Within the region, U.K. startups gained significant share in Q2, raising more than $10 billion. That marked the third-largest funding quarter for the U.K. on record, and came in at less than $500 million below its peak quarter in 2021.

蹤獲弝け startup M&A activity also picked up in Q1 and continued that momentum in Q2, even as public-market exits stayed subdued.

Table of contents

• Large rounds drive gains
• H1 2026 up 50%
• UK momentum builds
• By stage
• Looking ahead
• Methodology
• Glossary of funding terms

Large rounds drive gains

Four companies raised venture fundings of a billion dollars or more last quarter, accounting for 25% of all startup investment in the region in Q2, 蹤獲弝け data shows.

Those billion-dollar-plus rounds were raised by an AI-centric group: -owned AI drug developer , which was spun out of ; green steel production manufacturer ; , which is developing robots for home and industrial applications; and , an AI lab founded by former DeepMind researchers.

However, most of the growth in funding year over year and quarter over quarter was driven by rounds of $100 million and over. The majority of funding ��� 65% ���泭went to a group of 42 companies that raised rounds of $100 million-plus. Sectors that stood out for these companies include泭 biotech, quantum, financial services, AI labs, aerospace, semiconductor, robotics and energy.

H1 2026 up 50%

Funding to Europe-based startups in H1 was up 50% year over year to total $42 billion, 蹤獲弝け data shows. Still, the region���s startup investment for the first half of the year remained well below the 2021 H1 peak, when VC funding in Europe totaled $60 billion.

It���s also drastically lower than the $392 billion raised in North America���s record-setting H1, with that region���s funding up 158% year over year.

Europe���s funding deal count subsided last quarter, but mostly at the seed stage. Late-stage rounds were up a bit, while early-stage deals dipped slightly year over year. (It���s worth noting, seed stage rounds are often added to the 蹤獲弝け data set after the close of the quarter, so those numbers will increase over time.)

UK momentum builds

The United Kingdom widened its venture-funding lead last quarter, as startups based in the country raised $10.4 billion ��� not far from the peak in 2021 at $10.8 billion.

The region���s No. 2 startup market, Germany, trailed with $3.2 billion raised by its startups in Q2, and France followed in third place with $2.4 billion. Sweden was Europe���s fourth-largest startup market last quarter, with its companies raising $2 billion.

蹤獲弝け data shows funding to Europe���s AI-focused companies reached more than $10 billion in Q2 ��� the largest quarterly amount so far ��� but slightly below the Q1 percentage, when those companies raised more than half of the region���s startup investment.

By stage

Europe���s late-stage funding totaled $12.1 billion in Q2, up 90% year over year. Large Series C and D rounds were raised by Germany-based robotics developer Neura Robotics; Netherlands-based , which makes inspection tools for semiconductor manufacturing; U.K.-based quantum computing startup ; and Germany-based satellite launcher .

Early-stage funding reached $8.6 billion across 250-plus Europe-based startups last quarter, 蹤獲弝け data shows. Large Series A and Series B rounds were raised by London-based Isomorphic Labs, London-based AI self-learning lab , Germany-based fusion energy company , London-based semiconductor developer , and London-based quantum processor provider .

蹤獲弝け seed funding totaled $3.2 billion last quarter, with a billion dollars of that raised by just one company: Ineffable Intelligence.

Other large seed rounds were raised by , a London-based AI lab for science; Italy-based autonomous driving technology producer ; and Stockholm-based defense tech company .

M&A increase

While IPO activity for 蹤獲弝け startups was muted, M&A showed strong momentum following increased activity in Q1. A total of 154 Europe-based, venture-backed companies were acquired for a cumulative $11.5 billion or more in Q2, 蹤獲弝け data shows. That includes three companies acquired for more than $1 billion each in biotech, industrial AI and micromobility.

Looking ahead

蹤獲弝け startup investment has now steadily increased since the fourth quarter of 2024, with increased momentum in the just-ended quarter, driven by larger rounds of $100 million and over. The region���s startup ecosystem shows particular strength in deep tech and financial services as well as the formation of new AI labs, and M&A activity has fueled liquidity for the next batch of startups.

Now the question remains: Will it be enough to keep Europe competitive with the frontrunners, the U.S. and China?

Related 蹤獲弝け queries:

Related reading:

• 蹤獲弝け Data: Global Startup Investment Hit Record $510B In H1 2026 As AI Boom Accelerates Funding And Exits
• North American Startup Funding Shattered Records In First Half Of 2026, Driven By AI

Methodology

The data contained in this report comes directly from 蹤獲弝け, and is based on reported data. Data is as of July 6, 2026.

Note that data lags are most pronounced at the earliest stages of venture activity, with seed funding amounts increasing significantly after the end of a quarter/year.

Please note that all funding values are given in U.S. dollars unless otherwise noted. 蹤獲弝け converts foreign currencies to U.S. dollars at the prevailing spot rate from the date funding rounds, acquisitions, IPOs and other financial events are reported. Even if those events were added to 蹤獲弝け long after the event was announced, foreign currency transactions are converted at the historic spot price.

Glossary of funding terms

Seed and angel consists of seed, pre-seed and angel rounds. 蹤獲弝け also includes venture rounds of unknown series, equity crowdfunding and convertible notes at $3 million (USD or as-converted USD equivalent) or less.

Early-stage consists of Series A and Series B rounds, as well as other round types. 蹤獲弝け includes venture rounds of unknown series, corporate venture and other rounds above $3 million, and those less than or equal to $15 million.

Late-stage consists of Series C, Series D, Series E and later-lettered venture rounds following the ���Series [Letter]��� naming convention. Also included are venture rounds of unknown series, corporate venture and other rounds above $15 million. Corporate rounds are only included if a company has raised an equity funding at seed through a venture series funding round.

Technology growth is a private-equity round raised by a company that has previously raised a ���venture��� round. (So basically, any round from the previously defined stages.)

Per 蹤獲弝け data, and stood out among the most active U.S. venture and lead venture investors in the second quarter. The highest-spending investors, meanwhile, appear to be those who co-led ���s massive financing.

To get a more expanded sense of how busy startup backers spent their quarter, we put together several rankings for active investors. These include active venture backers, lead investors, highest spenders and prolific seed dealmakers.

Active venture investors

We���ll start with the most active post-seed investors. By this metric 1, the standouts for Q2 were General Catalyst, and Andreessen Horowitz, with 39, 34 and 28 deals, respectively. Of those, more than two-thirds were for AI-focused startups, per 蹤獲弝け data.

All are names that repeatedly show up high in active investor rankings. Y Combinator in particular regularly comes up near the top as it habitually makes nonlead follow-on investments in startups that participated in its accelerator program.

Below, we ranked the 19 most active investors for Q2:

Most active lead investors

When we narrow the ranks to only lead investors in venture rounds, the lineup changes some, but not dramatically.

By this metric, Andreessen Horowitz comes in first with 17 deals. and General Catalyst are tied for second, with 13 led or co-led rounds each.

Overall, at least 12 investors led or co-led six or more venture rounds in Q2. We rank them below.

Highest-spending post-seed investors

Of course, the investors with the largest number of lead deals aren���t necessarily the ones who put the most capital to work.

To get a sense of the highest-spending startup investors for Q2, we look at who led or co-led rounds with the largest aggregate value. It���s not an exact tally, as investors rarely disclose their share of a particular round syndicate. However, it does give a sense of who has been writing seriously large checks.

For Q2, the top slots in this spendy investor ranking were dominated by backers. This included 10 co-lead investors in ���s $50 billion May , as well as , which led a $10 billion separate tranche, and , which led a $5 billion tranche.

Overall, there were 23 investors who led or co-led U.S. venture rounds valued at $2 billion or more in Q2. We rank them below.

Active seed investors

Among seed investors, the usual front-runner, Y Combinator, retained the top slot in Q2. The famed accelerator backed at least 225 seed, pre-seed or convertible note rounds for newly formed startups during the quarter.

was a distant second in the ranks, followed by . Below, we rank the 20 most active seed-stage investors in Q2:

No slowdown

Overall, the active investor ranks paint an image of a startup funding scene still in high gear. All the elements are there: big deals, high round counts and vibrant activity across stages.

