The Scene
In 2019, SoftBank launched a $5 billion Latin America Fund to invest in technology startups, an unprecedented capital infusion equivalent to all the venture bets in the region in the prior two years. Then, it followed up with another $3 billion fund in 2021.
Those massive investments changed the startup landscape in Latin America. But when the global economy began to slow and interest rates rose, many young firms in the region couldn’t justify the lofty valuations that firms like SoftBank helped prop up.
The crisis didn’t scare the Japanese conglomerate away.
The company, run by famed entrepreneur and investor Masayoshi Son, is pumping more money into the region this year, including a new investment alongside Microsoft into AI conversational marketing platform Blip. The deal could create a strong global AI contender, because Blip has access to information through users of WhatsApp, the dominant messaging app in the region and in places like India and Southeast Asia.
Alex Szapiro, SoftBank Group’s head of Brazil and managing partner, and Juan Franck, managing partner at SoftBank Latin America Funds, told Semafor in an interview that they’re taking a more conservative approach, but remain bullish on the region’s prospects.
While Latin America has a reputation as an AI laggard, they argue there are emerging use cases that show the region is moving faster with the technology than widely believed.
They discussed with Semafor what AI developments could shift the dynamics of Latin America against leading players like the US and China, their appetite for compute data centers, and where they intend to plant their next investment flag.
The View From softbank
Reed Albergotti: SoftBank has a reputation for coming in with lots of money and flooding the Latin American market. What do you say to that criticism?
Alex Szapiro: SoftBank actually crossed a lot of lines for the good, inventing a market that didn’t exist in the region and cutting $30, $40, $50 million checks for a 10% or 15% position in those companies. And not only to invest in the first round, but also in the following rounds, so those companies could develop.
A lot of funds came after SoftBank. Some stayed, but I think a majority, which we call tourist funds, when the crisis came, they left.
Juan Franck: In 2021, we all made mistakes in terms of valuation, which got adjusted in 2022. But in 2021, there was an average annual investment of $17 billion, and we were a fraction of that. I think we were part of the craziness of 2021, but by no means were we the trigger, or the ones to blame for that.
Some say Latin America is not adopting generative AI as quickly as other parts of the world. Do you think that’s true?
Alex: There’s no question that there are a lot of things going on, more on foundational models, that I think is going to be very hard for us to see in LatAm. The US, Asia, Israel, and some of the other countries are ahead of their time in terms of everything that has been developed. And companies like Perplexity, or others that we’re very close to, like OpenAI.
On the foundational models, I don’t think we’re going to see a lot of activity in the region, given all the CAPEX necessary in terms of processing, in terms of data centers. Although there’s one great thing about LatAm, which is the cost of energy.
But having said that, there are things on the application [front], that’s where I think we’ll see a lot. Let me give you some examples.
We invested in Asaas, a company that handles collections for [small- and medium-sized businesses.] For SMEs, one of the most important things is getting paid. If you charge a 1% take rate on all the invoicing of the company, it is a great business, but it’s also really great value for an SME, which sometimes has delays on their invoices. Brazil, Mexico have interest rates of about 10%, 12% a year, which is 1% a month. So just imagine that a month’s delay costs you at least 1%. This is the kind of trade-off that we need to think of all the time.
Another example is a company called Blip. It’s what we call a conversational platform. In LatAm, 99% of the things that happen in the region happen on WhatsApp, like WeChat in China.
This company has done a fantastic job serving big corporations and SMEs on all the communication that the companies have with their customers, starting more on the customer service area. Once they automated this, they said, ‘What if we also do e-commerce on WhatsApp?’ where basically, if you’re talking to Nestlé and you want to buy an espresso machine, you just start a conversation. You say, ‘I’m looking for a machine. It is for my home. I like this color,’ and then the alternatives for you to choose pop up, then comes the link, and you can pay.
I’m just talking about the tip of the iceberg in terms of the things they’re developing, but this is an example of a company that has access to data because they are allowed to see all the conversations between the corporations and their customers. [WhatsApp owner] Meta isn’t allowed. So even for Meta, this is a great asset. [Blip] is now considered one of the preferred partners in the world in terms of what they are developing for WhatsApp.
I’m just putting some examples where everybody thinks LatAm is behind the game on AI, but there are a lot of things that we only see in the region that AI is actually solving much faster than probably other parts of the world.
I thought they [Blip] had some access to communication with customers that other people didn’t have?
Juan: That’s exactly it. Companies are interacting with customers through WhatsApp. And Blip builds on top of those interactions between the company and the customer, so you’re exactly right. They have proprietary access to that information. It’s not their information. It’s the company’s information, but they help the company digest and analyze that information to convert it into intelligence.
Are they able to use it in the aggregate, in any way, or do they have to do it in a one-off way with each individual company?
Alex: They do this with each individual company. They use anonymized data; they get it back to their models to improve the models; they do have some walls.
I’ll give you an example: América Móvil is one of their customers. It is the biggest cell phone carrier in the region, so the quantum of data that goes back and forth is gigantic. And it trains the overall Blip capacity to be used elsewhere. But of course, all the data is anonymized.
So you start improving communication, which is a cost-driven initiative, using AI to bring more efficiency. But as you learn and as you teach your system, you go more to the conversion side, which is using AI to increase sales.
One good example for that. This is the first time in history Google and Meta are connected together. How so? Because for Google, what matters is the return on advertising for the customers, which means the more they can convert the market, the more they can charge.
