How Salesforce Chooses the AI Companies It Wants to Buy
- Salesforce’s acquisition logic pairs scarce capabilities with distribution at scale. Somorjai pitches access to 15,000 sales reps and faster global reach—“as widely as possible and as quickly as possible”—but buying and integrating a company can still take six to nine months. Cheaper software development therefore strengthens the need to compare acquisition timelines against internal delivery.
- Outcome pricing gives Salesforce a way to capture AI value as customers need fewer seats. Somorjai reports that Fin deployed internally in 14 days, handles over 40% of help.salesforce.com traffic and delivers 65% resolutions without a human. Through “outcome-based pricing,” software can capture more of the labor budget while reducing customers’ overall costs, as Immerman frames it.
- Slack’s acquisition thesis gained an AI role Salesforce had not anticipated. Somorjai says revenue is almost three times its acquisition level and Slack is Salesforce’s fastest-growing product, though integration initially took time. The original distribution thesis—“great technology that was starved for distribution”—now benefits from an interface where humans, applications and agents work together.
- Salesforce Ventures seeks future enterprise markets, with acquisitions representing a small subset of its investments. Salesforce has acquired 29 of more than 750 companies backed; its $1 billion AI fund, established at the beginning of 2023, sits alongside an expanding interest in robotics. Asked about investing in potential Salesforce competitors, Somorjai says, “I think we probably should.”
- AI CRM remains a disagreement between incumbent confidence and startup adoption potential. Somorjai has yet to see enough novelty to support a large multibillion-dollar challenger and argues, “There are too many of them.” Immerman counters with faster deployment, adoption further downmarket and Stripe’s example of growing alongside initially small customers.
- High private valuations can leave companies with too few viable exits. Somorjai estimates “probably one percent or less” go public and says the buyer universe becomes very small above $10 billion; a reckoning is “inevitable,” although he does not know when. Immerman pushes back that today’s largest outcomes are substantially bigger, while Somorjai distinguishes rare generational companies from ordinary businesses receiving similar valuations.
- ARR growth can conceal the problems that kill an acquisition. Somorjai focuses on customer attrition, financial controls and security, which has veto power across Salesforce’s 14 diligence functions. One penetration test reached source code, the CEO’s laptop and employee information; the exposure was “unrecoverable.”
1. Salesforce buys time, reach and entrepreneurial talent
- Immerman puts acquisition activity during Somorjai’s tenure at nearly $100 billion, with 12 deals in the past year. Customer requests and Benioff’s “beginner’s mind” inform three tests: strategic fit and customer success; whether Salesforce can accelerate the target with compatible technology and people; and value for shareholders and customers, including valuation and time to accretion. Somorjai calls the last test hardest because “everything is expensive.”
- The distribution pitch is concrete: Salesforce has 15,000 sales reps and more than 8,000 ISV partners. Founders can partner, sell or keep building, but recreating sales coverage across small businesses, enterprises and global markets takes years, additional fundraising and dilution. Somorjai’s question is whether to undertake that work or reach customers “as widely as possible and as quickly as possible.”
- Cheaper development does not settle build versus buy. Product teams map three-year roadmaps against available expertise and scarce AI engineers; an acquisition can take six to nine months from idea through integration, so internal development might still win. Listen Labs supplied the opposite case: Somorjai’s team launched a survey in ten minutes with “no enablement whatsoever,” then used the product to survey its customers during diligence.
- Integration also refreshes management: acquired founders have become Salesforce’s chief product officer four times. Somorjai’s example is David Schmaier’s Velocity, which brought six industry verticals; Salesforce subsequently added seven, reaching 13. The acquisition changed how Salesforce understood industry-specific use cases, while Schmaier moved through chief product officer to chief strategy officer.
2. Outcome pricing connects AI efficiency to Salesforce revenue
- Salesforce’s headless platform raises Immerman’s seat-count question: customers using AI may need fewer seats. Somorjai points to consumption pricing—already used by MuleSoft—and a shift toward paying for outcomes, such as resolved support requests. Immerman’s economic framing is that software can capture more of the labor budget while lowering the customer’s overall cost of selling.
- Service Cloud is almost $11 billion in AOV and rests on a 2008 acquisition, but Somorjai says Salesforce could not build the new combination of AI applications, outcome pricing and fast deployment quickly enough. Fin supplied that capability and an AI team in Dublin. He also values founder Eoghan’s willingness to challenge the incumbent: he is “really trying to shake us up.”
- Fin’s average implementation is 16 days; Salesforce’s internal pilot took 14. Somorjai reports that it now handles over 40% of help.salesforce.com traffic and delivers 65% resolutions without a human. Qualified extends labor substitution into sales: its Piper agent answers website visitors through a lifelike avatar and books meetings automatically. It “can replace your sales development rep organization, essentially,” he says.
