being used for domestic mass surveillance and autonomous weapons.
The two are fighting it out in court at the moment, though Anthropic in March won an injunction against the Pentagon’s move to brand the company a “supply-chain risk.”
Advertisement
“The Department will continue to build an architecture that prevents AI vendor lock-in and ensures long-term flexibility for the Joint Force,” the statement reads. “Access to a diverse suite of AI capabilities from across the resilient American technology stack will give warfighters the tools they need to act with confidence and safeguard the nation against any threat.”
The DOD said the companies’ AI hardware and models will be deployed on Impact Level 6 (IL6) and Impact Level 7 (IL7) environments to “streamline data synthesis, elevate situational understanding, and augment warfighter decision-making.” IL6 and IL7 are high-level security classifications for data and information systems that are deemed critical to national security and require that these systems be protected physically, through strict access controls and audits.
The Pentagon said more than 1.3 million DOD personnel have so far used its secure enterprise platform for generative AI, GenAI.mil, which provides access to large language models (LLMs) and other AI tools within government-approved cloud environments. It is designed to help primarily with non-classified tasks like research, document drafting, and data analysis.
Techcrunch event
Advertisement
San Francisco, CA | October 13-15, 2026
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Europe should be known for BottleCap AI, not bottle cap memes. With its tongue-in-cheek name, this Prague-based AI startup is one of the teams that VCs think you should know.
It is not that European startups never cut through the noise — Lovable and Mistral AI are proof of it. But there are many more that don’t have nine digits in annual recurring revenue yet and that insiders are still tracking very closely.
That’s where this list comes in. Over the last few weeks, we asked investors at some of Europe’s best known venture funds to recommend two startups each: one from their portfolio (because they liked the startup well enough to invest) and one outside of it (because they are the startup experts but can’t invest in them all). We also threw in a few picks of our own.
From pre-launch to unicorn, these startups are at different stages in their journey, and from different sectors. Due to our methodology, they may not reflect where the region’s hottest hubs are, but they do reflect how deep tech talent could help Europe play its own cards in the AI race.
Advertisement
Alta Ares
Recommended by Julien Codorniou, general partner, 20VC.
What it does: Alta Ares develops AI-powered counter-drone systems.
Why it’s worth watching: Defense tech has gone from pariah to trending, particularly in Europe, where the war in Ukraine was a wake-up call for armies to modernize. Alta Ares’ interceptors answer a need for cheaper solutions to detect and fight drone incursions.
Techcrunch event
Advertisement
San Francisco, CA | October 13-15, 2026
Apron
Recommended by Jan Hammer, partner, Index Ventures (investor).
Why it’s worth watching: SMBs can be a lucrative segment for fintech companies; business owners are willing to spend at least some money to save time, and there are millions of them.
Botify
Recommended by Claire Houry, general partner, Ventech (investor).
What it does: Botify helps brands increase their visibility in AI searches.
Advertisement
Why it’s worth watching: Companies are still scrambling to replace SEO with generative engine optimization (GEO) — but this Disrupt NY 2016 alum has already embraced the shift. Botify has competitors in its new field, such as Otterly.AI and Profound, but also big customers, from Macy’s to The New York Times.
BottleCap AI
Recommended by Julien Codorniou, general partner, 20VC (investor).
What it does: BottleCap AI develops efficiency-focused foundational LLMs and apps.
Why it’s worth watching: With a founding trio that includes an entrepreneur who sold his previous company to Meta and two AI researchers, BottleCap adopted a dual approach. The startup is building its own models and releasing apps built on top of them, including Pulse, an AI-powered news app.
Advertisement
Cailabs
Recommended by Flavia Levi, investment manager, Join Capital.
What it does: Cailabs develops photonics for aerospace, defense, and industrial applications.
Why it’s worth watching: Cailabs is based on advanced research on the science of light, which it now applies to faster and more robust data transmission. Backed by public and private investors, it plans to deploy 50 optical ground stations to support growing demand for laser communications with satellites.