And they’re using Blip, together with some of the customers, to say: If you go to a Google ad, you read the ad, in some cases you will see the WhatsApp symbol. You’re going to click there, and the chances that you see that ad and you start talking to the company to buy a product is much higher. In a way, Google is using Blip and Meta to increase their conversion.
You might say those two companies compete with each other, but it’s a way for Google to charge more on the auction process for the ad. Going back to your question about how everyone thinks LatAm is behind the game, there are super interesting things here because of how advanced we are in some of those platforms to leverage and be used as a lab, so that later on, those tests in Brazil or Mexico will be used in other countries.
You mentioned earlier the cost of energy being lower. Is there any talk of trying to do these big, accelerated compute data centers in the region?
Alex: We are probably layers behind that. The discussions we are seeing now are about a lot of investment in data centers, not necessarily the supercomputers. The question comes to: As you build your data centers in places where energy has a much lower price, what are you going to put on the ground? Do we really need supercomputers to process things very fast? Or can you process smaller models during the night, and bring this data back to where it needs a latency type of effort?
What we see in Brazil and Mexico comes two or three steps earlier, which is a ton of investment on real estate and data centers, not only from startups, but also if you go to the Magnificent Seven, you see the announcements from Google, from Meta, from AWS. I think AWS probably is the biggest, doing huge investments in Mexico and Brazil, building data centers, because that’s much closer. If you look at Brazil as an example, about 65% of the energy generated in Brazil comes from hydropower, so not only is it cheaper, but it’s cleaner.
What’s the investment thesis there? The accelerated compute data centers have more people, higher level people working on them, because of all the complex networking. But for traditional data centers, they don’t spin off much in terms of innovation.
Alex: On the way we run our LatAm business, we are more focused on the asset-light type of investments. That’s why we have a lot of fintechs, health techs, ed techs, e-commerce companies, all those things with an AI type of focus. Data centers are not part of what we are really spending time with, but certainly we see a lot of activity from PE, and from infrastructure funds and sovereign funds.
When you talk about data centers, developing emerging markets are almost like battlegrounds between East and West, spheres of influence. You’re seeing it in the Middle East, in Africa. Is China coming in wanting to do this stuff as well, or is it really a US and Western play down there?
Alex: We have 77 companies in our portfolio in many different stages of investment. We deployed close to $8 billion in the region with the tech AI vision. Some of those companies are now starting to be in the moment of exiting. We always say a good investor is the one who invests, but also who disinvests, who can sell the assets.
The fact is that not only are we seeing a lot of companies from Europe and the US [show] interest in our portfolio, but we also see a lot of companies from Asia. Deals won’t happen overnight, but I think it shows a lot of interest of the usual suspects in Asia and China coming to the region, looking at the opportunities and exploring, sometimes participating. Some of our companies already have them on the cap table. That is to say that the neutrality that we have in the region, I do think is a big advantage.
You really can’t take Chinese money in the US at this point, those days are over.
Alex: A good example of that is, while EV sales are going down, with some countries putting tariffs on the Chinese manufacturers, BYD reported the biggest increase in the [LatAm] region in the last month because those countries are more open.
Juan: We started this fund in 2019… when it was a region that was competing for capital against other emerging markets that also had interesting growth opportunities, where industries were generally underpenetrated from a digital penetration perspective. Today we still have that along with other emerging markets, but we also have this geopolitical stability that puts us in a very good place to continue to attract both growth equity capital, but also private equity and strategic capital. Investors and strategics from China, from the US, from Europe are all spending more time in the region because it’s exciting from a scale perspective, but also just interesting from a relatively macro geopolitical stability perspective.
You did say in 2019, you planted this flag and said, ‘We’re going to do $5 billion in LatAm,’ and then there was another $3 billion. So that’s $8 billion. What now?
Juan: The difference there is, as of 2022, the LatAm funds were consolidated into the Vision Fund structure. All the investments that occurred after 2022 are pulling from the same pool that Vision Fund II is pulling capital from. Before that $5 billion initial commitment in 2019, the region, on an annual basis, was seeing an average investment pace of around $1 to $2 billion per year. That skyrocketed in 2021 to around $17 billion [and] obviously came down in 2022; 2024 is trending to be around $6 to $8 billion of average investment pace. That’s this steady state that hopefully will stay going forward. And in that context, I think we will continue to be, if not the [one], one of the most active growth equity investors in the region.
If you’re the biggest one in the region now, does that make it easier to get into the deals that you want to get into?
Juan: When you have people on the ground, you are closer to the entrepreneurs, and hopefully you’re also establishing better relationships with the ecosystem. That gives you the chance to be the first port of call for any interesting company that’s thinking about raising.
A very important point as well is that our team is composed of both operators and investors. We bring both those profiles to the table when we propose ourselves to be potential partners of interesting tech companies. Alex has vast experience in a bunch of different operational roles and really interesting tech companies in Brazil. I come from the investment world. And that complementarity at the managing partner level also flows down to the rest of the team.
It’s more than just capital. It’s people on the ground that can also have a nuanced view. They have a different perspective compared to the traditional investors.
A lot of funds talk about being long-term committed to the region. The last two years have helped people prove if your long-term commitment is there, during the good, but also the bad of 2022 and 2023, where we were on the ground having very, very uncomfortable conversations with founders about optimizing OPEX (operating expenses) structures, about extending cash runways, about potentially participating in follow-on rounds. Coming out of that — in what could be a better 2025 in the context of decreasing interest rates and potentially capital being more available — is that we were in the trenches during that whole time with them.
That is a key differentiation that the past five years have given us, versus a lot of investors that were parachuting in and out, especially in 2021, without having a local presence.