3. Slack’s distribution thesis gained an unforeseen AI upside
- The 2020 Slack thesis was distribution and a common interface for Salesforce products; agents came later. Slack was “losing to Teams,” Somorjai says; a bad quarter and stock decline helped trigger discussions about combining a beloved product with Salesforce’s reach.
- Somorjai acknowledges the outcome was uncertain: pandemic diligence happened largely through Slack, without many face-to-face meetings, and integration “took a little bit of time” to hit its stride. He now says revenue is almost three times the acquisition level, roughly five years and a quarter later, and calls Slack Salesforce’s fastest-growing product. His explanation: “great technology that was starved for distribution.”
- The AI payoff came later: Slack is “the perfect conversational interface for AI,” in Somorjai’s view. Immerman calls it the “unofficial org chart,” containing reporting relationships, interactions and decision-making context. Somorjai connects that work history to richer Slackbot answers; the collaboration layer originally intended for humans and apps now also accommodates agents.
4. Ventures follows future enterprise value beyond software
- Salesforce Ventures began in 2009 with ecosystem investments such as DocuSign, nCino and Twilio, plus regional services partners. The mandate widened to Snowflake and cybersecurity through Wiz, then a $1 billion AI fund at the beginning of 2023. Somorjai names Anthropic, Mistral and Cohere, ElevenLabs in voice and Runway in video; the through-line is future enterprise value.
- Physical AI pushes that mandate beyond software. Somorjai highlights Figure AI, Physical Intelligence, Dyna Robotics and Zipline’s drone delivery, linking robotics progress to improved dexterity and mobility. Immerman adds that early signs of scaling laws and more vertical go-to-market strategies are beginning to translate into revenue growth. Customer introductions and operating advice remain Salesforce Ventures’ main pitch to founders.
- Asked whether he would back a future Salesforce competitor, Somorjai says, “I think we probably should.” Staying close can lead to a partnership, an acquisition or neither. Of more than 750 companies funded, Salesforce has acquired 29—a relatively small subset. The ventures team’s thousands of annual company meetings also feed information about traction and technological direction back into Salesforce’s decisions.
5. AI CRM challengers face a disputed opening
- Somorjai’s skepticism is sharper on new CRMs: “I have yet to see something so novel that it’s going to turn into a large, multibillion-dollar company in that space.” Existing products are strong, Salesforce keeps improving, and small differences are insufficient. Immerman counters with Lightfield and Monaco, then suggests fast deployment and bottom-up adoption could create an opening further downmarket.
- Somorjai concedes, “I could see that,” but calls it a small opportunity. Immerman’s stronger counterexample is Stripe: start with small companies, then grow alongside them. Somorjai does not resolve that analogy; he returns to market crowding—“There are too many of them”—and says venture capital should fund new, important categories rather than numerous niche CRM variations.
6. High valuations can exhaust the buyer pool
- The valuation objection is exit arithmetic: Somorjai estimates “probably one percent or less” of companies go public, while valuations above $10 billion leave very few potential buyers. He calls a reckoning inevitable but explicitly says, “I don't know when.” He also thinks unsustainable valuations will hurt employees; maximizing the next-round price is “usually not the best decision.”
- Immerman tests whether power-law outcomes justify the pricing, positing Anthropic above $1 trillion and citing SpaceX at $2 trillion. Somorjai calls such businesses “generational companies” and “very rare.” Salesforce invested in six Anthropic rounds because its first meeting with Dario and Daniela convinced the team they were exceptional; he contrasts that conviction with ordinary ideas receiving $10 billion-plus valuations.
- Immerman argues that today’s top-1% outcome is larger than a decade ago, citing Wiz and Cursor. Somorjai agrees he is an optimist but sees thousands of companies that don’t make it; many come to Salesforce seeking acquisition, a part of the industry the press doesn’t cover. Those unpublicized cases, he argues, belong in founders’ fundraising and valuation decisions.
7. Churn and security can outweigh headline ARR
- Deal trouble often appears after price and major terms are agreed, during deep diligence. Somorjai flags missing customer-contract data, weak financial controls and fixation on ARR without understanding attrition. High churn can indicate unhappy customers and product-quality problems; ARR growth alone does not answer whether customers will renew. Organized data can materially accelerate the process.
- Security has an explicit veto across Salesforce’s 14 diligence functions: it can say, “I'm killing this deal.” In one abandoned transaction, penetration testers accessed the company’s source code, the CEO’s laptop and employee information. Salesforce shared its findings, but the exposure was “unrecoverable.” Somorjai’s preparation advice is to red-team the technology before the buyer runs its own tests.
- Not every vulnerability kills a deal. Salesforce can distinguish P0, P1 and P2 problems, require some fixes before signing and others before closing. Somorjai says the outcome depends on how severe the problems are.