We���ll see if it keeps up in Q3.

Illustration:

The firm says entrepreneurs starting their second or third companies have created more than $2 trillion in enterprise value across Europe, and expects the coming wave of AI exits to produce another generation of experienced founders.

���There���s beginning to be this super cycle of repeat founders in Europe,��� co-founder and managing partner said in an interview with 蹤獲弝け News. He recently relocated from San Francisco back to London, where the firm has also opened a new flagship office.

Repeat founders bring not just experience, but connections and the ability to hire quickly to the table, according to Murphy.

From its new fund, Tapestry plans to invest in a similar number of companies as it did with prior funds: Around 30 companies at pre-seed or seed.

Prior fund check sizes were around $1 million but checks from the new fund will trend larger, from around $1 million to $3 million, according to the firm.

The team seeks out previous founders even before they have decided what’s next. ���Let���s spend time together before you start your new company. Let���s ideate, let���s brainstorm,��� said Murphy. ���We���re not taking anything for that ��� we���re not an incubator, we���re not an accelerator.���

The firm���s earlier bets include smartphone and earbud developer and AI customer service startup , which was recently acquired by 1泭for $3.6 billion.

Other investments over the years include drone delivery startup and , which works to automate manufacturing. It also has a renewed focus around AI security with investments in , and .

New investors in this fund are sovereign investor , alongside pension fund and fund of fund . Notably, , CFO at , is also an investor in the fund.

���I think encouraging a vibrant boutique seed environment for funding is very important for encouraging creative new people to start interesting, different and weird businesses,��� said Murphy.

Related 蹤獲弝け query:

Illustration:

founded after departing as CTO in 2022. The firm invests in companies rebuilding the physical economy. As Schroepfer and his partners at the firm see it, surging demand for AI, power and industrial capacity is creating a once-in-a-generation opportunity to rebuild the physical economy ��� from energy infrastructure and advanced manufacturing to materials and robotics. And as AI makes software cheaper and easier to create, the competitive advantage increasingly shifts to the hardware, energy systems and supply chains that underpin it all.

Key to starting the fund was Schroepfer’s experience building out the infrastructure to support Meta���s business. ���I could see the trends coming. We’re going to need all the compute,��� he said. ���I don’t know where we’re going to get the power, so it’s going to create this massive supply-demand crunch.���

Gigascale raised its first institutional fund this month, a $250 million investment vehicle. The firm has already made more than to date.

Schroepfer���s partners at the firm are , previously an investor at climate-focused investor , and , previously at .

Before raising the fund, the firm made 22 investments funded by Schroepfer���s family office in order to prove the model. At the time, the broad perception was you could not make money investing in the hardware layer.

���Not software with higher capex���

Gigascale invests at pre-seed through Series A with some later-stage investments. Its check size is anywhere from $1 million to $10 million.

Hardware businesses are not the same as software businesses. ���It’s not a software business with higher capex,��� said Schroepfer. ���The failure modes are very different. The way you plan and test and iterate, and what you understand is very different.���

In our conversation, we spoke about an array of topics, including energy as a major investment focus, his learnings from running Meta, why now is a great time to build a hard-tech company and what excites him about the IPO.

Gen矇 Teare: What is Gigascale’s thesis?

Mike Schroepfer: It’s really simple. We are backing companies that are rebuilding the physical economy. This is how things are powered, built, moved, manufactured and how people are fed.

The belief is there is a confluence of technological changes that are bringing new products and new companies to market that are better, faster, cheaper than what’s out there. This is the biggest part of the economy.

Another way to say it is that we think the future is atoms, not bits, and it’s a really exciting time to be building these companies.

What did you see that made you decide to set up the fund in 2023?

Schroepfer: A lot of the tech trends I have been part of ��� from the web transition when I worked on Firefox, to the early web infrastructure at , to the mobile transition in the early 2010s, to founding the Facebook AI Research Lab in 2013, well before ChatGPT ��� were looking at the very shallow part of exponential curves. These technological changes did not seem that prevalent yet, but they were on this massive upswing.

I saw the same set of curves in solar cells, batteries and electrolyzers. They were all going through massive exponential cost downs, and at the same time a massive increase in demand. We had electric vehicles showing up, onshoring and manufacturing, and this was pre-data centers. I knew compute demand was going to grow. Where are we going to get the electrons to fuel all of this? It’s going to create an immense supply chain crunch.

Demand and supply were converging at the same time to create massive tailwinds. It just felt like this opportunity to rebuild the entire physical infrastructure in a way that our kids are happy about. Meaning, the new solution wins because it is cheaper, better and faster.

The other co-benefit it brings along with it is that because it is simple and cheaper, it is also less polluting, so it doesn’t hurt humans. I can build a solar farm way faster than I can build a gas power plant. I can live next to a solar farm and get zero pollution. I do not want to live next to a gas plant.

What I understand about the firm is that you are very focused on energy specifically. Is that a misunderstanding?

Schroepfer: It is probably the single biggest area that we invest in. A large chunk of our portfolio is energy. It is a $2 trillion market and it is the place where I think all the disruption is happening. But we also invest in industry, including materials from neodymium to copper, production and recycling, to a lot of AI in the physical world. That includes everything from how I use AI to make my house more efficient with , to how I build power-efficient AI inference chips with .

Then there is the built environment, in terms of buildings, and a little bit in food. We do a little bit in everything, but if you look at our portfolio, the two biggest hunks are really energy and AI in the physical world.

When do you think Silicon Valley woke up to the focus on the physical world?

Schroepfer: In the broad consensus, it happened recently ��� in the last six to 12 months. There were some folks who were looking at it early, but I think the broad consensus has just happened recently.

The other thing that I saw is, if AI is going to make software nearly free to write, then I think software businesses might be challenged, and the moat moves to the hardware. The game becomes: How do I get the infrastructure built to have a better AI? That is mostly an infrastructure hardware problem, less of a software coding problem, and that is going to filter through a lot of businesses.

When I started, frankly, three years ago, I had many people ��� I am thinking of someone sitting in my office ��� saying, ���don’t do this.��� All the money is in software. You can’t make money in hardware.

It doesn’t hurt that , , , and are now household names of companies that have had massive valuation runs because they are such a core part of the physical economy. I used to use Nvidia as my example, but now I can use SpaceX. Talk about a company in the biggest market that is running away from the competition. It’s a really hard company to compete with.

How should we understand the energy needs in the U.S.?

Schroepfer: We’ve been at relatively flat demand over the past 20 years or so, meaning each year that goes by, we don’t need much more power, close to 0%. We are now growing at at least a few percent a year.

Something has gone from almost no growth to relatively high growth. You’ve got hundreds of gigawatts of data centers planned to be built over the next five years alone. That doesn’t count EV charging stations and electrification of homes and factories. It’s a massive supply-demand imbalance right now, and building power takes a long time. If you’ve got to build a power line, if you’ve got to permit a gas power plant, these things take years, not months. It has created massive demand, but everyone wants compute yesterday.

Meta has used tents instead of buildings for their servers because cutting out the time erecting steel for the building gets them compute faster. Everyone is thinking about how to get power faster and how to get compute faster because, again, it’s a competitive advantage when infrastructure is the moat.

Which technologies are you focused on in the shorter term, and then the longer term?

Schroepfer: We have companies deploying things now. In the power crunch, one of the big problems is that the demand for power swings much more widely than it used to. It used to be fairly steady. Now you have big training runs, you have solar that comes on and comes off, and you need a shock absorber to dampen the power or deal with three or four days of clouds or no wind, if you’re depending on renewables.

is a company that has a new kind of battery that lasts for four days. You charge it up, and it’s there for 100 hours. In any event where a power plant is offline or the sun is not shining, Form Energy is there. Utilities think of this instead of building a gas power plant. There are these gas power plants called peakers, which you only turn on when you really need them. They sit there all the time, and then you fire them up in these intervals. Instead of doing that, which is very expensive, you have this Form Energy battery: zero emissions, much cheaper to operate, and built from the ground up for utilities using a totally different technology. They are going to be deploying batteries this year, as an example.

