GP @Benchmark Director @confluentinc @Cerebras @Contentful @Benchling @CommerceLayer @acuitymd @FireworksAI_HQ @quilterai @pomerium_io @greptile @sundayrobotics
- Tweets 2,990
- Following 853
- Followers 21,235
- Likes 9,176
Pinned Tweet
.@benchmark co-lead the initial round for Cerebras 10 years ago. Over the following 5 years, the team amazed, delivering the technological marvel of a wafer-scale chip, the system to heat and cool it, and more recently the software layer that allows giant fleets of Cerebras systems to work together for very large MoE models.
But even more impressive, is they just never fucking quit, despite kissing death like 3 times, getting made fun of for unusual early customers, and getting passed over by virtually every respected semi investor (who have all converted now).
The team knew, IF they could stay alive, it was just a matter of time…. In tech, speed ultimately wins, and nothing is close to as fast as Cerebras.
@OpenAI and @Cerebras have signed a multi-year agreement to deploy 750 megawatts of Cerebras wafer-scale systems to serve OpenAI customers.
This has been a decade in the making.
Deployment begins in early 2026, and when fully rolled out, it will be the largest high-speed AI inference deployment in the world.
OpenAI and Cerebras were both founded in 2015 with radically ambitious goals.
OpenAI set out to build the software that would push AI toward general intelligence.
Cerebras set out to rethink computing hardware from first principles.
Our teams met as far back as 2017. We shared ideas, early work, and a common belief:
there would come a point when model scale and hardware architecture would have to converge.
That point has arrived.
ChatGPT set the direction for the entire industry. It showed the world what AI could be.
Now we’re in the next phase - not proving capability, but delivering it at global scale.
The history of technology is clear on one thing:
speed drives adoption.
The PC industry didn’t operate at kilohertz.
The internet didn’t change the world on dial-up.
AI is no different.
As models grow more capable, speed becomes the bottleneck.
Slow systems limit what users can do, how often they engage, and whether AI becomes infrastructure or remains a novelty.
Cerebras was built for this moment.
By keeping computation and memory on a single wafer-scale processor, we eliminate the data-movement penalties that dominate GPU systems. The result is up to 15× faster inference, without sacrificing model size or accuracy.
That speed changes product design, user behavior, and ultimately productivity.
For consumers, it means AI that feels instantaneous.
For the economy, it means agents that can finally drive serious productivity growth.
For Cerebras, 2026 will be a defining year.
With this collaboration with OpenAI, Cerebras’ wafer-scale technology will reach hundreds of millions - and eventually billions - of users.
We’re proud to work alongside OpenAI to bring fast, frontier AI to people around the world.
This is what a decade of long-term thinking looks like.
17
117
375
177,718
Has there been a bank run in the US over the last 50 years where depositors actually lost money?
Fed can't let it happen. Would trigger more runs.
Bank equity to 0? Sure.
Funds tied up for a few days? Maybe.
Lost money? Can't find an example.
102
94
1,071
928,315
1/ Company proposes ambitious, seemingly unrealistic, plan. Board member makes “if you do that, I'll eat my hat” type comment which leads to a silly bet. Company obliterates plan. CEO shaves board member's head. I'm the board member. @Confluentinc the company.
37
98
781
If SVB depositors aren’t made whole, then corporate boards will have to insist their companies use two or more of the BIG four banks exclusively.
Which will crush smaller banks.
AND make the too big to fail problem way worse.
(SVB equity holders are wiped. That’s fine.)
88
52
573
322,477
This gave me chills. Awesome lens.
From @InvestLikeBest @patrick_oshag via @JamesClear
Investor Rick Buhrman on the kindness of mastering your craft:
INTERVIEWER: What is the kindest thing that anyone's ever done for you?
BUHRMAN: ... our oldest son, Theo, who just turned seven, spent the first six months of his life in several NICUs. He was eventually helicoptered to Indianapolis at Riley Hospital for Children. And while we were living in that NICU for almost a half a year we saw a lot of kids who passed away. Most of those kids were not as sick as Theo was.
I don’t know exactly why Theo survived, but I know that a major part of how he survived was because for several decades leading up to that moment, numerous nurses, nurse practitioners, respiratory therapists, doctors, surgeons had committed themselves wholeheartedly to mastering their craft. I can give you tons and tons of examples of these people. And I know that in the moment, it wasn’t necessarily viewed as kindness.
But maybe in some sense, the kindest thing that all of us can do is to pursue something radically that in some way is in service to others, because you just don't know how it's going to change the trajectory of human life. And so for all of those medical practitioners, none of whom I'm sure are listening to this, I owe everything to, because they gave me the gift of being Theo's dad.
16
48
462
62,596
Sequoia and Benchmark are often competitors, but more often we’re partners. I’ve had the privilege of serving on boards with four Sequoia partners (including Pat) across four companies. And there are many more examples across our broader portfolio. They’re the rare venture capitalists who actually do the work and, importantly, add real value.
Over the years I’ve learned a tremendous amount from Pat. Whenever I’m stuck, he’s my first call outside our own partnership — and he personifies what you’d want in the leader of a firm.