Full transcript
We’ve built an incredible Service Cloud business. It’s now almost $11 billion, and the foundation of it was an acquisition.
How do you describe Salesforce as a platform today?
At its core, it’s delivering a trusted foundation for all of your enterprise applications. You can interact with it through Slack. You can interact with it through CloudForce.
Software development is faster and more affordable than it’s ever been. In this era of software, what is worth actually buying?
There will be a reckoning. A lot of these companies have valuations that aren’t sustainable, that they won’t be able to grow into. We’ve invested in more than 750 companies. We’ve acquired 29, so it’s still a relatively small number of companies. But if we ever see something really exciting, I think we want to stay close to that company.
If you were going to start a company today, and it was going to attack Salesforce head-on, what are the openings that you see for it?
John, it’s great to be with you today.
Thank you for having me. It’s wonderful to be here.
You’ve been at Salesforce now for over 20 years. Marc Benioff is a Mount Rushmore-style figure for cloud software. He led the transition from on-premise to cloud. He’s an incredible marketer. At Dreamforce this year, he frame-mogged Jensen, Dario, and Sam. I’m curious: You’ve had a long career with him. What’s it like to work with him?
I would also add “visionary” to that point as well.
Yes, yes.
Marc is quite a visionary, and he really can see where the industry is heading. That’s how he helps the company transform itself. When he thinks about growing the business, he thinks about organic innovation, but also inorganic innovation. That’s really been the key to how we’ve grown the company: marrying the two together. Acquisitions play a huge role in that.
When Marc sees that the world is moving in a certain direction, he’ll point me and my team to go figure out, go find the best asset to buy in that category, and come up with those ideas all the time. I’d say the other thing that’s very unique about Marc is his level of curiosity and how he thinks about things with a beginner’s mind all the time. He asks that of us as well, and I think it’s really helped us find some remarkable companies to bring into Salesforce, and some really exceptional entrepreneurs.
With that in mind, I really like that idea of a beginner’s mind, the curiosity. How do the two of you approach acquisition and investment ideas together?
One of the things that’s always been the hallmark of Salesforce is that we listen to our customers quite a bit. We get a ton of feedback about what customers want to buy from us, where they feel like our products are missing something, and where the gaps are. We take that and work with our product teams to identify the best opportunities. That’s certainly true on the acquisition side.
On the investment side, my Ventures team meets with thousands of companies every year. I think we get a very good sense of where there’s traction and interest, and what areas of technology are really growing the fastest. We can bring all of that feedback and learning back into the company and, I think, make better decisions as a result.
You alluded to Marc as this big visionary, and of course he is. How do you and your team think about balancing Marc’s big-vision mindset with strategic and financial discipline? Maybe there have been times when you’ve had to tell Marc no on his big, grand idea. I’m curious how that all comes together.
We have an internal framework that we follow, which we’re pretty disciplined about. We look at 3 elements. The first is: Where is the strategic fit, and how much customer success can it drive? The second is: Can we accelerate the business? Will it fit into our technology stack? Is there good cultural alignment with the team? Do we see the talent there as people who will work well within Salesforce?
The third part, which is probably the trickiest in this day and age, is: How can we deliver value to our shareholders and to our customers? What’s the timeline for value accretion? Is the valuation reasonable? That last point is probably the hardest right now, because everything is expensive.
We’re in a unique moment in time, and we should use some time later to chat about that.
Yeah.
Maybe to take a step back: Salesforce is a really large company today. You’re much more than CRM now. How do you describe Salesforce as a platform today?
I think it’s always been a place where customers can go to buy technology to help them interact with their customers in new ways. That’s been foundational from the beginning. We think of it as a platform that, at its core, delivers a trusted foundation for all of your enterprise applications. Then you can interact with it from different surfaces. That’s why we made Salesforce headless, so you can interact with it through Slack, through CloudForce, and through other places. I think it’s as broad as a company wants to make it: a place where you can really deliver a lot of enterprise value.
Absolutely. I think it’s been really cool to see you guys leading the charge on the headless transition. As we all enter this brave new world with AI and new interfaces—
Mm.
You referenced Claude; there’s obviously ChatGPT, and a broad swath of them today. You’re able to access Salesforce via them. A good portion of your customers may be aspirational about requiring fewer seats in the future because they’re able to accomplish more work today with AI than they previously needed to. Under this transition, how do you think about Salesforce capturing value in this new world?
First of all, we have different ways to price our products. Most importantly, consumption is a big part of how we price. Many of our products are consumption-based. MuleSoft was always consumption-based, for example.