Going in a different direction, the entire supply chain for how we get electrons to a building. I’m going to build a new data center, and I have to hook it up to the grid to get electrons there. There is all this equipment in the middle called power transformers, these big green boxes or big metal boxes. It’s literally 1930s technology. We haven’t changed much since then. They are back-ordered for years now because they are these exquisite hardware machines.

There is a new company, , that said, ���wait a second, we’ve been shipping this new generation of technology called solid-state power electronics in electric vehicles ��� the Model 3, Model Y, and more ��� for millions of units a year, with very fast ramps. We’re taking that same technology and putting it on the grid.��� We’re replacing this 1930s technology with 2020s technology. It’s more efficient, it’s a third the size and, most importantly, they’re going to start shipping lots of units next year. They’re building their factory right now. In 2027, they’ll be shipping lots of these Heron Link units.

A little bit further out, we have a company called that said, ���we’ve got about 10 terawatts, which is an immense amount of power, in the Southern Ocean in waves sloshing around with nothing else going on down there. If we can harness that, it is an untapped resource.���

They’re building autonomous buoys that float in the ocean. They bob up and down and turn that wave motion into electricity. Then they use that to power, on the buoy, a compute node to do AI inference and use to send the bits back. They’re kind of exporting electrons via tokens in the Southern Ocean.

They’ve been testing off the coast of Portland, and they’re going to deploy their first units next year. People have talked about data centers in space. My big pitch for this company is that it’s 100x cheaper to put a ton of capacity in the open ocean than it is to put it into space. If you think data centers in space are a good idea, you might want to look at the ocean.

Then you can think about , a company in El Segundo, California. They are building a compact, next-generation microreactor, or nuclear reactor. You can think of it as something you put on a truck or on an airplane, and it can run and power something for five years straight. Instead of, in a remote region in Alaska or on a Pacific island, doing what they do now, which is shipping diesel fuel there to run a diesel generator 24/7, you install one of these boxes, and it produces power for five years before it needs refueling. Most importantly, again, you would not want to sit next to a diesel generator while it’s operating. It has very toxic emissions. This thing has no emissions. It’s good for humans, and it’s actually going to be cost competitive with those things. Those are some examples of things we’re doing in the power sector that I think are really affecting the future.

Is there an unlock in this industry that has made development cheaper and faster at this moment in time?

Schroepfer: The analog I’d use is from computing. We used to build mainframes, these big building-sized computers. Then we had minicomputers that were still really big. This is the motherboard for the first server we designed at Meta that we deployed in 2011, called Freedom. It was a Type 1 server. It was the web server.

I installed millions of these, maybe tens of millions. I don’t even know how many. They’re all the same, every single one of them. They go in a pizza-box-size thing that goes into a rack in a building. That building comes in four units. Each of those is the same. That building is next to another building, which is exactly the same. We build four of those on a site. They all look the same. I did that in 17 places around the world. They all look the same.

The technique we use to make things cheap is mass manufacturing. Everything in your life that has gone down in price or improved in price-performance is mass manufactured: your iPhone, the servers and data centers. They’re all the same. They’re mass manufactured.

The world is full of custom, bespoke stuff that’s wickedly expensive.

In the power grid, for example, all of the stuff I talked about, you custom order it. I want a transformer. I do engineering design. I send it off to someone. Four years later, a truck shows up with the crane and all the rest of it. That’s inherently expensive and gets more expensive every year. Everything that is custom gets more expensive every year, so I think the biggest thing we’re seeing is this move to things that are mass manufactured.

Solar panels are mass manufactured. Batteries, the things that go in your phone or in your electric vehicle, are 99% cheaper than they were 20 years ago. That’s because we manufacture them at a massive scale. Every time you double the size of manufacturing, you get a 10% to 20% reduction in cost, and there are so many other problems like that.

In this case, the power electronics, the transformer, are all special-purpose. Heron Power is going to make the same box for a data center, for an EV car charger, and for a solar farm. It’s the same box. No changes. That’s how we’re going to get a cost curve down for these things. That is the most exciting trend underneath this: the idea that generalization and mass manufacturing of things allows you, year over year, to reduce costs.

When you’re competing in the power industry, fossil fuel costs have been basically stagnant. They go up and down a little bit, but if you average them over 50 years they are not on a cost-down curve. It doesn’t get cheaper to get oil out of the ground. My competition is flat, and I’m getting 10% to 20% cheaper every year. That’s a great business to be in. That’s the big trend behind all of this. We saw it first in solar and in batteries, but it enables a whole bunch of other things in other industries, like power electronics and more.

Are we at this time very dependent on China for mass manufacturing?

Schroepfer: A lot is coming from China, but I visit a factory a week in the United States that is getting spooled up with robotics, with really smart founders from and SpaceX. It turns out that when you start in 2026, you can build a much more efficient, much faster factory. You can use modern technologies.

Right now, China has the industrial base, and we’ve let it go. But I think we have a shot at rebuilding it in the United States, and I see brilliant founder after brilliant founder running at this problem inside the United States every day and every week.

It’s one of the reasons I started this firm, too. I think we have a shot to rebuild that industrial base in a next-generation set of technology. Just like regions around the world that didn’t have landlines went straight to cellphones, we’re going to go straight to fully automated robotic factories with 3D printing, laser milling and the latest technology set. It is not going to be a cut-and-paste of what happened in China, but a next-generation set of technologies that allow the U.S. to be self-sufficient in what we’re doing.

We’ve seen new techniques. As an example, rare earths were something no one ever talked about. Neodymium is this rare earth material that is key to making a magnet. Who cares about magnets? Well, magnets are in every electric motor in anything. Anything that has an electric motor, you care about magnets. Almost all the neodymium is made in China, and it is made in this very polluting, dangerous process. You do not want to visit one of these factories with fluorinated gases ��� it’s awful.

We’ve got a company making neodymium in Alameda, California. That is not an easy place to permit polluting things, which is fine for them because their process doesn’t pollute at all. It’s very simple. It’s two reactors. I walked around the facility. You don’t need any protective gear. Because it’s so simple, they are cost-competitive with Chinese imports.

To their customers who are saying ���I’m trying to make magnets,��� they’re saying ���great, I will sell you neodymium. I have it. It’s cost-competitive.���

Everyone is excited, but the thing we’re whispering in the background is, it’s also not polluting. This is how we’re going to win. It’s not a cut-and-paste of that technology over here, but saying, ���How do we approach this in a way that’s simpler and cheaper, and then likely cleaner as well?���

We’re doing the same thing in copper. We’ve got a whole bunch of bets in different kinds of materials where I think we can do it better in the U.S. We’ve got a company, , in South Carolina that’s doing this for copper recycling. We’re doing it in cement manufacturing. There is a whole variety of opportunities. I don’t have enough time to meet all these entrepreneurs.

We talked a lot about some of the companies in the energy sector. What are the other areas of investments that you’ve made that you’re excited about?

Schroepfer: I mentioned this a bit, but worth going a little deeper on is applications of AI to the physical world. I talked about one: Fractile, which is building a next-generation AI inference chip that’s much more power efficient.

Another example is a company called , which is using AI to put a simple piece of hardware on a power line, on both sides of a power line, to detect if there is a fault in that power line that might be causing a fire. The idea is that if you detect that fault sooner, you can prevent the fire before it’s a problem instead of waiting for it to happen and then having to respond. Using AI plus hardware to figure these things out is another example of that.

We have another company called that’s using AI to help with the nuts and bolts of how people make transactions to build energy projects. There is a lot of due diligence work and other things that need to happen. You can build, very much like for legal or for doctors, these vertical AI companies. This is a vertical AI company for energy developers. There is a lot to happen there.

Then is doing industrial automation with robots, using next-generation models to train robots to be more effective in factory environments, back to my point of how we are going to do this in the U.S. with advanced robotics. I think AI for the physical world is a big area.

I talked a bit about materials: neodymium, copper. We have a company called that’s making clean chemicals. Those would be the big areas I would highlight.