Few leaders can put the organization’s success ahead of their own — balancing the confidence to make controversial calls with the humility to listen. Those traits are rare anywhere, and especially in our business, where success is ultimately measured by money made.
Congratulations to everyone at Sequoia on this prescient evolution.
Oh — when your competitors get stronger, you have to get stronger too. We welcome that challenge ;-)
See you on the field. ⚔️
With immense gratitude and appreciation for @roelofbotha and his legendary leadership over the better part of a decade, @Alfred_Lin and I are honored to accept the torch and become the next stewards of @sequoia
Here is the note I shared internally:
14
13
450
213,017
THIS IS THE BEST FEATURE EXCEL HAS ADDED IN AT LEAST 15 YEARS.
Announcing STOCKHISTORY, the newest member of the @msexcel function family. Easily pull historical stock prices into your workbook to analyze your portfolio! Get started: msft.it/6018Tllcn
14
25
442
Remember when 3x, 3x, 2x, 2x, 2x was the gold standard for annual growth in SaaS companies?
That was cute.
16
10
366
100,073
Interestingly, I was surprised to learn I counted as diverse from a Nasdaq perspective (based on the rules at the time), but did NOT count as diverse according to California’s rules (at the time).
9
7
272
99,826
In terms of things VCs directly control - employees & investments - @A16Z has done better on inclusion than any other VC firm. @bhorowitz @NaithanJones @ChrisLyons @JorgeCondeBio @martin_casado I’m sure others. Let’s not measure action by tweets. nitter.app/kateclarktweets/status…
10
14
237
Vishria at Benchmark sums up the hope: “All the pretenders and the speculators will get wiped out. We’ll have the believers and the builders.”
ft.com/content/6395df7e-1bab…
9
31
234
I agree with all of this (especially the “good enough AI” concept which @zeynep articulated well) but there is a new dynamic that shouldn’t be ignored imo:
For the past couple years the model was the product. Now, there is distinctly a product layer and a model layer. Deep research, voice mode, the system prompts, the UI controls, the artifacts, integrations, APIs, reliability…. all tremendously impact the user/developer experience but are beyond the model itself. Reminds me of Geoffrey Moore’s “whole product” concept.
OpenAI’s lead on the model layer may have narrowed, but they are pushing the product layer better than anyone, right now anyway. There’s a good argument that once a threshold has been exceeded, the best product, not best model, wins.
A test-the-theory thought experiment I’ve been asking folks, including OpenAI folks, could you imagine a world 3-5 years from now where OpenAIs underlying models aren’t all home grown?
It would be a wild identity shift internally to be sure, but like the shifts in chips in Macs or Azure running Linux and various open source databases, we’ve seen crazier things!
14
8
235
39,311
Most of our tech industry doesn’t get it either. Intel’s stumble means we depend 100% on chips from a tiny island not far off the coast of China. Yes, Nvidia’a chips too. Increasingly AMDs. Everybody’s.
13
24
219
Today, I find myself simultaneously agreeing with @MittRomney @RoKhanna @LHSummers @DavidSacks and… @profgalloway (?!) which is all extremely unusual and very disorienting :-)
Meanwhile ppl on the far right and far left bizarrely want to test containment, which seems… risky.
Kudos to @RoKhanna. Every word of this is true. Decisive action is needed to protect the regional banking system. The run on other banks has already started.
17
8
213
70,595
While our regulators chase Amazon and meaningless tiny acquisitions by Facebook, the greatest monopolist today - Apple - has taken a Trump-ian tact, pushing a privacy narrative while stifling competition, and favoring their own ads business.
As only @benthompson can...
8-15-2022
•Apple and Facebook’s Pre-ATT Negotiations
•Timing and Documentation
•ATT’s Impact
stratechery.com/2022/apple-a…
11
22
199
By and large, SaaS leaders dropping into rapidly scaling AI-native co's are struggling. It is like dropping rural Germans into the middle of modern Tokyo.
The AI CEOs only know this pace, tech, sales cycle.. and can’t explain how these leaders should reframe. The SaaS people talk about quota capacity, qtrly roadmaps, & enterprise, which don’t mean anything to the AI CEOs and are generally irrelevant, at least for now.
Bridging this is hard but critical. The productive combo of this wave & scaling expertise would turbocharge these already ripping companies.
13
17
193
55,550
We all know that organic and SEO traffic to our sites is dropping rapidly. And a huge part of user research on which products to use, justification for mgmt, etc. is done inside ChatGPT or the like.
One tidbit from one of our companies - if people ask the LLM how different products compare, the main data source for those questions often is the websites of the companies and how they compare themselves against competition.
So you need to start creating content on your B2B websites for those types of queries (even if it's not visible to the user).
You need a whole set of hidden pages that get crawled, that compare us vs. each competitor, in detail, in ways that will then get surfaced to the user in an LLM (and crazily enough will be even more authoritative and seem less biased).
The LLM crawlers and indexes aren’t nearly at good at fighting the game play - Google has had 25 years to do this. And if you don’t do it, I guarantee your 25-something year old competitors are
We just made Amplitude's first free product.