What we try to do is make sure that we’re delivering value that our customers really appreciate and want to continue to pay for. Whether they pay on a per-seat basis or on a usage basis doesn’t really matter. The key thing is making sure that they’re happy and successful and will come back. One of the critical things about where the technology industry has been moving is toward outcome-based pricing, where you’re paying for resolutions. This is what Fin does so well, for example, versus just paying for a seat. That’s definitely a trend we’re getting really behind, and more and more of our products are going to be delivered that way.
Yeah. Fin was certainly an early pioneer in outcome-based, more resolution-based pricing.
The reason they can do that is because the time to value is so fast. You can get Fin up and running on average in 16 days and immediately start seeing the agents handling more and more of the calls that humans might otherwise have been escalated to handle.
We decided, “Okay, well, why don’t we just pilot Fin internally and see what it does for Salesforce?” We found that we had it up and running in 14 days. On our help.salesforce.com website, it’s now handling over 40% of the traffic.
Oh, wow.
It’s delivering resolutions in 65% of cases without going to a human. That’s really quick time to value, and it’s a hallmark of what a great acquisition looks like.
Yeah. With enterprise sales in particular, you’re earning a larger share of the labor budget. In your Fin example, that could be customer support, a little bit of sales. In a Salesforce context at large, that can be: “My AI team is prospecting, finding leads, qualifying leads, following up, and closing deals.”
Right.
As you’re earning a greater share of that labor budget, your customers may be paying you quite a bit more than they did in the past, but their overall cost of selling has gone down.
Absolutely. Qualified is probably one of the great examples of what you just said. It can replace your sales development rep organization, essentially. It’s taking all the traffic that’s coming in through the website and answering questions with its Piper agent. It’s a very human-like experience where you’re speaking to a lifelike avatar. They can explain what the products do and help set up meetings for you automatically, without touching a human.
It makes a ton of sense. We just started to touch on a number of your acquisitions. You have a long history of acquisitions—I think nearly $100 billion over the course of your tenure, 12 over the last year. Really amazing. As you think about M&A as your strategy, there were the large ones like Slack, MuleSoft, and Informatica, and then more recently, you’ve gotten to the AI-native companies. You mentioned Fin; Listen Labs is another. Your framework for acquisitions that you touched on: financials is the third component.
Mm-hmm.
The second is strategic, and the first was, “What are your customers asking for?”
And also, how can we accelerate them—
Accelerate growth.
—with our distribution. Remember, we have 15,000 sales reps. If you can plug one of these acquisitions into that—in the case of Listen, with only a handful of reps—you suddenly make it available to your entire sales team. You’re going to grow that business.
So maybe imagine I’m on the other side of the table. I’m Alfred, I’m at Listen Labs. I’m a really ambitious AI founder. I get approached, I’m spending time with Salesforce, and it sounds great to partner with your 15,000 sales reps and get pushed through your channel. I would love that. How do I think about partnering with Salesforce versus selling to Salesforce?
Well, I think we approach partnerships in a very open way. We have probably more than 8,000 partners that we work with at this time, and these are ISV partners. Sometimes we invest in the partners as well, by the way, and that’s part of the role of Salesforce Ventures.
We want to have an open ecosystem where we’re bringing as many solutions as possible to our customers, and they make the decision. They can buy a partner solution or they can buy a solution from Salesforce. I think the hallmark of our company is making sure we have as broad an ecosystem as possible, because you want to have as much to offer as you possibly can. There are definitely ways to partner with us. We have multiple types of partner programs and a great organization that does that.
So maybe, as I think about it as an ambitious AI founder selling to Salesforce, why is that the more ambitious choice for me?
Imagine having to build that distribution channel on your own, and how long it will take to hire the reps, ramp the reps, enable them, and get them productive. It takes a very, very long time to build that at scale. And, by the way, you’ve got to do that across small businesses, medium-sized businesses, and enterprises. You also have to do that globally. It’s a heavy, heavy task for anyone to take on.
Entrepreneurs have to make that decision: Is this the right time to sell and plug my technology into a company that already has that, or can I continue and go it alone, try to build it on my own, continue to raise money, and dilute myself further and further and further? It’s many years of undertaking versus getting your technology in the hands of customers as widely as possible and as quickly as possible.
And—
And that’s the decision you have to make.
Fair enough. As you try to integrate these folks into your culture, they bring a certain entrepreneurial spirit and their own way of thinking—
Which we love.
—of thinking.
Yes.
So how do you think about bringing them in and integrating them appropriately so that that spirit can continue to spread and thrive while being part of the larger platform?
I think that’s one of the most exciting things about what we do when we buy a company: bringing in this exceptional talent to help transform the company and, in many ways, refresh us. It makes sure we’re bringing that entrepreneurial spirit into the company and thinking about new ways of doing things.
A key focal point of the integration team is the people: making sure they’re successful in onboarding and learning how things work within Salesforce, but also growing their careers internally. We’ve had many stories of people who have come in, sold their companies to us, and risen to become our chief product officer. It’s happened four times now. Others have taken on big roles within our sales organization or other leadership roles. I think it’s been an important part of how we’ve grown the company, too, through bringing in new talent.