I know there are a lot of investors that you partner with or work with that are similarly focused in this area.

Schroepfer: The thing that’s been most interesting is that there is a set of folks who have been doing hard tech or climate for a while, and they are great partners of ours, from to to to many others. But what’s been interesting to me is the generalist firms coming in. A very common co-investor for us is , , or . We’re seeing them come in large amounts, because they’ve seen the economic opportunity here.

What did you learn from spending 14 years at Meta?

Schroepfer: I learned a few things. When I joined in 2008, the company had fewer than 100 million users, was not profitable, and had about a 150-person engineering team. We relied on outside parties to do all the hardware work. We were leasing data center space.

Over the next 14 years, we grew dramatically in users and profitability and in the size of the team. But we also moved into the physical world. As I showed you the server, we built our first data center in 2011. I built 10 million-plus square feet of data centers in 17 places all over the world. We then moved to consumer hardware, so we built the smart glasses, the Oculus Quest VR headset, and the Portal. Then we moved into AI research with the Facebook AI Research Lab in 2013.

That shift into the physical world brought a lot of really humbling lessons. There were a lot of times where stuff just went wrong. At the very first data center, I remember touring it under construction, and we had wood blocks on the loading dock because they had graded the loading dock wrong, so the trucks couldn’t back up and unload properly.

It’s this new, awesome, state-of-the-art data center with a free-air cooling system, and we got wrong the thing that every in the country has five of. It’s a million small challenges.

This is the thing I bring to the founders that I see: having learned how to build stuff in the physical world builds an appreciation for the risks and scale, and for how you need to emphasize speed and learning rate.

People learn the wrong lesson. They think hardware means spending a lot of time designing on paper. Wrong lesson. You have to get out there because you don’t know which part is going to blow. You have to get out there and learn as fast as you can and as cheaply as you can, so that when you’re in mass production, you’re not learning things, you’re just repeating.

That lesson, from data centers to consumer hardware, matters. When we build consumer hardware, you spend 18 months building this exquisite pair of glasses or this exquisite headset, but before you sell it, you have to do this drop-test thing, where you literally say, what happens when someone takes it out of the box at home and drops it on the ground? If it breaks, they return it, and we eat the cost. So you sit there and drop this thing with high-speed cameras over and over again to make sure it will survive a drop from head height. You don’t think of these things when you’re designing it. You have to make sure someone can drop it and it’s fine, or spill some wine on it and it’s fine.

Those problems in the real world, plus the challenges of building an executive team and scaling an organization, are the fun part of my job: working with our founders and having their back when things are tough, when they need to recruit someone, or when they’re running into a challenge in the real world, because I’ve seen it. I’ve seen it all.

What’s your reaction to the SpaceX IPO?

Schroepfer: I’m honestly pretty excited about it, because we have a lot of SpaceXers in our portfolio. I have a lot of friends who are alumni or work at SpaceX. Having more people in the world with the financial resources to work on audacious engineering projects is going to be really good.

I think it’s also a lesson in building and hardware. How many companies can land rockets the way SpaceX can? They’ve been doing this for a decade, so they have a very large technical moat in terms of what they’re able to deploy in the world. Starlink is another example. Everyone is racing to catch up. If you’ve ever used Starlink on an airplane, you don’t ever want to be on an airplane without Starlink. It’s hard to describe other companies that have such a singular product as SpaceX. I think it’s exciting that the markets are rewarding that. I can’t wait to see what SpaceX alumni do next.

I imagine there’s going to be a lot of company formation coming out of that IPO.

Schroepfer: It’s going to be an exciting five years. I met you after I started my first company in 2000 and sold it off. We looked at starting another company, and then I worked at and Facebook, so I’ve been through a couple cycles of this. I think it is the most exciting time to start a company, in terms of the capital available, the AI tools available to you, and the physical tools to build things quickly in the physical world. It’s the bet I made: I think this is the most interesting time to be building new companies. That’s the smaller version of why I did this. I think this is the time. This is the thing to be doing.

Related 蹤獲弝け queries:

Illustration:

The Menlo Park, California-based firm highlighted its early investment in , which last month overtook rival as the top-valued frontier lab in the world with a staggering $965 billion valuation. While Menlo Ventures��� investment in Anthropic���s was not its first bet on artificial intelligence, the firm described it as its ���flag-planting moment.���

���We made our first investment in Anthropic in 2023, when the company was pre-product, pre-revenue. By then, ChatGPT was a household name, and many believed the LLM race was already decided. We saw it differently,��� the firm wrote in published Tuesday. ���In and his founding team ��� arguably the most accomplished researchers in the field ��� we saw the rare mix of technical depth and clarity of purpose that defines a category leader. We were convinced there was room for another independent foundation model company, that Anthropic was the team to build it, and that an investment in Anthropic could anchor our broader AI strategy.���

The firm went on to lead Anthropic���s the following year.

���That early relationship gave us a rare vantage point on the model layer and on the infrastructure, workflows, and application opportunities forming around it,��� the firm said this week.

Two new funds

The firm���s new capital is across two funds: , earmarked for seed and Series A startups, and , a growth fund for Series B and later startups that are ���already pulling away from the pack and on their way to becoming the breakout names of the AI era.���

Along with Anthropic, other notable Menlo Ventures investments over the years include , , , and . Anthropic, which has filed plans for a 2026 IPO, would be the largest exit to date for one of its portfolio companies by far, with an expected IPO target of $1 trillion or more.

Related 蹤獲弝け query:

Related reading:

• Anthropic Nears $1T Valuation And Leapfrogs OpenAI On 蹤獲弝け With Massive Funding Round

Illustration:

Since its launch almost two decades ago in 2007, the firm ��� which had an initial fund of just $3 million ���泭 has invested in around 550 companies. With this latest fund, its assets under management have reached $1 billion.泭

蹤獲弝け News spoke with , the firm���s managing partner who joined Seedcamp in 2010 and , who rejoined the firm in 2022 to head up the select fund and establish a New York presence.泭

Seedcamp invested early in , , , and .

Since fund 2, it has invested in 100 companies per fund. ���What we���ve learned is that you need a community to support each other,��� said Espinal. The tipping point for the firm was 70 companies where it became clear that founders were helping one another, becoming customers, and teams starting new companies.

���We realized early on that the best thing a founder can get is access to another founder who just went through that experience ��� not necessarily a founder who is successful 10 years down the road and is a great figurehead, but someone just a little bit ahead. That���s effectively our secret sauce,��� said Espinal.泭

Seedcamp also invests in software and vertical AI, but is careful about what is compelling and unique. ���We���re trying to monitor so we���re not one of eight bets in one area that���s been overinvested within the AI vertical space, and making sure that you���re not betting on number 100 in a space that���s hypercompetitive,��� Espinal said.泭

Seedcamp plans to invest in 35 new companies per year, totaling 100 to 120 for the new fund. It invests up to $1.3 million in its initial check, and will lead roughly 70% of those deals with a 5% to 10% ownership target.泭

The firm reserves 40% for follow-on seed and Series A rounds. Its select fund will invest in portfolio companies from Series B onward.

���Building is so much easier and faster now,��� Howe said. ���Signals of product-market fit are there earlier. The founder DNA is still the same, but the ability to see it in action earlier is there with the AI lift.���

New York presence

Howe, who heads up the New York office, noted that 蹤獲弝け companies are heading to the U.S. earlier. ���Historically, maybe we���d see a company raise a round and stay in Europe, dominate their local market, raise a few more rounds, and then come to the U.S.��� she said. ���Now we���re seeing them come right from the get-go.���

From fund 3, its 2014 vintage fund, the firm’s return is 13x distributions to paid-in capital, with Revolut, UiPath and seed investments from that fund.

Related 蹤獲弝け queries:

• 泭

Illustration:

His investment mandate primarily focuses on early-stage technology companies from seed to Series B that align with or impact the global telecommunications, network infrastructure and enterprise software sectors.