Introducing AI Visibility: SEO for LLMs.
Customers are using search engines less and AI more for finding what companies they should work with.
AI Visibility queries @ChatGPTapp, @claudeai, and @GeminiApp, and scores how your company appears across hundreds of prompts. It shows competitor rankings, sentiment, and trends. It gives you recommendations on how you can improve your company's visibility. Ranking highly with LLMs is the biggest untapped opportunity in marketing.
@_leojiang shares more:
25
14
185
80,992
This is the definition of VC that we aspire to.
And I have a different definition of “venture capitalist” than the world has recently adopted. To me, it’s someone whose personal involvement and efforts can materially impact your chance of success. The work not the $$. Others are investors. There are few true VCs in the world.
15
14
183
As SaaS multiples fully embrace the “can’t stop, won’t stop” stage, it was fun to look back 20+ years at the evolution of SaaS from Applications Service Providers (ASPs) of the late 90s to API companies of today. Thank you @patrick_oshag for having me on!
My conversation with @ericvishria on all things software.
- why public software valuations are so high
- the "competitive frontier"
- the three generations of SaaS
- APIs as products
- zones of opportunity
Eric is among the best out there. Enjoy!
investorfieldguide.com/eric-…
9
18
181
Thoughtful take.
As I’ve said for the last couple years, FMs are the fastest depreciating asset in human history.
1) DeepSeek r1 is real with important nuances. Most important is the fact that r1 is so much cheaper and more efficient to inference than o1, not from the $6m training figure. r1 costs 93% less to *use* than o1 per each API, can be run locally on a high end work station and does not seem to have hit any rate limits which is wild. Simple math is that every 1b active parameters requires 1 gb of RAM in FP8, so r1 requires 37 gb of RAM. Batching massively lowers costs and more compute increases tokens/second so still advantages to inference in the cloud. Would also note that there are true geopolitical dynamics at play here and I don’t think it is a coincidence that this came out right after “Stargate.” RIP, $500 billion - we hardly even knew you.
Real: 1) It is/was the #1 download in the relevant App Store category. Obviously ahead of ChatGPT; something neither Gemini nor Claude was able to accomplish. 2) It is comparable to o1 from a quality perspective although lags o3. 3) There were real algorithmic breakthroughs that led to it being dramatically more efficient both to train and inference. Training in FP8, MLA and multi-token prediction are significant. 4) It is easy to verify that the r1 training run only cost $6m. While this is literally true, it is also *deeply* misleading. 5) Even their hardware architecture is novel and I will note that they use PCI-Express for scale up.
Nuance: 1) The $6m does not include “costs associated with prior research and ablation experiments on architectures, algorithms and data” per the technical paper. “Other than that Mrs. Lincoln, how was the play?” This means that it is possible to train an r1 quality model with a $6m run *if* a lab has already spent hundreds of millions of dollars on prior research and has access to much larger clusters. Deepseek obviously has way more than 2048 H800s; one of their earlier papers referenced a cluster of 10k A100s. An equivalently smart team can’t just spin up a 2000 GPU cluster and train r1 from scratch with $6m. Roughly 20% of Nvidia’s revenue goes through Singapore. 20% of Nvidia’s GPUs are probably not in Singapore despite their best efforts. 2) There was a lot of distillation - i.e. it is unlikely they could have trained this without unhindered access to GPT-4o and o1. As @altcap pointed out to me yesterday, kinda funny to restrict access to leading edge GPUs and not do anything about China’s ability to distill leading edge American models - obviously defeats the purpose of the export restrictions. Why buy the cow when you can get the milk for free?
12
10
175
99,818
As we work through our 2nd 48+ hour power outage of ‘23, I’m reminded of an under discussed reason the gas stove ban is so dumb: having our lighting, heating AND cooking dependent on an already unreliable electric grid.
Increasing single points of failure is not a good idea.
27
10
171
47,891
In thinking through future implications of AI, it is more important than ever to be able to hold multiple opposing ideas in your head at once.
Because the probability distribution for any outcome has widened and flattened, dramatically in many cases.
11
13
172
23,593
100%. Another example, listen to @mntruell talk about the history of dev tools.
1
6
152
27,886
In 1999, at the time of Google’s series A, the Internet ran on Sun Solaris and Sparc and Oracle.
In 2004, at the time of Google’s IPO, the Internet ran on Linux, x86 and mySQL.
Unimaginably dramatic shift. In 5 years.
And the current cycle is mostly moving faster.
9
18
146
19,782
Inspired by @andrewdfeldman sharing his Thanksgiving story, here’s mine.
As a child of Indian immigrants, our household was vegetarian. I didn’t understand the big deal about Thanksgiving, or turkey but was curious, as every elementary school activity in Nov is centered around it.
In second grade, my dad, under extreme disapproval from my mom, took my kindergarten brother and I to the grocery store on Thanksgiving morning. He ordered a small plastic bag’s worth of shaved turkey from the deli counter.
We excitedly rushed home, and the three of us stood around in the freezing garage and ate that delicious deli turkey straight out of the bag with our hands.