You have 80,000 employees. I can imagine you bring in these companies and they help turbocharge certain areas.
Mm-hmm.
Is there an example of one of these acquired companies that you point to as having a big impact on the culture of the organization?
One that springs to mind is David Schmaier’s company, Velocity. We didn’t really have much of an industry focus within Salesforce. Yet there are so many important verticals that are a key part of the growth of the entire software industry, and we now have 13 verticals.
Oh, wow.
That focus on the industries and the specific use cases for each industry came through that acquisition. David came on board and brought Velocity in. It brought in six verticals; we’ve added another seven since. He went on to be chief product officer and is now chief strategy officer. I think he’s definitely helped us think about selling to those industry verticals in ways we never would have otherwise.
That’s very interesting. One of the pieces that’s also been impressive about Salesforce over the years is that there are large companies built on top of you that are just focused on verticals as well, like Viva and the like.
nCino was probably—
nCino is another great example.
—a great example of that, and that was one of our early investments in the Ventures program. They created an incredible company just based in financial services.
Today, as you’re thinking about build versus buy, you’re probably told a lot more, “We can just build that.” Software development is faster and more affordable than it’s ever been. Part of your job must be figuring out when that’s actually true. In this era of software, what is worth actually buying?
When we think about whether to build, buy, or partner, we go through a lot of analysis with the product teams around what their roadmap should look like for the next three years. Do they truly have the people who can build what they’re trying to deliver? Do they have the internal expertise, for example? And can they go out and easily hire for it?
It’s a competitive market these days.
You know how hard hiring AI engineers is right now: very, very hard. In many cases, you can really speed up the development process through an acquisition if you can find the right one and integrate it quickly. That’s also one of the things you have to consider, because a typical acquisition, end to end, from the moment you start with the idea to when it’s actually integrated, could be six to nine months. Could you deliver something on your own much faster than that? Those are the types of calculations we make and try to think about.
When the right asset shows up, you really know it. You can have a sense. That was the case when Alfred and Florian came to meet with Marc and demoed Listen Labs to us. They were just brilliant, and we thought not only was the technology outstanding, but the interface was so easy. It was so easy that my team started using it with no enablement whatsoever. Within 10 minutes, we had a survey out and started. It just could not have been easier.
Is that how you diligenced Listen Labs? A survey on Listen Labs?
It was part of what we did—to their customers, actually. That was incredibly effective. By the way, our diligence process is pretty rigorous, and the customer element, the customer feedback that we get, is one of the most important ways of judging how good the product is and whether these customers are likely to renew.
We’re happy Listen Labs customers ourselves and use it as a diligence tool as well. Very cool. Maybe on this piece of accelerating capabilities, another acquisition that you were talking about earlier was Fin. You guys have long had a role in customer support. Maybe talk about what capabilities Fin brought to you all, and the story behind that acquisition.
Mm-hmm.
Yeah, I think it goes back to what we talked about earlier: Mark’s focus on having a beginner’s mind. We’ve built an incredible Service Cloud business. It’s now almost $11 billion in AOV. The foundation of the business was an acquisition we made in 2008: a company called Insternet, which was Alex Dayon’s company. He went on to become our chief product officer.
A sequel. Amazing.
He’s exceptional. He’s one of the greatest entrepreneurs to come out of France. What was very clear to us is that the whole movement toward native AI applications, the way they were priced based on outcomes, and the very fast time to value—how quickly you could get them implemented and up and running, with happy customers using them—wasn’t something we could build quickly enough.
We had a wonderful relationship with Eoghan, whom we’d known for years. We got to know his AI team in Ireland, who are outstanding—some of the best AI talent in Dublin. We realized this could turbocharge what we were trying to do with Service Cloud. We were able to buy the company, and it’s proving to be a very valuable acquisition for us, as I said earlier.
Yeah, that one’s an amazing story from a company-rebirth perspective, too, with the transition from Intercom to Fin and him coming back as founder.
Yeah. He’s having a big impact at the company. Eoghan is someone with a loud voice, and he makes his views known.
Free spirit.
He’s really trying to shake us up, and that’s another benefit of bringing these people in.
Amazing. Another product that I use every day, and certainly you do, is Slack. I’d be curious: if you rewind the clock, you had a rationale for acquiring Slack. Slack’s role in AI today is pretty important and different from maybe what you could have expected then. I’m curious how your thinking about Slack has evolved over the years.