Under his leadership, AT&T Ventures targets investments in software, hardware and infrastructure sectors where AT&T’s network scale and internal engineering resources provide a distinct commercial or technical diligence advantage. Portfolio companies include enterprise and deep-tech firms such as , , , , and .

Prior to his current 12-year stint directing AT&T Ventures, Taneja spent more than two decades working across corporate development, venture lending and investment banking. He previously managed M&A and strategic investment activities for during ownership.

Taneja also served as a director at , where he focused on growth-capital debt and equity investments in mid- to late-stage technology businesses, as well as holding corporate finance and investment banking roles at and .

In an email interview with 蹤獲弝け News, Taneja shares why he believes that while AI has drastically lowered the barrier to building software, it has also shifted the definition of seed-stage technical risk.

The new dynamics, in his view, gives AT&T Ventures an opportunity to differentiate itself by offering immediate, real-world technical validation and network integration rather than just capital.

The interview has been edited for brevity and clarity.

蹤獲弝け News: If startups are building fully functioning apps by the seed round using AI, what does that mean for the traditional definition of technical risk? Is tech risk dead at seed, or has it just evolved into something else?

Vikram Taneja: The old definition of technical risk was ���can they build it?��� Although not entirely absent at the seed stage, I���d say it is becoming less relevant given the dramatically lower barrier to building software with AI tools.

But what replaced it is actually harder to answer: ���Is the tech defensible?��� Not just ���does it work?��� but ���does it compound?���

Data moats, proprietary training sets, network effects built into the architecture ��� that’s the new measure of durability.

In prior cycles, technical complexity alone created some natural protection. As a result, the technical risk conversation has shifted to focus on how a company defends itself over the next three to four years, especially as frontier labs move down the stack into application layers and start targeting entire verticals.

Similarly, the distribution question shows up much earlier. ���How can you get this to market?��� is increasingly asked at the seed stage rather than later in the cycle.

We���re also seeing increased competition for investors to secure larger stakes at seed that they would have previously pursued at the A round. This is driving investors to be more thorough at the seed stage, and founders have to be prepared to meet higher expectations across the board.

When anyone can use AI tools to spin up a working app in a weekend, product execution happens fast, but moats can be incredibly shallow. At the seed stage, how are you separating a truly defensible platform from a beautifully executed wrapper?

Taneja: In early 2025, we saw a wave of AI wrapper companies built on top of frontier models like ‘s GPT, ���s Claude or LLaMA, and a lot of capital flowed into them. What���s changed is that frontier LLMs have now clearly started to take more of a platform approach ��� moving into the application layers and beginning to pick off the low-hanging fruit.

This is why defensibility becomes critical in AI investing. No platforms are totally defensible, but on some level, you have to ask that question now at the seed stage.

We���re looking for platforms using proprietary data that can���t be replicated by AI, companies that have embedded deep domain expertise ��� areas where general-purpose AI still lacks industry context ��� into their workflows, or highly specialized ecosystems or niche markets that provide another layer of insulation in categories that are too targeted for frontier labs to pursue directly.

Are you seeing a change in the actual headcount or makeup of seed teams? If AI handles the heavy lifting of the initial code, are these founders spending their seed capital on engineers, or are they shifting resources immediately to distribution and go-to-market?

Taneja: There is still an engineering focus in the early stage, as there should be, but we are increasingly seeing product, sales, or partnership roles becoming sought after earlier than in the past. And the reason is, as you stated, that it���s easier to build a working prototype, or even a production-ready application, so the focus very quickly turns to establishing trials with customers or exploring distribution paths to dial in the product features.

For strategic investors like AT&T Ventures, where we often do proof-of-concepts with potential portfolio companies, this is very exciting. We get a chance to work with companies earlier in their formation, can get real technical validation much earlier than otherwise, and can similarly try to find a path to collaborate more quickly.

AT&T Ventures has traditionally played heavily in the Seed to Series B space. If institutional VCs are rushing to seed to grab larger stakes because the tech is mature, how does that change the competitive landscape for CVCs? Are you finding yourself competing directly with traditional multistage funds earlier than before?

Taneja: The makeup of seed rounds has definitely changed. Multi-stage funds used to show up at Series A or B when there was enough traction to underwrite. Now they’re at seed because, as we discussed, the companies are mature enough, and they are trying to find winners earlier in the cycle. So yes, we’re in the same rooms as before.

But I’d push back on the idea that we’re competing directly.

A Tier 1 financial VC���s seed check and an AT&T Ventures seed check are different instruments. They are offering capital, brand, guidance and pattern recognition from backing hundreds of companies.

We’re offering something a financial VC structurally does not: our network teams working with your product in a production environment, oftentimes before we even write the check, for example. That’s free diligence running in both directions. We’re validating the company, but it’s also receiving a real-world signal from one of the world’s largest network operators.

For a seed-stage company that’s already solved the building problem and now needs distribution, that���s tangible value and complementary to what financial VC firms are providing. So that competitive pressure has actually sharpened our value proposition. It forces us to bring more than just capital to the table.

Historically, corporate partners want to see enterprise readiness, security compliance and scalability ��� things a seed startup rarely has. If a seed startup has a fully functioning product but is still a two-person team, can an enterprise like AT&T actually run a pilot with them, or does the corporate integration timeline become a bottleneck?

Taneja: It starts with strategic rationale. That has always been the entry point for us at AT&T Ventures, and that hasn���t changed. If that is in place, then it doesn���t always require full enterprise readiness to start a pilot. It can be a structured trial or a highly targeted engagement, depending on the company’s stage.

We have a number of ongoing proof of concepts with portfolio companies across areas such as AI-RAN, connected infrastructure and computer vision.

The key is clarity upfront ��� clarity on what the objective of the engagement is and how we measure success. Once that is clear, even early-stage companies can be integrated into a learning or testing environment without unnecessary delay. The goal is to make the AT&T relationship feel like an accelerant to further adoption.

If seed is the new Series A in terms of product maturity, are you seeing Series A pricing bleed into the seed round? How are you disciplined about valuations when the product looks like a Series A, but the company infrastructure is still very early?

Taneja: Seed pricing indeed looks different than maybe four or five years ago. We���re routinely seeing seed deals priced in the low- to mid-single-digit-million range at about $20 million to $25 million post-money. This is pretty much where Series A deals were a few years ago. But it���s not necessarily unjustified ��� the makeup and traction of seed-stage companies are much further along than predecessor vintages as we���ve discussed.

We stay disciplined by being explicit about what we’re actually underwriting. We’re not just underwriting the financial return on this round ��� we’re underwriting the strategic value of the relationship over a five- to 10-year horizon.

Does this company make AT&T’s network more intelligent? Does it open up a new customer segment? Does it validate a thesis we’re building around? Are there commercial opportunities beyond our initial thesis? When you frame it that way, it gives us a longer horizon to work with and provides multiple levers to pull.

And honestly, that’s where our engineering and product teams play a key role. They help us decipher whether the product that looks like a Series A is actually built like one, or whether it’s a great demo sitting on a foundation that hasn’t been stress-tested. That technical read bolsters our conviction when making investments.

A functional AI app at the seed stage still requires massive infrastructure. When you evaluate these early-stage companies, how much does their underlying architecture and how they handle data processing or edge computing factor into your decision?

Taneja: Architecture is a key part of our diligence process. The way we think about it really depends on the ultimate use case. Is it for internal use ��� i.e., a tool that AT&T will be working with in our environments ��� or is it something we���d be distributing or incorporating into some form of product offering?

If the former, all aspects of the architecture will be reviewed, and this is most likely to occur throughout trials and proof of concepts as we develop a technical understanding of the application or product. If it���s the latter, then we���re likely most interested in understanding how this product architecture scales over time and what it means from a cost, latency and infrastructure perspective. We love to see companies embracing edge-related technologies, but that doesn���t preclude us from working on applications that use traditional data processing methods.

You���ve spoken before about your interest in ���physical AI��� and robotics (like Apptronik). The software lifecycle is easily compressed by generative AI, but hardware and physical deployment take time. Does this ���seed is the new Series A��� trend apply to pure-play software strictly, or are you seeing AI accelerate physical tech and IoT at the early stage too?