And all I remember thinking is, “omg, I’ve been missing out, this is so awesome.”
Happy Thanksgiving!🍁
12
1
163
31,920
Venture capital is like looking at a kindergarten playground and trying to pick out LeBron. 😂
This one was fun - thanks @thogge @sterlingmsnow for having me.
7
9
145
16,250
Every once in a while, entrepreneurs propose an idea so provocative it completely changes how you think about an entire domain. 3.5 years ago the founders @cerebras did that to us @benchmark.
2
21
178
For decades Bezos masterfully played "your margin is my opportunity," and now the table is 180 degrees turned - Amazon's margin is GCP/Azure's opportunity. They are happy to buy cloud business, funded by the best business model of all time. Under discussed irony, thx @chetanp
"Your margin is my opportunity": really interesting symmetry in Microsoft and Google leveraging their high margin core businesses to compete against the very high margin Amazon AWS business.
4
5
133
This kid is going to be a legend. Learners mentality.
Carlos Alcaraz says he wasn’t mature enough to handle the match tonight against Daniil Medvedev:
“These kind of matches can happen. Even if I feel I’m a more mature, different player. He played really really great. A great game. I couldn’t find solutions. I thought right now I’m a better player to find solutions when the match isn’t going the right direction.. after this match, I’m gonna change my mind. I’m not mature enough to handle these kind of matches.”
8
3
133
33,509
While the co-pilot approach is all the rage, I think we’re going to see a bunch of areas where AI just does - like completely does - tedious work humans historically have done. No AI-assist, just AI-and-done.
@quilterai is taking this approach for PCB layout, which if you’ve ever seen, is about the most tedious work you could imagine. We’re excited to join @sergiynest on the journey.
Special thanks to @mcgd for introducing us!
Quilter is building generative AI for electronics that fully automates the process of designing circuit boards.
Our objective is to build a “compiler” for electronics design that makes it fast and easy to generate high quality, physics-validated, fully optimized PCBs with AI.
ALT https://app.quilter.ai in action!
11
7
130
53,034
Benchmark has been investing at formation for 25 years. NextDoor, NewRelic, Docker, Discord, Confluent, Cerebras, and tons of others. We aren’t making small check option “bets.” They are full commitment, board seat, and decade+ commitment to help build.
11
130
"What would you say is the biggest sacrifice that you’ve had to make in running Nvidia?
The same sacrifices other entrepreneurs make. You work really, really hard. And for a long time, nobody thinks you’re going to succeed. You’re alone in believing that you’re going to make it. The insecurities, the vulnerability, sometimes humiliation, it’s all true. Nobody talks about it, but it’s all true. CEOs and entrepreneurs are human like anybody else. And when they fail publicly, it’s embarrassing.
So when somebody said, “Jensen, with everything you have today, you wouldn’t have started it?” Like, “No, no, no, of course not.” But if I had known then that Nvidia would become what it is today, would I have started the company, are you kidding me? I would’ve sacrificed everything to do it."
wired.com/story/nvidia-hardw…
2
12
127
28,836
The AI Cloud is up for grabs.
The business will far surpass the CPU Clouds we have so far (which are huge). AWS started with object storage, what will the AI Cloud start with? There are many shots on goal - starting with DC space, starting with GPUs, starting with training,...
@FireworksAI_HQ starts with the "operational" software layer - literally run models, tune models, RL models... This is the most differentiated and defensible layer today - turns out, running large models efficiently with performance and quality is very hard. E.g., on the exact same GPUs, with the exact same OS model, Fireworks consistently gets 5x+ the performance of others.
And, to go from standing start to $280M in revenue run-rate in less than 3 years, and extremely cash efficiently is a testament to the customer love and market pull.
Kudos to this incredible team.
Now on to chase the big prize! :-)
I'm very excited to share that @FireworksAI_HQ has raised $250M in Series C funding co-led by @lightspeedvp and @IndexVentures , and participation from @sequoia Capital and @EvanticCapital, bringing up valuation to $4 billion. In total, we raised $327M from prior rounds led by @benchmark and @sequoia, with participation of strategic investors including @nvidia , @AMD , @databricks, @MongoDB , and many angel investors.
When we founded Fireworks in 2022, the vision was simple, but the problem was complex: give builders the speed, cost, and control they need to win. Our mission is to reach Artificial Autonomous Intelligence – automated product and model co-development to reach maximum quality, speed and cost-efficiency using generative AI. This round propels our execution to expand our current product of tuning and inference platform towards Artificial Autonomous Intelligence.
↗️ We have onboarded hundreds of thousands of developers to customize the latest models to create unique applications and impactful user experiences. 10x from Series B.
↗️ 10K+ organizations are running on Fireworks. Companies including Notion, Shopify, Uber, GenSpark, Vercel have built and scaled on Fireworks.
↗️ We process more than 10 trillion tokens daily. More than 20x from Series B.
We will use the new funding in the following investment:
📖 Deepen our research in post-training and inference alignment to maximize quality, speed and cost efficiency of computation, including R&D in system research and algorithmic research.