I think the core thesis for why we bought Slack holds true today, but it’s even bigger than I think we all imagined, because it’s the perfect conversational interface for AI. We bought the company in 2020, so it’s now been—
There’s a rumor out there that Mark only spent an hour with Stewart before the acquisition. I don’t know if that’s typical, but—
A lot more time than that, definitely. We looked at it very closely. We’d been using Slack. It was clearly a beloved product among its users, and it needed more distribution. It was losing to Teams, and they were stuck. I don’t know if you remember 2020, but there was one bad quarter, its stock went down, and that’s really what triggered a lot of the discussions: How can we bring these 2 companies together and turn Slack into what it really needed to be, which was the user interface for everyone and the collaborative tool where humans and apps would work together?
And now agents.
And now agents, yes. We didn’t know about the agent part; that came later. But you can see it was the perfect interface for that. We also knew it would be the interface for all of our products. You can access all of Salesforce’s products through Slack, and that was always one of our key strategic rationales from the beginning.
Since then, Slack revenue is now almost 3 times what it was when we bought the company, and it’s our fastest-growing product. It’s taking off like wildfire because it’s perfect for agents.
Yeah, that makes a ton of sense. I think of Slack in this current era as the unofficial org chart. It knows who reports to whom and how people talk to each other. It has all the context on how decisions are made.
Yeah.
So it’s a really important corpus for AI.
Yeah.
But it’s an interface product, as you alluded to. As other products become more headless, it can be the central repository, the central ground for decision-making.
Absolutely. Because it has all the context from all of your work, if you use Slackbot, you can get the richest answers you’ve ever seen from AI right there within your interface. It couldn’t be easier, and we’re thrilled we were able to buy the company. It’s been 5 years and a quarter since we’ve owned it, and I think it’s transformed Salesforce.
Amazing. You said it’s grown nearly 3 times in a bit over 5 years. That’s really strong growth at scale. What about that story and the integration has gone so well, relative to other acquisitions?
I think it wasn’t clear from the beginning that it was going to go so well, especially since we were buying it during the pandemic. We basically negotiated the deal and did all the diligence through Slack, buying the company without ever having a lot of face-to-face meetings. It took a little bit of time for the integration to hit its stride, but once it did, it took off.
Again, it’s great technology that was starved for distribution. That’s the benefit an entrepreneur can get from a great acquisition partner.
How many of those face-to-face meetings happened to happen in Hawaii?
Well, that was a lot later, actually.
That wasn’t the common ground in COVID times?
No, not then. There were so many restrictions on even traveling to Hawaii then. But Hawaii is very useful now. Think about Braze, for example, which is our most recently announced acquisition. I think the meeting the 2 CEOs had in Hawaii was critical in jump-starting that process, building camaraderie, and seeing the alignment. Hawaii is a magical place.
All good things start in Hawaii. We’ve spent a bunch of time talking about buying companies and partnering with companies. You have a third option, which is investing in companies and letting them run independently. I used to think of Salesforce Ventures as partnering with your ecosystem, your close partners. Today, it seems you’re investing much more freely and broadly. Anthropic and Mistral are close partners of the company. Fall, Basepower seems a little more out there in terms of strategic fit. I’m curious how the strategy has evolved over the years.
That’s a great question, because there’s been an evolution for sure. But the through line has always been the same: How can we make sure we’re finding the companies that will deliver enterprise value, whether today or in the future? It started in 2009, when we invested in companies that were part of our ecosystem, like DocuSign and nCino—
They should closely fit with your software.
Twilio was another early investment. We integrated with them and partnered closely with them. We also made a lot of services investments to build out our whole ecosystem of regional services partners, and those were very successful.
We then ventured more into new areas where the industry was heading, like Snowflake. That was a very critical investment we made over several rounds, including the IPO round. We also invested in cybersecurity because we could see it was going to become a massive problem for the industry, and we needed to stay current. Wiz was one of our investments.
When AI became the obvious next wave of the industry, we wanted to invest very heavily in the best companies there. We created the billion-dollar AI fund at the beginning of 2023. In addition to Anthropic, we invested in Mistral and Cohere, and in several companies more at the infrastructure level. We’ve invested in voice with ElevenLabs and in video with Runway.
Now we’re venturing into physical AI, too, which I think is a new, fun, and exciting area. With improvements in robotics, dexterity, and mobility, you can see this is going to be a huge part of the industry’s future, and we want to be there.
Totally makes sense. Very broad mandate. It seems to cover all of technology wherever there is growth in the sector at large. How are you spending your time personally?
I divide my time between finding new companies to acquire, working with the integration team and making sure the acquisitions are successful, and the Ventures program. I’m split a lot. As I said to you, I’ve never been busier than I am now, but I think that’s a hallmark of how rapidly the industry is evolving and how quickly technology innovation is happening these days. You just have to move faster, and however fast you move, it’s not even fast enough.
In all these areas on the venture side, what are you most excited about right now?
