Updates
Magma Summit 2024 Wrap-Up: Lima, Peru
We hosted Magma founders, LPs, and partners in Lima, Peru, for Magma Summit 2024. Here's a rundown of all the activities from each day at Magma Summit 2024.
Magma LPs,
Magma turned 10 in Q1 2024, which feels like a significant milestone. In 2014, Francisco and I started Fund 1 in Chile when the VCs only invested ~$250M per year in all Latin America. I didn’t go full time until 2015 and was the only full time Magma team member until 2017.
Fast forward to 2024, while VC is down from its 2021 peak of nearly $20B, the entire ecosystem has matured into a functioning industry. Magma is a distributed team in Mexico, Colombia, Argentina, and Chile, and we’ve backed founders in nearly every Latin American country. We’re on our 4th fund and have invested ~$75M into ~120 Latin American companies.
Thanks to everyone who’s supported us on our journey, especially Francisco who made a bet on a gringo living in Chile back when Latin American VC was scary. Special thanks to all of the founders who have made us part of their journey. Although Latin American tech has come a long way in 10 years, it still feels like day one for helping founders solve Latin America’s biggest problems.
Anecdotally, it feels like there’s more momentum in the market and more of our companies are successfully fundraising, but the data does not back up a VC warm up. In fact, Q1 2024 Latin American VC funding fell to $576M according to Crunchbase, a 17% decline from Q1 2023 and a 39% decline from Q4 2023, reaching an annualized $2.3B.
While down from the peak, $2.3B would still be ~9x higher than when we got started in 2014. We expect 2024 to be closer to the $4B from 2023 than to the $2.3B run rate from Q1 2024, as larger rounds get announced this year.
At the same time, we’re seeing a trend, confirmed by the data, that bridge rounds are up. Bridge rounds are where VCs invest at the same valuation as the last round, without a markup. You can see the biggest difference at Series A and Series B, but all stages are affected.
The chart from Carta shows that ~22% of 2021 Series As were bridge rounds, and it nearly doubled to 43% in 2024.
17% of 2021 Series Bs were bridge rounds, and it’s now 38% in 2024, up 123%. The chart shows the slow motion reset happening in real time. We have portfolio companies that grew 2-5x from 2021 to their last round and the valuation stayed the same as the companies grew into their valuations.
Extensions between primary rounds continue to attract VC attention
Our portfolio has seen its share of bridge rounds, but even bridge rounds require good unit economics, decent growth and a business that makes sense. Except in AI, which is still behaving like it’s 2021. Read on for more on AI later in this letter.
As we suggested in previous letters, Latin America seems to be lagging the rest of the world’s Q1 2024 VC comeback. Although Latin America’s lag could be noise in the data, it’s more likely that everything in Latin America takes longer: the boom took longer to get to Latin America, and it's taking longer to reset.
Wall Street has been pricing in 3 US rate cuts, which haven’t happened, and the stock market is behaving like they have. Now, Wall Street consensus is 2 rate cuts in 2024, with the first one in September. We continue to be skeptical about rate cuts this year.
With Q1 US inflation at 4.5%+, more than double the FED’s 2% target, there’s even a non zero chance the next move could even be a rate hike. While unlikely, it's important to at least acknowledge it could happen, which would put a damper on Wall Street and potentially the economy as debt continues to reset.
We continue to believe founders should be more open minded about this market. There are many risks on the table, between a stock market at or near highs, inflation picking back up, US and Mexican elections, China’s economic issues and geopolitical risks in Europe and the Middle East that haven’t changed much.
Higher rates make it harder for VCs to raise money for new funds, leading to smaller funds, and less capital available for founders, so we believe it’s prudent to take a bit less risk and plan for the possibility that rate cuts don’t happen this year.
In 2022, Lux Capital’s Josh Wolfe predicted that the VC fund world would contract significantly. The biggest funds would get bigger, differentiated funds with a track record would do well, but everyone else would suffer and many would go out of business. QED’s Frank Rotman wrote his essay The Three Body Problem where he predicted that there would only be 4 ways for VCs to thrive: Scale (big funds), Late Stage Generalist, Solo Capitalist and Non-Consensus Alpha.
In 2024, their predictions are coming true. According to OpenLP and Pitchbook, only 12.4% of first time fund managers have been able to raise a second fund, down from a pretty steady 60% from 2013-2019.
Total dollars invested into first time VCs has gone down, and the first time VCs that raised in the 2021 boom will need to come to market in 2024/2025 for their second funds.
According to Pitchbook, LPs are concentrating capital into the largest VC funds, up to 45% from a low of 10% during ZIRP and according to OpenLP, “2023 was the second consecutive year established managers (Fund 4+) took more than 70% of total committed capital.”
It’s likely that many of the new ZIRP-era funds will struggle to raise future funds. We think this retrenchment is likely to lead to better returns as an industry, but also in the funds that outperform. Fewer dollars chasing the same deals likely means higher returns. We think it will take another 2-3 years to shake out, since VC funding resets even slower than startups do.