🛠️ Expand our product towards a comprehensive tool-chain for the new user experience creation lifecycle, centered around product and model co-design and automation, from model evaluation, reinforcement learning to ultra fast inference engine.
🌎 Grow our computation footprints 3x higher in the next one year, and continue R&D to minimize computational cost and maximize system utilization.
This round is not just evidence that our bet three years ago is driving a big impact, it's concrete proof of the trust and commitment we've built with our customers, partners, and our community.
Join our mission to build Artificial Autonomous Intelligence fireworks.ai/careers
4
9
126
54,613
WOW! @TimescaleDB v Amzn Timestream, @elastic Cloud v Amzn Elastic, @confluentinc Cloud v Mgd Kafka, @MongoDB Atlas v DocumentDB - the AWS rip off is slower, more expensive & less innovative.
The originals are just better.
AWS services are the Rocket Internet of open source.
I just spent one week benchmarking @TimescaleDB vs. Amazon Timestream. My results surprised even me:
✍️ 6,000x higher inserts
📈 5-175x faster queries
💰 150x-220x cheaper
...all in favor of TimescaleDB
Full results: tsdb.co/timescaledb-v-timest…
TL;DR w/ 💯pretty graphs 👇 1/
8
9
120
I think this is one of the main reasons almost no “pre-AI” companies have been able to catch an AI tailwind, despite loads of product initiatives and even more marketing. And it isn’t well understood or appreciated.
From @ericvishria on how the top AI founders are building products completely opposite of the SaaS era:
"One of the things that is really different in the AI world versus the SaaS world, is that in the SaaS world, over and over again, you had people who really understood the customer. And the problem. And then they understood a domain.
They understood what the technology was more or less capable of. But it wasn't a real question of if you could build something or not.
For example, take Salesforce, Workday, and ServiceNow. CRM existed before Salesforce. HR management existed before Workday. Same thing with ServiceNow.
So in every case, Salesforce followed Siebel. Workday followed Peoplesoft. ServiceNow followed Peregrine and Remedy, and others. So they were just kind of, cloud SaaS versions of the prior generation product.
They just understood the customers. They understood the problem. And they were just like, here's a better version. And that evolved a little bit over time in SaaS land. But that's what it is.
And so product development in that way was done by people who really understood the customer and the problems. And then just took advantage of the next wave.
And this is almost diametrically opposite of product development in the AI era. When I look at the teams that are having the most success today, they have intimate knowledge of the models. They are right on the frontier of understanding which models are better at what, and why, and when. And what they're going to be good at and what they're not going to be good at.
And what they're spending their time on, is figuring out how do I apply this capability of this model to this domain or to this user. So they're actually working inside out or technology out, versus customer problem in.
And of course, they understand the customer problem. And a lot of times they have firsthand knowledge of it. But they're really close to the metal and capability, and they're applying it. And I think this is a really different way to develop products than in SaaS.
I started my career as a product manager a long time ago, and it's almost the complete opposite of everything you learned. "Listen to the customer, understand it, then bring it back to the engineering and product teams."
If you did that right now, ask a bunch of customers what they want out of AI, and you brought it back, for the most part, it may not be possible today with today's technology. Whereas the teams that are winning right now really understand the technology and are applying it out.
And so I think this reversal matters. I think it's a big difference in terms of how companies are getting built. And maybe even the types of entrepreneurs that will be successful. I'm not sure. You're seeing some real change there.
Look at the Bret Taylor's at Sierra. That's a super, super technical founder who really gets it. Brett and Clay really get it. You look at Michael and his co-founders at Cursor. They're super technical founders and they get it.
They all really understand what these things can and can't do. And that's a pretty different dynamic relative to the way the best SaaS companies got built."
Link in bio for the full conversation going deep on the current class of startups going from zero to $100m+ in ARR within 12 months.
7
11
121
31,602
1/ Company proposes ambitious, seemingly unrealistic, plan. Board member makes “if you do that, I'll eat my hat” type comment which leads to a silly bet. Company obliterates plan. CEO shaves board member's head. I'm the board member. @Confluentinc the company.
7
13
106
Aped into $shop, $zm, $mrna.
Great time to pick up great businesses, with strong moats and amazing leaders like @tobi @ericsyuan
wagmi
5
1
104
Recently a very successful person pointed out that kids learn competition through sports (& elections). Those are zero sum, where one wins and another loses.
But that’s an awful (& wrong) mentality for most of life.
What are systemic ways to teach kids a win-win mentality?
39
4
100
Insane. "Brady, 44, is older than the three opposing NFC head coaches remaining in the playoffs."
5
1
98
“Every new CEO is good for at least one big, dumb acquisition.” – probably the best M&A banker of all time
Astonishing number of companies for sale right now.
Will be interesting which new CEOs will get New Owner's Syndrome!
5
4
93
16,477
Excited to be working with @benchmark's first company in Italy, ever: @commercelayer Especially right now, just inspiring to see entrepreneurs doing what they do. techcrunch.com/2020/05/28/it…
3
11
90
Woody Allen said “80% of success is showing up.” To which @howes28 added, “and the other 20% is refusing to leave.”