I think the robotics companies are fascinating to see. I absolutely love that. We have an investment in Figure AI and its humanoids. We’ve got investments in Physical Intelligence, Dyna Robotics, Skilled, and a variety of other companies. We have Zipline, which does drone delivery, for example. So we’re venturing off into new areas.
We love many of those companies.
And we’re doing things in new areas that aren’t just about software, but it’s very clear they’re going to be really important to the industry. I’m enjoying that. I’d also say we have investments around the world, and we’re seeing some really interesting companies in Japan, for example. We invested in a robotics company there called Mugen, which is really big in logistics and in the Japanese market. We have investments in many companies in Europe, Israel, other parts of Asia, and Australia, for example. Seeing the globalization of technology has been a really interesting vantage point.
On the robotics side, I agree. We’re starting to see the early signs of scaling laws within robotics, which is really exciting. A bunch of these companies have gone a bit more vertical in their go-to-market, and we’re starting to see revenue scale up, which is really exciting. As you think about your investment aperture, how often are you looking at companies and thinking, “Hey, potentially one day they could compete with Salesforce”? Would you invest in a company that might take that position?
I think we probably should. It’s good to stay close to where the technology is going, and maybe that company may turn into an acquisition target someday. To give you some numbers, we’ve invested in more than 750 companies. We’ve acquired 29, so it’s still a relatively small number of companies. But if we ever see something really exciting, maybe it’s an acquisition target in the future, maybe not; maybe it’s just a great partner. We want to stay close to that company.
You mentioned that a good portion of those 750 are in areas you wouldn’t say are core strategic areas. Robotics today doesn’t strike me as the next thing in line in the Salesforce roadmap.
But we want to be partnering with them.
But you want to be partnering with them.
Yeah.
How do they think about the value proposition of partnering with Salesforce Ventures?
I think the biggest value add is probably how we can help them access customers. We have the broadest set of enterprise customers that are interested in buying technology, and we help our portfolio companies get in front of these customers. I think we also provide a lot of meaningful advice from an operational lens around what it’s like to build a business in this particular area. Having been through so many stories, we can really help these entrepreneurs grow their businesses.
If you were going to start a company today and it was going to attack Salesforce head-on, what openings do you see?
What I would say is, I don’t know why there continue to be new CRM companies at this point. It’s interesting how—
Why do you feel that way? There are a bunch of AI-native CRMs. There’s Lightfield, Monaco. I’m curious about your take.
I just think it’s very hard to—First of all, there’s so much great CRM technology in general, and we are continually improving our products. It’s just very hard to build something that’s particularly specialized and niche in that area right now. I have yet to see something so novel that it’s going to turn into a large, multibillion-dollar company in that space.
My perspective would be that maybe there’s an opportunity downmarket from where you traditionally play. I know you guys play in SMB too, but in terms of getting up and running quickly, bottoms-up, there could be an opportunity.
Maybe more very downmarket, yes.
There could be an opportunity.
I could see that. But it’s still a small area, and venture investing—we only invest in big opportunities, and I think most VCs look at it that way. I just don’t get investing in small, niche opportunities that might be a little different from what the incumbents are doing.
Fair enough. The parallel story might be that Stripe started with small companies and was able to grow with a lot of them over time. I imagine these startups might have similar aspirations.
I just think there are too many of them. There are too many of them. There’s too much funding going into them, and I think the VC community should focus on really new, important areas that haven’t been done before.
As you think about the VC community at large, we’re in a unique funding moment in time. I’m curious what you make of it and how you’re making sense of valuations these days.
I think there’s a lot of irresponsibility going on. The valuations are, of course, way too high. But also, the buyer universe is so small. When you think about investing, you want to think about, “What are my exit opportunities?” Not every company is going to go public. In fact, very few companies go public—probably 1% or less. So the exit opportunity is critical. You want to think about who could be buying it. Once you get above $10 billion, for example—
There aren’t that many buyers.
The buyer universe is so small, and yet you’re seeing so many enormous, multibillion-dollar-valued companies out there. There will be a reckoning that happens. I don’t know when, but it’s inevitable because a lot of these companies have valuations that are not sustainable, that they won’t be able to grow into, and that I think are also going to be bad for their employees. This is why I say it’s a little irresponsible. Unfortunately, you try to have this dialogue with entrepreneurs, and they just want the highest valuation they can possibly get. But that’s usually not the best decision.
Do you think so much of it is a result of power law thinking? Anthropic is likely to go out at a valuation north of $1 trillion, let’s just say. Obviously, SpaceX at $2 trillion. Those companies grew to those valuations in a relatively short period of time, certainly in Anthropic and OpenAI’s case.
Right. Those are generational companies.
Absolutely.