According to Theory Ventures’ Tomasz Tunguz’s 2024 AI report, AI investment now represents 20% of all VC dollars invested, up 4x in 24 months.
Up 4x in 24 months
He projects AI investment will reach $70B-$80B+ in 2024.
Up 2.6x from 2023
But just as we’re seeing concentration into the 5 biggest VC funds, we’re seeing concentration into the top 3 AI deals, with xAI, Anthropic and CoreWeave making up 60% of all AI VC investment so far in 2024, and 53% of AI VC dollars were invested in the San Francisco Bay Area.
Heavy Concentration in a Few Names
We expect this trend to continue, as AI continues to move extremely fast. The battle between open source models like Meta’s LLaMA 3 vs. closed models like OpenAI’s will be a big global battle, with open source seemingly able to catch up to close source competitors quickly.
We believe nearly all startups will need to incorporate AI, but generally in ways that use Latin American distribution to get AI into people’s hands more quickly, rather than building foundational models in Latin America.
Although not directly in our wheelhouse as an early stage VC fund, we are very interested in electricity. Up until 2022, electricity was a boring business. US electricity capacity didn’t grow much from 2006 to 2021.
Utilities didn’t have to invest in growth. They spent time trying to green their infrastructure, but not much else. For example, in 2021 Virginia’s power utility Dominion Energy, which powers a big share of US data centers, predicted it would need to grow capacity by less than 10% in total by 2036. In 2023, they said they’d need to grow 2x in the next 15 years. Many experts think their estimate is low and could be as much as 3x.
Electricity is very regulated, has long lead times and doesn’t respond quickly to new demand. Utilities are out of practice, since they haven’t had to grow capacity in 15+ years.
AI data centers are eating power all around the world, and according to a guest on Odd Lots, companies like Google, Microsoft and even Amazon could really be thought of like power companies: They transform natural resources into power, which they use for data centers, compute, AI and infrastructure that runs our entire world. According to JP Morgan:
“Alphabet, Amazon’s cloud arm (AWS), Meta and Microsoft consumed 90 terawatt-hours (twh) of electricity in 2022, as much as Colombia. And that was mostly before ChatGPT touched off the AI revolution in November 2022. The ensuing hoopla led the International Energy Agency (IEA) to predict that data centers (including those dedicated to AI and equally energy-hungry cryptocurrencies) will eat up more than 800twh globally in 2026, double the amount in 2022.”
For example, Amazon spent $650M to buy a datacenter next to a nuclear power plant that had a long term contract for 480MW of power, enough to power between 300k-400k homes. Meta will spend $40B on capex buying AI chips and data centers, which is close to Saudi Aramco’s $50B capex spend as the largest oil company in the world. According to Collabfund, "The US data center footprint is expected to absorb 35 gigawatts of electricity by 2030, more than twice the total demand in 2022—growing compute is draining not only electricity but also water."
Governments, utilities and even private companies are trying to figure out how to meet demand not just for data centers, but also for air conditioners, washing machines, electric cars and more, especially in emerging markets.
In Latin America, power demand has steadily increased, and companies are used to building new capacity, unlike many US utilities, which are out of practice.
Mexico and Colombia are already being pushed to the breaking point, with Mexico experiencing an extended heatwave and drought and Colombia a drought that is hindering its hydroelectric power generation that represents ~70% of Colombia’s electricity. Ecuador had power cuts after 13 years because of the same drought. Mexican politics are not helping, with ~2000 MW of completed solar and wind projects unable to come online until the government approves them, plus another ~5000 MW stalled in various stages of development.
Companies are building data centers and AI infrastructure around the world, chasing cheap, reliable and renewable energy. Latin American governments and private companies have an opportunity to take advantage of this trend, especially as the US regulatory framework makes it hard for power utilities to grow.
We expect more investment from government, big tech and private power companies like Vistra in the US, which now supplies 20% of Texas’ power, to increasingly operate in the market.
Many Latin American countries could benefit if they are able to keep up with growth in demand. We’re interested to see what happens and if there’s any investment opportunities in Latin America’s energy sector, whether it's from software, renewables, data centers or something else.
Even though VC funding has dropped, we are excited by the number of high quality founders we’re meeting. We believe there’s a very good chance the 2023/2024 vintages of VC funds will outperform, as only the truly committed founders are starting companies now, and there’s less available capital to fund them, lowering the entry points to active funds like Magma. At the same time, we’re likely in the next big wave of progress in AI, which could remake software and service across industries.
We’re advising companies to be vigilant about the market, start investing in testing out how they can add AI to their products, and continue to solve Latin America’s biggest problems. If they continue to execute, there will be money available, and big rewards for those who are able to solve them.
Best,
Nathan, Pedro, Mak, Francisco and the Magma team