I think about that a lot when companies run into tough times. So much of long term company success is built on the team simply persisting.
1
10
87
I think that is wrong actually. If it takes 4 ppl, 4 months to get to $10M, the next team in the same space will do it even faster. No moat.
Super quick growth can just mean the product is magic to the customer, not that it is a durable business.
Otoh, there are AI applications where it takes a bit of time to get capable enough. I'm more bullish on the moat possibilities there.
5
1
83
6,010
Neat! @Amplitude_HQ Announces Confidential Submission of Draft Registration Statement for Proposed Direct Listing amplitude.com/press/amplitud…
3
7
82
As the cost of code gen goes to zero, the amount of code generated goes through the roof, and human code review falls apart.
AI has to verify all the code generated, whether from the various AI code gen tools or humans:
1) does what it is supposed to,
2) is high quality,
3) is secure.
@greptile does exactly that. Give it a try!
Greptile has raised $25M to Kill The Bug.
The round is led by Benchmark with @ericvishria joining our board.
Today, code is written by humans and a variety of coding agents like Devin, Claude Code, and Cursor.
Greptile serves as the independent and universal code review layer.
This month alone, Greptile reviewed 500M lines of code for top companies like Brex, Substack, PostHog, Bilt, and even YC’s internal software team, helping prevent 180,000+ bugs.
Alongside our Series A - we’re excited to announce Greptile v3 - a brand new agent architecture capable of catching >3x more critical bugs than Greptile v2. Available now to all users. 🧵
4
10
79
32,423
Reading the responses, many people don't think about how their paycheck comes to them.
Paychecks come through a bank acct - doesn't matter if you work for a hospital, school, or company.
& that acct likely has more than $250k in it. Not a risky investment acct, a *bank* acct.
The Treasury Department and FDIC must communicate loudly and clearly that depositors will be protected. The investors and executives of SBV should bear risk and lose. But this should not mean that workers go without paychecks on Monday or small businesses collapse.
8
12
74
45,491
No. But it isn't an acid test at all. Equity holders may get wiped out. That is totally separate from depositors.
3
76
13,562
My 10 yo son wrote a letter to Santa.
- the handwriting and spelling are 6 yo quality (and he actually typed it first so no excuses)
+ really clever and excellent use of QR codes
+ intuitive management technique
+ big hearted kid :-)
9
74
Yep!
As an investor, there is another important lens. With co-pilot approaches, you better have a damn good theory of how your startup is going to overcome the MASSIVE distribution advantage incumbents have. 50% better co-pilots will get smoked by the crappier-but-good-enough incumbent solution.
With AI-and-done approaches, or Sell the Work approaches (as my partner @sarahtavel so eloquently laid out), the tech may be harder (maybe), but your startup is up-ending the very business model of incumbents in addition to a much better proposition for customers.
Sarah's post: sarahtavel.com/p/ai-startups…
My taxonomy is:
AI-based Copilots serve as a supportive assistant, enhancing a user's decision-making and efficiency, without autonomously taking control of a task(s).
AI Agent exercises some level of autonomy. They can be special purpose or general, but general = harder.
2
7
67
23,079
This morning in a board meeting, @spenserskates referred to me as a “finance person” and it cut me to my soul.
😂
AND it was in the context of saying I was right about something, which makes the whole thing worse.
8
68
7,607
I hated this.
Not because I didn't like the content, I loved it. Not because I don't like @GuptaRK22, I love him too.
But b/c he articulated concepts - painful focus, inevitability as investment criteria, the grind required of founders - way better than I can.
Great listen.
My conversation with @GuptaRK22
Our whole conversation is about focus and the value of keeping the main thing the main thing
This is a business discussion, but also a life discussion, with someone that’s thought a lot about both
Enjoy!
joincolossus.com/episodes/87…
3
5
64
I’m watching football players play golf. Such is the depths of my desperation for live sports.
5
2
67
Parenting is mostly some exhausting shit but staging Santa’s visit in anticipation of their morning elation is one of the most joyous experiences of my life.
63
.@adidas, @boltthreads, @benchling, and mushrooms.
This morning @adidas announced a super cool new Stan Smith, made with Mylo - a biologically derived material.
Mylo was developed in conjunction with innovative materials company @boltthreads.
Introducing Stan Smith Mylo. The upper is partially created from Mylo material. Crafted from lab-grown mycelium, the root-like structure of mushrooms. Harnessing natural technologies, billions of years in the making. #MadeWithNature #EndPlasticWaste #Mylo ///
1
9
66
It is goosebump-inducing to join *1000s* at a company’s user conference, when less than 4 years ago they were little more than ambitious founders pitching @benchmark an exciting vision. @Amplitude_HQ’s Amplify this week & @confluentinc’s Kafka Summit next. So cool.
1
64
Loudcloud did this. It did not work for many employees. Also with today’s rules the loans have to be paid back BEFORE the S-1 is effective, which makes it even worse.
1
3
61
I'm no expert in banking, so YMMV but my pragmatic advice is have 6 months of operating cash in another, non-regional bank, and beyond that, don't contribute to the run.