They’re very rare, and when you find one of those, they’re amazing. That’s why we invested in six rounds of Anthropic. It was clear from the first meeting we had with Dario and Daniela that these were special people. They were some of the most brilliant minds we’d ever met, and they were going to build something great. From that vantage point, it made sense to keep investing. But those companies are rare, and I see a lot of mediocrity in entrepreneurs and their ideas that are not going to become generational companies. They shouldn’t be commanding valuations above $10 billion, but they get them. That’s where I feel a little more discipline is warranted right now.
I agree that in certain pockets of the ecosystem, that exists. Part of my optimism comes from how different a top 1% outcome was 10 years ago from what a top 1% outcome is today.
Yeah.
Even on the M&A side, you mentioned Wiz earlier, and Cursor. There have been these larger outcomes that drive returns for the ecosystem at large.
Definitely. I’m an optimist. I could not do my job if I weren’t an optimist. But I see the thousands of companies every year that don’t make it—
Because they all come to us to try to be acquired. We see that part of the industry that you don’t read about in the press.
Yeah.
The press doesn’t like to cover those stories. It’s an important fact to consider when an entrepreneur is thinking about raising money: What’s the right valuation? What’s the appropriate way to build this business? They have to think about all of that.
You’ve been involved in some of the largest enterprise software transactions. Sometimes those deals go south and don’t happen. What kills deals?
When you go through the diligence process, first of all, it’s really important that you’re ready and organized, so the data is flowing and it doesn’t look like you’re having trouble getting access to it. One thing we’ve found is that a lot of companies don’t have strong controls over their financials, or they don’t have data on their customer contracts. They don’t really understand what customer attrition is.
One of the mistakes a lot of founders are making these days is focusing on one metric alone, which is ARR. ARR is a wonderful metric, but it masks attrition. It’s really important that you understand the attrition of your customers and do something about it. When you meet with a company that has very high attrition, that usually means there are a lot of unhappy customers, and there might be some product quality issues as well. It’s one of the big red flags we look for.
Another area is how you’re handling your data privacy and security. We have 14 functions that we work with on all of our diligence processes, and security is the one function that can hold up the red flag and say, “I’m killing this deal.”
Hmm.
If you’ve built your technology stack so it’s easily open to hacking, that’s going to be a real problem for us. You also want to be prepared by red-teaming your technology stack and seeing what happens before you go into diligence, because we’re going to do our own penetration testing and find the problems. You definitely want to identify the vulnerabilities before we do.
Nothing goes better in a process than a company being totally buttoned up, with all its data accessible and its tech secure. Then the diligence process moves really rapidly, and we can get through it quickly and painlessly. When they don’t, that’s when deals can die.
Is there an example of a deal not going through at the 11th hour? You can’t name names, of course, but maybe a painful anecdote for you or for the founder?
I think it usually doesn’t happen at the 11th hour. It’ll happen after the term sheet is agreed and you’ve negotiated the price and the major terms of a deal. Then you get into the deep diligence process.
We found problems—for example, when we did the pen test, we were able to access not just all their source code, but even the CEO’s laptop and information about their employees.
When you shared that back, it was pretty understandable.
And we shared it back. Of course, we always share what we find and try to help them fix the problems. But it was so bad, some of the worst we’d ever seen, that it was unrecoverable.
We’ve had other security incidents that are recoverable. You can identify the P0, P1, and P2 problems, and make fixing some of them a condition to close—or maybe a condition of signing. We won’t sign until these are fixed; we won’t close until the others are fixed. It all depends on how bad it is, but you’ve got to have your house in order when you go through one of these processes.
Makes sense. There’s an amazing photo of you and your mother online, representing your father in 2023 for his Farami Award. Your father was an extraordinary figure. He survived the Holocaust, fled Europe, came here, and had an incredible scientific career at Berkeley. I’d love to close with your experience growing up with your father, what he thought of your career, and how he shaped you.
Well, my father wanted me to be a chemist, to be clear.
You came pretty close.
You could argue M&A is a lot of chemistry in what we do. He founded a field of chemistry called surface chemistry, and also spent a lot of time in catalysis. He won many awards. We got to go to the White House when he won the National Medal of Science.
It was incredible to grow up with a man like that and meet his colleagues, too. We had some of the most incredible minds in the world come to our house for dinner. My mother loved to host these incredible dinner parties, and that was what life was like growing up. You were really encouraged to be curious and to think about the future. That was a big part of the way I grew up, and I think it’s really shaped who I am today.
That does seem to have a lot of parallels to what you do. Maybe those dinner parties and hanging out with the chemists had some similarity to hanging out with the AI researchers of today.
Definitely. I also had a father who loved to teach all the time. He would teach at the dinner table, in the car rides, and when he was driving me to school. You were just flooded with information and knowledge all the time. He was a remarkable man.
Amazing. Well, thank you so much for spending the time with us today. This was really fun.
Yeah. Thank you, Alex.