Taneja: Physical AI is a sector we���ve been looking at quite a bit, particularly because inference and decisioning in autonomous systems, robotics and connected devices create a very different type of demand profile on networks.

The software layer is clearly accelerating ��� things like perception, control systems and decisioning are moving faster because of AI (the rounds show it!). That will ultimately help pave the way for the adoption of physical AI. However, the physical deployment cycle still takes time, so you don���t see quite the same level of time compression there.

What is interesting for us at AT&T is the intersection ��� how intelligence is moving closer to the edge and how that changes the way networks need to be architected to handle those workloads.

Related 蹤獲弝け queries:

Related reading:

• In Charts: Seed Deals Keep Getting Bigger As Odds Of Reaching Series A Fall Dramatically
• Amid Record Robotics Funding, Apptronik Raises $520M Series A Extension To Boost Production Of Humanoid Robot Apollo

Illustration:

“Silicon Valley has done very well with software, but while software was eating the world, they forgot about silicon,” Barrett told 蹤獲弝け News in an interview.

The firm recently closed a $475 million fund focused on investing in deep-tech startups at seed and Series A. In the decade-plus since its founding, it has built its investment thesis around the idea that breakthroughs in science and engineering ��� not just software ��� would create the next generation of valuable companies.

With demand surging for compute, semiconductors and energy, Barrett argues the rest of the industry is now catching up. “We’ve been at it for more than a decade,” he said. “In recent years, as AI is eating software, people are scrambling back to recognize that the energy, semiconductors and infrastructure they operate on all need capital too. We’ve been operating in that regime for a very long time.”

Barrett is originally from Australia and came to Silicon Valley in the 1980s. He’s been coding for 50 years, he said, after developing an early and deep respect for science and engineering as the child of two engineers. His childhood was steeped in punch cards, draftsmen and drawings of control systems and machinery, he said.

���Science lets you follow breadcrumbs from prehistoric plumage to semiconductors. One principle can be applied somewhere orthogonal and create extraordinary value,��� Barrett said in a lengthy interview with 蹤獲弝け News.

Barrett went on to found video game developer , joined to build the entertainment browser acquired by , and was subsequently CTO at prior to co-founding Playground Global in 2015.

Playground Global operates a lab in the former Palo Alto Research Building in Palo Alto, California. The location hosts 350 people, including those working at its portfolio companies and others with adjacencies working from the lab.

On a recent visit to the warehouse, I saw various models of robots, materials for aerospace construction, and a model of building powerful lasers to increase the speed of semiconductor manufacturing. The quantum computing startup , a Playground portfolio company, moved in when it had three employees and moved out when it reached 90.

The firm has four general partners. Along with Barrett, they are , the former CEO of and who architected CPUs at Intel that helped computing take off at scale, and who joined the Playground team last year as a general partner specializing in semiconductors; , who has made many investments in biotech, including ; and co-founder , who led the investment in .

What follows are highlights from a wide-ranging interview with Barrett that covered topics including sovereign technology, the need to invest in companies that operate on the physical plane, and why he believes putting data centers in space is stupid.

This interview has been lightly edited for clarity.

Gen矇 Teare: What is the thesis for Playground Global?

Peter Barrett: It is about reducing new results in science and engineering into commercial and societal value. That means operating at the boundary between computation and the physical world. We are very interested in new capabilities of computation driving civilization forward, and that inevitably means operating in the same physical plane that we live in.

We’re seeing in our data a huge amount of funding going into space, semiconductors and robotics. It seems as if the whole venture industry has pivoted to this much broader array of companies. Do you see that as a good thing?

Barrett: We lost a lot when people weren’t investing in things that strike us as important. It is good that there is capital chasing the things we care about and that have real consequence.

You can���t spin up a deep-tech practice overnight. You still need domain expertise. You still need to understand why investing in nuclear reactors is good, and why data centers in space are preposterous.

Silicon Valley hasn’t been very efficient with much of the capital it’s deployed over the past decade or so. But I do think it’s good that people recognize that software may be eating the world, but you can’t eat software. We have to operate in the physical layer.

Do you think Silicon Valley gets more efficient?

Barrett: We need to do the work. You develop the instincts and the platform to deploy capital efficiently into these places.

It’s important that people recognize there’s this unprecedented funnel of technical change. AI is an early indicator of it, but we have technologies like quantum. We know how to produce computation using things beyond transistors and semiconductors.

We’re scratching the surface in terms of AI models. We’re right at the beginning of an explosion and renaissance in materials science driven by things like quantum computing.

Now would be the time ��� and candidly, I feel the imperative ��� that anywhere there is science and capital, it needs to be turned into value, especially in liberal democracies, because the despots are doing a pretty good job of it. It’s incumbent on us to stay ahead.

We’re in the DOS age of AI. We’re scratching the surface, both in terms of the models we make and the hardware we run them on.

Now would be the time for people to write checks into things that are sensible and valuable. We spent a lot of time on NFTs. How are we doing with cancer? How are we doing with our most difficult challenges in terms of healing and feeding the world?

There are lots of new degrees of freedom that could take capital and turn it into value.

Do you think deep tech fits the venture thesis, despite the long time horizons and the amount of capital it requires?

Barrett: The long time horizons certainly exist. If you’re building PsiQuantum, we’re building million-qubit quantum machines. That takes billions of dollars and a decadal effort.

The corollary is that we’ve had hardware exits in two years. The timelines for hardware aren’t necessarily that different from software.

Therapeutics naturally take a longer time, because of clinical trials. But we’ve also seen exits there. One of our companies tested half a million drugs in a single animal and created a new corpus of AI input for building models to create therapeutics. That’s not a decadal effort ��� that’s a handful of years before exit.

We try to craft a portfolio that’s a mix of tactical and strategic. Some of these companies get to hundreds of millions in revenue within a few years. Others, like PsiQuantum or , may take a decade to reach full entitlement. That’s part of portfolio construction.

The biology company you mentioned ���泭what’s its name?

Barrett: . It did the largest pharma deal of its kind last year with . The deal could be worth $2 billion on the back end.

It’s a unique mechanism to create giant AI training sets by using physical systems ��� using animals and in vivo testing to create that dataset. It affords the ChatGPT and biology moment, where you can have large enough training sets to build big models.

You describe the firm as investing somewhere between improbable and impossible. Are there companies that really fit that thesis when you first met them?

Barrett: When we first met PsiQuantum, they were talking about building a machine which was 10,000x the state of the art. Using then-current technologies, it would have been the size of the Sierra Nevadas.

They required exponential improvements in both hardware and software, and they’ve achieved both. It’s the size of a warehouse, not a laptop.

The work we’re doing in biology, materials, quantum algorithms and superconducting logic ��� which will replace transistors and semiconductors ��� all of these things sound like science fiction, but they’re much closer to improbable. In many cases they’re entirely practical before we invest; they just seem improbable to those unfamiliar with the domain.

There are things that are not impossible but are still really dumb ��� data centers in space, small modular reactors (SMRs), or fusion. The physics may work, but the economics don’t, or the timelines don’t align.

I’m disappointed we haven’t invested in anything that turned out to be more impossible than we thought. None of our portfolio companies failed because the technology didn’t work.

We’ve had capitalization failures. We flew hydrogen planes. We’ve built things that were thought to be virtually impossible that turned out to be straightforward. They may have missed their market or may have been unable to raise the capital to continue.

I want to do something where the technology doesn’t work, and we���ve yet to do one of those.

Is there a company you missed out on where it looked impossible and you wish you’d invested?

Barrett: I wish I hadn’t taken ‘s word for it when was a non-profit.

We haven���t missed many. As the roadmap developed, we wish we had been earlier in a couple of categories that are really interesting. But overall, we haven’t missed too many.

In which sectors or companies have you invested where the time horizons have shortened due to AI?

Barrett: Adding Pat Gelsinger to the team reflects an interest in scaling semiconductors along various dimensions, including energy efficiency and how power is delivered.