4
5
55
27,872
I think this is right. Voice is the future.
My primary interface to AI is voice mode. I have to tell it to ignore background noise every time but it is really really good besides that.
Too bad Siri sucks.
I remember the day I received it : my first Blackberry. A few weeks later I lost it in the back of a taxi cab in Paris. But I haven’t forgotten the chiclet keyboard, its subtle click with each keypress.
A year later, the iPhone presented the world with an all glass keyboard. Who would type on a flat surface, detractors asked? In the end, all of us.
This week, I saw another user interface for a phone : dictation only.
I pressed the microphone to start dictating a message into Gmail, & the keyboard disappeared, replaced by a single thin bar at the bottom. There are three functions : delete, autosuggest, & full keyboard.
This is the future : no keyboards. Voice will become the dominant interface for humans with computers.
Speaking is at least 3 times faster than typing. It’s more human - since after all, it’s how we communicate with each other.
The technology has improved massively. Modern AI transcription models combine both voice-to-text transcription & AI to improve grammar structure & diction.
The result? Text that reads as eloquently as you think you sound in your head.
The next step : a laptop without a keyboard. Soon that collection of QWERTY squares will look as archaic as Hemingway’s Underwood typewriter - a relic of a bygone era.
8
3
61
25,667
1/ .@benchmark we've long worked with entrepreneurs from the (very) beginning. @newrelic @Nextdoor @SalesforceQuip all started when their founders were EIRs here.
Excited to continue that tradition with @ravisparikh and @__joshma and @AirplaneDev.
techcrunch.com/2021/12/09/ai…
3
7
58
I love that he published this in the Kansas City Star. Everyone in tech knows stories like this. We need to do a better job telling them where they aren’t as familiar.
3
54
Wonder if SendGrid is going to crush their quarter based on GDPR privacy policy updates going out.
3
1
57
Humbled, honored and overwhelmingly excited to be joining the incredible @Benchmark team w/ @bgurley @mattcohler @peterfenton @mitchlasky
17
17
55
Over the weekend, I used GPT-3 to write an investment memo and literally have never had more enthusiastic support for an investment. WTH.
I am completely floored. Someone run a thought of mine through GPT-3 to expand it into an explanation of what I had in mind, and it's like 95% meaningful and 90% correct. I don't think that I have seen a human explanation of my more complicated tweets approaching this accuracy :)
3
4
57
It is actually impressive how far behind Apple is with even the most basic AI capabilities. Like you have to try for Siri to be as bad as it is.
7
2
57
11,475
Many of the highest quality companies I know use the “homework assignment” in their interview process. My partners at @benchmark had me do one. And the team at @Amplitude_HQ regularly uses it. Shazam and Tatari founder @Inghelbrecht describes it well here. blog.tatari.tv/post/17241553…
2
16
56
15/ I believe this shift will continue to propel @Confluentinc to new heights. Ultimately, an Event Streaming architecture will just be how you build most applications, just like today a dev assumes virtualized compute. Which is why, in the end, it was worth shaving my head :-)
7
2
56
What do @stripe @zoom_us @elastic @servicenow and @confluentinc have in common? The great Jonathan Chadwick is on the board. Really excited we'll be able to benefit from his experience working with extraordinary companies!
We're excited to welcome Jonathan Chadwick to our board of directors as we scale and build a new category of data infrastructure around event streams. Read more here: confluent.io/press-release/c…
2
2
56
At board meeting this week - co.has 100 engineers, 70 with PhDs, from *29* countries (!) Unfortunately, visas have gotten much harder to come by. When a country of 400M people draws from the WW talent pool 20x the size, the leverage so obviously benefits America AND Americans.
6
52
Fun to be on one of my favorite podcasts with @jwangARK at @ARKInvest. Talked about a wide range of topics from AI asics to the pharma shift from chemistry to biology to SaaS and open source. @confluentinc @benchling @cerebras
🔉New Podcast Episode! In FYI | EP44, we talk to @ericvishria, General Partner at @Benchmark, about why he believes #SaaS is larger than ever before. That, and much more in #VC. Tune in!
🎧: arkinv.st/2Q79hrK
11
54
We tend to think of network effects for consumer or social apps like Snap or Slack, or marketplaces, but there are a handful of enterprise infrastructure network effects that are going to result in monster companies.
5. More importantly though, data streaming is about connectivity, so like the telephone system you want to use the network with the most connectivity. It’s about the ecosystem.
7
51
Astonishing. And against 0 interest rate backdrop, the net present value of those companies’ future cash flows is worth so much more. Which at least partially justifies the massive valuation disparity btw growth and non-growth.
Fun fact: Only 10pct of s&p companies grow 20pct or more. They have 100B of cash. Meanwhile the other 90pct have 1.8T in net debt. All the growth companies have the cash, the old companies the debt. Good luck competing against the innovators!