Cailabs’ turnkey optical ground station.Image Credits:Cailabs
Cala
Recommended by TechCrunch’s Anna Heim.
What it does: Knowledge graph for AI agents.
Advertisement
Why it’s worth watching: Cala plans to build the knowledge layer that AI agents are missing. Its founder is Elisenda Bou-Balust, a high-profile Spanish entrepreneur and AI expert who sold her previous company Vilynx to Apple in 2020.
Flower
Recommended by Pär-Jörgen Pärson, partner, Northzone (investor).
What it does: Renewable energy management.
Why it’s worth watching: Wind and solar energy are inherently variable. Flower leverages AI and battery energy storage systems to make their use more predictable. This Swedish company also recently raised over $60 million in bonds to keep on scaling.
Advertisement
Fundamental
Recommended by Jonathan Userovici, general partner, Headline (investor).
What it does: Foundation AI for big data analysis.
Why it’s worth watching: Fundamental’s foundation model, Nexus, focuses on helping enterprises draw insights from their data. The company just emerged from stealth in February, but it is already valued at $1.4 billion following a $255 million Series A.
Gradium
Recommended by Jonathan Userovici, general partner, Headline.
Advertisement
What it does: AI voice models.
Why it’s worth watching: Gradium’s AI models can be used for real-time text-to-speech that gives AI agents a voice in multiple languages. A spinout of French AI lab Kyutai, this ElevenLabs challenger raised a $70 million seed round of its own.
HappyRobot
Recommended by Pablo Ventura, general partner, Kfund.
What it does: AI agents for complex use cases.
Advertisement
Why it’s worth watching: HappyRobot, a startup backed by a16z and Y Combinator, is one of many building AI agents, but its focus is on making sure that these can be deployed and deliver ROI. It is headquartered in the U.S., but its three co-founders and part of its team are Spanish.
Inbolt AI robot in deployment.Image Credits:Inbolt
Inbolt
Recommended by Claire Houry, general partner, Ventech.
What it does: Physical AI for factories.
Why it’s worth watching: Mixing AI and robotics, Inbolt improves and expands automation in manufacturing, from the automotive industry and electronics to home goods production lines. The startup says it is already active in more than 70 factories.
Legora
Recommended by Pär-Jörgen Pärson, partner, Northzone.
Recommended by Floriane de Maupeou, principal, Serena Data Ventures.
What it does: AI training data infrastructure.
Advertisement
Why it’s worth watching: “Every strong model starts with great data,” Macrodata Labs claims on its “coming soon” landing page. But the startup won’t build that data; its upcoming platform will provide other companies with tooling to create solid training datasets.
Multiverse Computing
Recommended by TechCrunch’s Julie Bort.
What it does: Offers compressed versions of open weight models like OpenAI, Meta, DeepSeek, and Mistral AI.
Why it’s worth watching:Multiverse Computing‘s tech takes a proven model and makes it smaller and less expensive to operate, especially on a company’s own hardware. Co-founded by CTO Román Orús, a professor at the Donostia International Physics Center, the Spanish startup has raised $250 million.
Advertisement
Optics11
Recommended by Flavia Levi, investment manager, Join Capital (investor).
What it does: Fiber-optic sensing systems.
Why it’s worth watching: Optics11’s technology makes it possible to monitor equipment underwater and in similarly harsh conditions. Its potential in preventing disruptions to subsea infrastructure and energy grids helped the startup secure venture debt from the European Investment Bank.
Pennylane
Recommended by Jan Hammer, partner, Index Ventures.
Advertisement
What it does: Finance management platform for SMBs.
Why it’s worth watching: Pennylane started out with accounting, but it has bigger plans. Like many other growth-stage fintechs, this French unicorn has expanded its scope, with the ambition to build a unified financial operating system for SMBs in Europe.
PLD Space
Recommended by TechCrunch’s Anna Heim.
What it does: Launches rockets.