We do everything from nuclear reactors all the way through to transmission, energy conversion outside the data center, inside the data center, under the chip, what kinds of chips you���re running, what models run on top of those chips, what architectures those chips are made from, and what materials those chips are made from.

At every layer of the infrastructure ��� optical interconnects, memory systems ��� we have a best-in-class company at every point. We built the first AI accelerator a decade ago, and we���ve broadened that to encompass the entire ecosystem, from the creation of electrons to how they expend themselves doing useful software work.

There are bubbly aspects of the current AI moment, but the bubble is being modulated to some degree by the unavailability of energy.

We���re in the DOS age of AI. LLMs are embarrassingly incompetent compared to what comes next, but we believe in the durability and growth of AI, and are making investments in model architectures and the ways AIs are trained. We see demand for compute, energy and infrastructure continuing to grow.

We have technologies that can reduce general-purpose compute workloads by 100x to 1,000x over state of the art. We believe we know how to make the energy and deliver it. We know how to connect these systems.

So quixotic pursuits like putting data centers in space are unnecessary.

Talking privately to hyperscalers and Fortune 50 companies, they all say there is way more demand for AI in its future incarnation than exists today. It���s incumbent on us to figure out how to do it 100x, 1,000x or 10,000x more efficiently, because that demand turns into GDP growth and better solutions to our hardest problems.

What are the companies in energy and semiconductors that you are betting on?

Barrett: One example is the wild superconducting logic company . We can make things that are post-semiconductor and post-transistor, with devices that switch five orders of magnitude more efficiently than transistors.

They operate at cryogenic temperatures, but quantum computers do that, and our extreme ultraviolet lithography system does that. The future of computation is cryogenic. Even after you pay to make it cold, you���re still 100x to 1,000x more energy-efficient on compute.

This technology has been around since last century, but it���s mainly been used for secure signals intelligence and radar applications. We���re generalizing it for compute.

Another example is . People talk about SMRs, which are a physics solution to a financial problem, or fusion, which is still decades away. Alva instead uprates the existing nuclear fleet to get hundreds of megawatts out of each unit by replacing 1970s steam generators with a 2020 steam generator.

We can deliver power in a handful of years. No new fuel, no new regulatory path, and a business model that makes sense for operators. We can put gigawatts onto the grid without moving a fence line of an existing reactor and without upgrades to the electricity grid.

We know how to make AI training wildly more efficient. We know how to train different kinds of AI models that we���ve been unable to train.

The last supercomputer at uses something unlike a CPU or GPU to run existing software. We���ve been running software the same way for 70 years, but there are other ways, with dataflow architectures. We have a company doing that ��� [].

The degrees of freedom from materials, systems, code and models have never been greater. We���re exploring all of them. But most require rolling your sleeves up in the physical world.

LLMs feel like brute-forcing something ��� like a drunk looking for keys under the streetlight. We���re pushing more and more into that, and I think that���s a dead end. We know other ways of moving forward.

Are you seeing new model companies, separate from LLMs, that are going to solve things?

Barrett: Our brains are not LLMs. They���re not transformers. Transformers are effective, but they are one of a long line of soon-to-be-extinct models that get replaced by something that works better.

That millionfold gap between our brains and GPUs is an architectural gap. Meat is much worse at computation than hardware can be, so biology shouldn���t be better.

Physics allows a million times a million more efficiency, and we should start chipping away at that.

Intelligence is useful and can be pressed into service against basic things like photosynthesis. Plants were invented by accident of evolution 3 billion years ago. They���re pretty, but not efficient. They shouldn���t be green; they should be black. We know how to make photosynthesis twice as efficient, and probably 5x more efficient.

We���re not stuck with the physical constraints of our technology or of nature. Nature is beautiful, but cobbled together by a process that we can have agency over.

All the materials that operate our civilization are discovered, not designed, because we can���t design things we can���t simulate. Our best computers cannot simulate the quantum nature of nature. That���s about to change.

We���re stumbling around in the dark, relying on serendipity and the occasional magical material. Whereas we can construct any number of materials with magical properties that are currently hidden from us by our inability to simulate the quantum mechanical processes that animate chemistry.

We are right on that threshold of unlocking all of these dimensions. And at the same time, we���re putting money into NFTs, the metaverse and other things that will come and go, without anybody ever caring.

Are you talking about the mix of quantum with biology and model-focused companies?

Barrett: Quantum allows us to directly design materials, directly explore the method of action of drugs, and directly design drugs.

AI has a role to play in biology and understanding structures we can measure. We think there are quantum wet labs where we can measure the performance of small-molecule drugs against models of nature and then verify in nature.

We don���t know how many things that animate our industry actually work. We don���t know how Tylenol works. We don���t know how the Type II superconductors we���re building fusion reactors out of work. We know that if you take iron and nitrogen and arrange them in a certain way, they produce magnets stronger than rare earth magnets, but we don���t know why.

There are mysterious things we���ve stumbled across that hint at an Aladdin���s cave locked behind a wall of computation. That wall is coming down.

Which sectors do you think are going to take a lot longer to come to fruition?

Barrett: Civilization will operate on fusion eventually, but right now the only reactor that works using gravimetric confinement is the sun. I think that���s a long way off.

Data centers in space are stupid. You can���t operate a gigawatt data center in a thermos. We have terrestrial answers to those questions that we should pursue.

I���ve always been a detractor of self-driving cars, which are starting to work. Now we need an economic model that makes them sensible and doesn���t drown our cities. The problem with transportation in cities is not the degree of autonomy. If we cared about traffic deaths, we���d worry about roundabouts.

There���s also nonsense with NFTs and the metaverse which have sopped up enormous amounts of capital. Small amounts of capital using these tools against our most difficult diseases would yield results. Small modular reactors are an unwarranted innovation.

There are lots of things that, at first blush, seem good and valuable, but there are far better solutions that are simpler and more imminent. We need to be practical about where the money goes.

There was a company that just joined the 蹤獲弝け, valued over $1 billion this past month, doing orbital data centers. Are you saying this whole category doesn���t make sense?

Barrett: To his credit, will show you a picture of what a 100-kilowatt data center looks like, and it���s bigger than Starship. A 100-kilowatt is a small rack from that is human-sized.

The arguments are that there are a lot of renewables in space. But there are a lot of renewables on the ground too. North Western Australia has solar and wind that are 70% naturally firm, and on the ground, so you can build things on it.

Put a data center in North Western Australia, which we are doing. We have a renewable site 35x the size of Manhattan.

Energy generation and compute in space is a nonstarter because space is not cold. You���re building things in a thermos and need to get rid of heat. A single human-sized rack is 100 kilowatts, which is about the size of the International Space Station���s radiators and solar panels.

Starship has yet to actually put anything in orbit. It���s made some fireworks, which are pretty, and it���s a beautiful thing. is an amazing company because of Falcon 9 and Starlink. But data centers and power generation in space makes no sense.

We know how to build arbitrary amounts of energy generation on the ground with very safe, very large nuclear reactors. We���ve been doing it for decades.

For all the talent and genius rattling around the Valley, we do spend money on silly things.

Do you think now is the most exciting time to be investing, or have some of those investments already been made and are going to come to fruition?

Barrett: We���ve already made investments in things on a really steep trajectory.

Snowcap will take a decade before we���re building GPUs with that technology, but we���ll have commercial product from them next year. We���re getting better at early, undeniable signals.

PsiQuantum is a long journey, but some things just take that amount of time.

X-Lite seems like a ridiculously long journey, although we���re building the prototype facility now, and it received the first money from the new CHIPS Act.

Some hardware companies making silicon or systems are getting significant revenue in a handful of years.

There���s a sleeper in Fund I. Its first trick was to make MRI machines 100,000x more sensitive, and they���re shipping those. In the background they���ve also been developing that core physics to build a new quantum computing modality. So we actually have two quantum computing companies in Fund I.

Even though that���s a 10-year-old company, there are about to be two companies, one of which will be a unicorn virtually overnight.

There are wild things bubbling under the surface that people are going to wonder where they came from.

Companies like ��� the only co-packaged optics on TSMC ��� we���ve been working on that for a long time. Now people are waking up to silicon photonics and co-packaged optics.