3
2
53
Incredibly smart acquisition by @Cisco. Can’t think of an easier to use, more loved product in security than @duosec. (Nor an enterprise business that scaled more efficiently and effectively). Congrats to @dugsong and the entire team!
cnbc.com/2018/08/02/cisco-bu…
8
51
"Optimism is the engine of capitalism. When u look @ big successes its bc some1 tried something they shouldn’t have" compoundingmyinterests.com/c…
21
49
One of the untold stories of the SaaS explosion is as new areas are "quantified," new "systems of record" are created. And those SoR have created the largest enterprise companies. Am loving working with @Benchling, a new SoR for life sciences R&D. @forbes forbes.com/sites/amyfeldman/…
6
52
That is a very useful skill for an entrepreneur! Necessary, but definitely not sufficient :-)
3
1
50
I'm 4 years in on @tonal and still a ~3x a week user. It is the best. Variety of exercises/workouts, not having to think about weight settings, different modes... it is excellent.
3
3
50
23,767
Another 10+ year “overnight” success story! Early traction, merging with a competitor, changing the business model - through it all awesome leadership and execution. Congrats to the team!
While we pause today to celebrate, tomorrow we’ll be back at work creating a better future of work. Thank you to our hardworking employees, freelancers, clients, and @nasdaq for an incredible day. Onward!
2
51
Every day there are new comparisons of AI to internet, mobile or cloud. I think ppl compare to those tech shifts b/c they happened to live through them, not b/c they are actually good analogs.
Transistors are a much better analog. Enabling technology, impacting everything, in extremely unpredictable ways. But of course that shift started before most of us writing were born...
6
4
51
8,487
At the @Amplitude_HQ DL I compared a co. going public to an athlete going pro. Very few get the oppty to compete at the highest level for way bigger stakes.
Even rarer is when it’s obvious the rookie is going to be a 🌟. Just wow.
Congrats again to @spenserskates & the team!
So @spenserskates' AMA (on the occasion of @Amplitude_HQ's IPO) is just an all-timer HN fairytale:
news.ycombinator.com/item?id…
- disrupting competitor @Mixpanel
- Mixpanel founder @suhail congratulating them
- @patio11 "Charge more" taken to heart
- @bgurley style IPO rant
2
1
49
It seems like ultimately builders don't want low code, they want "easy code" - prototype quickly and easily, but then open it up, fine tune, tweak and tinker to get it exactly right for real-world use.
Love the @AirplaneDev approach.
5
50
.@lqiao, CEO of Fireworks, said to me "finding leaders who have the right balance of Pattern Matching and Path Finding has been particularly difficult."
I thought it was an excellent articulation of the difficulty of leading in AI natives.
And related to @rabois' post on figuring out the new org in the face of AI.
By and large, SaaS leaders dropping into rapidly scaling AI-native co's are struggling. It is like dropping rural Germans into the middle of modern Tokyo.
The AI CEOs only know this pace, tech, sales cycle.. and can’t explain how these leaders should reframe. The SaaS people talk about quota capacity, qtrly roadmaps, & enterprise, which don’t mean anything to the AI CEOs and are generally irrelevant, at least for now.
Bridging this is hard but critical. The productive combo of this wave & scaling expertise would turbocharge these already ripping companies.
1
4
45
9,814
So true.
Two interesting ones:
Chemistry-based drugs > biology-based - changes all tooling, development process - once in a century shift
People as primary users of sw > Other sw as primary user of sw - drives rise of API-everything, digital automation, importance of real-time
Breakout technology companies are often created by leveraging one (or more) major platform shifts.
Personal Computing > Apple, Microsoft
Home Broadband > Youtube, Netflix
Smartphones > WhatsApp, Instagram
What are the most interesting (non-obvious) recent platform shifts?
1
2
44
Everyday more and more traditional companies realize that getting digital right is existential, not nice-to-have. And their IT stacks will have to turn over for them to be competitive with digital natives. So as much as cloud and SaaS have run, there is probably a ton yet to go.
"Nike's CEO Donahoe said on Thursday it would now aim for digital to account for 50% of its overall business, up from 30% recorded in the reported quarter."
When management knows big change is coming the job is to get ahead of it. Denial is a river in Egypt, not a strategy.
3
43
OR, maybe competition from DLs (and even SPACs) has forced more integrity in the pricing process which is resulting in more favorable terms for the company, employees, and shareholders.
Turns out capitalism and competition work!
4
1
42
Really enjoyed doing this podcast. Think we covered a lot of the excitement and magic around AI startups, and also the challenges investing in hypergrowth.
Thank you so much for having me on!
This convo with @ericvishria is one of my favorite ever.
I am so grateful to learn from him and the @benchmark approach.
-Just *how fast* are some of these AI companies growing?
-is the revenue durable?
-What does Benchmark look for?
(00:07:28) Exponential Rise of AI Startups
(00:10:21) Diverse Revenue Paths in AI Startups (00:14:10) Vertical AI / LLM
(00:18:25) Product Success through Market Awareness
(00:20:05) Balancing Humans with AI Automation
(00:35:22) The Value of Venture Capital Partnerships (00:37:48) Partnerships in Early-Stage Investing
(00:47:10) The Core of Investing Decisions
(00:54:22) Supporting Founders
3
1
42
11,358