Advertisement
Why it’s worth watching: PLD Space is part of Europe’s push for space autonomy. After successfully launching a suborbital rocket in 2023, it is currently developing a reusable orbital launcher for small satellites. Last month, the Spanish company secured a $209 million Series C round led by Mitsubishi Electric that brought its funding to more than $350 million.
PLD Space’s MIURA 1 space rocket during its presentation in Madrid in 2021.Image Credits:Eduardo Parra / Europa Press via Getty Images / Getty Images
Proxima Fusion
Recommended by Daria Saharova, general partner, World Fund.
Recommended by Floriane de Maupeou, principal, Serena Data Ventures (investor).
Advertisement
What it does: Software for AI model deployment on advanced chips.
Why it’s worth watching: University spinout Roofline bridges the gap between AI and an increasingly fragmented hardware layer with software that lets users deploy models efficiently on different types of chips.
Space Forge
Recommended by Daria Saharova, general partner, World Fund (investor).
What it does: Space Forge manufactures semiconductor components in space.
Advertisement
Why it’s worth watching: In-space manufacturing is on the rise — for pharmaceutical applications and for chips, which are Space Forge’s focus. With extra tailwinds from geopolitics, the startup is already forging ahead: It recently generated plasma in low Earth orbit.
Theker
Recommended by Pablo Ventura, general partner, Kfund (investor).
What it does: Robots as a service.
Why it’s worth watching: Theker is one of several startups backed by Zara owner Inditex through a dedicated fund managed by Mundi Ventures. Theker’s AI-enabled robots could help the retail giant improve its logistics, but the startup is also pursuing use cases in waste management and food and beverage production.
Advertisement
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Instructure, the company behind the widely used Canvas learning platform, has disclosed that it recently suffered a cybersecurity incident and is now investigating its impact.
The U.S.-based education technology company is best known for developing Canvas, a widely used learning management system that helps schools, universities, and organizations manage coursework, assignments, and online learning.
“Instructure recently experienced a cybersecurity incident perpetrated by a criminal threat actor. We are actively investigating this incident with the help of outside forensics experts,” reads a statement from Steve Proud, Chief Security Officer.
“We are working quickly to understand the extent of the incident and actively taking steps to minimize its impact. Maintaining your trust is our highest priority, and we are committed to transparency throughout this process.”
Advertisement
Instructure says that it will provide new information regarding its investigation as it becomes available.
Since May 1, some services, including Canvas Data 2 and Canvas Beta, have been under maintenance, with customers warned they may experience issues with tools that rely on API keys.
The company has not stated whether this maintenance is related to the security incident.
BleepingComputer contacted Instructure earlier today with questions about the incident, but has not received a response.
Advertisement
BleepingComputer previously published and retracted an earlier report about this incident after determining it was based on incorrect information from a prior disclosure.
Targeting education technology firms
Threat actors have increasingly targeted education technology firms due to the large amounts of personal information they hold on students and teachers.
AI chained four zero-days into one exploit that bypassed both renderer and OS sandboxes. A wave of new exploits is coming.
At the Autonomous Validation Summit (May 12 & 14), see how autonomous, context-rich validation finds what’s exploitable, proves controls hold, and closes the remediation loop.
GameStop is reportedly preparing a potential offer for eBay, an unusually ambitious move given that eBay’s roughly $46 billion market value is nearly four times GameStop’s. Reuters reports: GameStop is preparing an offer for eBay as CEO Ryan Cohen pursues plans to boost the struggling videogame retailer’s market value more than tenfold, the Wall Street Journal reported on Friday. Shares of eBay, which has a market capitalization of about $46 billion, soared about 14% in extended trading. GameStop gained 4%. The company has a market value of nearly $12 billion.
GameStop has been quietly building a stake in eBay’s shares ahead of a potential offer, the report said, citing people familiar with the matter. If eBay is not receptive, Cohen could decide to take the offer directly to the e-commerce company’s shareholders, the Journal said.
.PNG)
You must be logged in to post a comment Login