There are also stealth companies that are indistinguishable from magic. Some of those will come out of stealth this summer.

Is there anything we haven���t chatted about that you think is worth noting?

Barrett: It���s a sobering note, but globally there is a need and desire for sovereign capability in tech ��� in Western Europe, Australia, Canada and elsewhere.

There are extraordinary pools of capital, pension funds and Australia���s superannuation fund. Given the things we can invest in, globally the West needs to do a better job translating that capital into societal and economic value.

The safety and durability of liberal democracies depends on creating wealth and staying ahead.

We see a resurgent desire to do that in Europe and Australia. Around those pools of capital, there���s ambition. We need to drive that ecosystem globally, not just in the U.S.

The pace of innovation in Ukraine, driven by need, is indicative of changes that can be made in parts of the world less friendly to the tenets we hold dear in liberal democracies.

We can���t operate under the assumption that everybody clever lives in Palo Alto or that we can only invest in things we can drive to. We need to deploy capital globally, and we do. We���re going to do more of that.

Do you feel encouraged by the amount of infrastructure build-out that���s going to happen over the next few years? It feels like it will create a boom in all sorts of technologies because the drive for efficiency will become much stronger.

Barrett: LLMs are not the end. We���ll run LLMs on these data centers initially, but we���ll run their descendants and other more useful things on these machines and on quantum machines.

It���s going to be hard to overbuild because computation is incredibly useful. There���s no upper bound. We���re not in a Malthusian zero-sum game for resources.

We know how to make everything more productive. We know how to grow GDP arbitrarily large. But we need food, energy and medicine there, and we need to normalize the distribution of wealth.

There is unbounded abundance we can unlock if we spend capital on the right things. We know how to do much more of that than people suspect.

The fact that sensible people are considering data centers in space indicates they���re not paying attention to the things we already have in hand that can move the needle.

We do need compute in space. We need AIs in space, sensing in space, and Starlink is great. But we need to use technologies that make sense, not try to make skyscrapers out of toothpicks.

蹤獲弝け queries:

Related reading:

• Sector Snapshot: Quantum Computing Startup Investment Slows In 2026 While Public Markets Hold Strong
• Sector Snapshot: Semiconductor Startup Funding Still Running Hot

Illustration:

蹤獲弝け News spoke with co-founder , who describes the firm���s thesis as using technology to bring capabilities traditionally available to the top 1% to the other 99%.

Portfolio companies that fit that thesis include LatAm neobank ; fleet safety management startup ; , which is a tool for travel agents; , which develops agents for enterprises; and coffee chain .

The firm has a strong focus on logistics, payroll, construction and other real economy sectors.

It is also exploring vision models and world models ��� the equivalent of LLMs for visual understanding. If AI could truly understand every pixel and atom in a construction site, that will unlock robotics, Ajao said.

Manufacturing intelligence is another area of interest.

Ajao asks: Can AI understand manufacturing processes the way LLMs understand text, whether it’s perfumes, pharmaceuticals, chips or concrete, for real economy applications?

Stage focus

The firm invests at seed through Series B. From the early-stage fund, Base10 plans each year to make 10 to 15 seed investments, and two to three at Series A. The Series B fund, roughly equal in size, will make three to four investments each year.

Base10 is research first, spending months analyzing sectors before investing.

���We might ask what IT support firms look like when you have AI, or what the software stack of the modern restaurant is,��� said Ajao.泭 The firm tries to meet every company globally operating in that space. It spends roughly 50% of its time with companies that are not fundraising, with 90% of investments made due to its research.

For the recent batch of 160 companies, the firm only meets with those that align with their research. Along with too much happening, founders are better prepared.泭 For the firm, being informed allows them to get to conviction fast.

Base10 has created an internal AI system called Base11 to classify companies, and automate research. However, ���the actual decision-making and winning is more human than ever,��� said Ajao.

That means spending more time understanding founders as people and talking to customers, said Ajao.

Competition among venture firms is also higher than ever. ���It forces all of us to articulate a lot more why someone should partner with us,��� he said.

Through its Advancement Initiative, Base10 donates up to 50% of carried interest to underfunded colleges and universities to support financial aid.

Related 蹤獲弝け query:

Illustration:

Size increase

Median seed round sizes have been climbing since 2023, 蹤獲弝け data shows, with the median U.S. seed round last year now standing at around $3 million. That���s 3x larger than it was in 2018.

The upper quartile median last year was around $5.6 million ��� more than double what it was in 2018泭 ��� and the lowest quartile was $1 million. (Although, underlying those medians is a much wider range of deal sizes.)

At seed, ���What we see is everything from the inception stage, which is typically $3 million to $5 million, unless it’s a truly unique and obvious founder, all the way through to $8 million to $10 million-plus rounds,��� said , managing partner at , one of the earliest institutional Bay Area seed funds founded in 2004.

McLoughlin noted that the typical check size his firm writes for a seed round has almost doubled from 18 months ago. ���We’re still trying to buy at least 10% ownership, ideally more, and our average check has grown from $2.5 million or less, to $4.5 million,��� he said.

The speed at which those round sizes have accelerated is mind-bending, he said. ���If you’d asked me 18 months ago, would the $8 million to $10 million-plus seed round become de facto, I would have said you were crazy���

Series A rounds have also grown in size, per 蹤獲弝け data. Last year, the median U.S. Series A deal was $15 million, with the upper quartile at $25 million and the lower quartile at $7 million. That trend has continued into 2026, with median Series A rounds moving still higher.

Longer time frame to Series A

But while companies that are funded at the seed stage are typically raising larger checks, they���re also taking longer to move on to Series A and face lower odds of graduating to that phase at all, 蹤獲弝け data shows.

Since 2023, U.S. startups have been taking longer to raise a Series A round following an initial seed round of $1 million and over, per 蹤獲弝け data, with that time frame now stretching to more than two years.

That continues a general upward trend since 2018 of startups taking longer to raise a Series A round after seed, with notable exceptions in the previous peak funding years of 2021 and 2022, when the timeline shrunk by six months.

The threshold for raising a successful Series A is no longer $1 million in annual recurring revenue, said McLoughlin. In the AI era, startups are expected to show $2 million to $3 million ��� even $4 million ��� in ARR as proof that the business has the momentum to scale, he said.

���When you’re fundraising for your [Series] A, you’re not in competition with the startups you deem to be competitors,��� said McLoughlin. Rather, he noted, you’re in competition with every other deal floating around in the venture ecosystem ��� not just the partner you’re talking to and their ability to do the deal, but what the entire team is doing, how far along they are, how far ahead of pace they are on their investment cycle, and whether they’re being pushed to only do things that truly look like breakouts.

Fewer graduates

Since 2021, drastically fewer companies that raised an initial seed round of $1 million or more have progressed to a later-stage funding or exited, 蹤獲弝け data shows.

Through 2020, companies that raised a seed round of $1 million-plus had a typical graduate rate of 55% or higher.

Since then, graduation rates appear to be falling dramatically. Of the companies that raised a $1 million-plus seed round in 2023, only 24% have progressed further, 蹤獲弝け data shows. For the 2024 cohort of seed-funded companies, that���s even lower: just 16%.

While these cohorts are staying at the seed stage longer, still McLoughlin predicts, ���we’re going to see the mortality rate from seed to A will be much, much higher.���

As the dynamics of seed funding change, investors are being forced to rethink their portfolio strategies ��� adjusting to the right number of bets, reserving enough capital for follow-on rounds, and deciding whether to invest earlier or in larger seed rounds with potentially less ownership.

���We’ve also got to be comfortable with this notion that there will probably be more early outcomes or failures in the portfolio, but if we do our job well, the big outcomes will be bigger than they’ve ever been before,��� said McLoughlin.

Related 蹤獲弝け queries:

Related reading:

• Data: The Seed Funding Boom Is Concentrating Capital In The San Francisco Bay Area
• Seed Funding Is Bigger Than Ever ��� And Harder To Get

Clarification: The rates of graduation from seed stage in 2023 and 2024 have been updated.

Illustration: