Alex Heath (@alexeheath) header
Alex Heath (@alexeheath) profile picture

Sources / Sound Ventures

Joined May 2008
AWS CEO on building a $1 trillion business Amazon thinks its cloud business could generate $1 trillion a year within the next five to ten years. I sit down with AWS CEO @mattsgarman in Seattle to understand what gives him that confidence and what could get in the way. Garman explains how AI is pushing companies to move their core operations into the cloud, accelerating the business AWS spent two decades building. We discuss why he thinks inference, the computing required to run AI models, is still below 1 percent of its eventual scale. He says AWS is adding one GPU every 14 seconds and expects capacity shortages for another couple of years. I ask how he decides who gets the GPUs, why he holds capacity back for startups that don’t exist yet, and where he sees risk in the industry’s spending. Recorded right after the midterm elections, our conversation also covers the backlash against data centers, what the industry needs to do to win over communities, and why he sees deserts as more promising than space. I also ask about the reset inside Amazon’s Nova team and whether it still intends to compete at the frontier of AI. We get into selling Trainium chips outside AWS, the rise of inference providers like Baseten and Fireworks, and Amazon’s big research bet on using mathematical proofs to help AI agents follow rules. Timestamps: 00:00 AWS’s trillion-dollar bet 02:24 How AI is accelerating cloud growth 09:33 The cost of scaling AI 14:10 The risks of the compute boom 17:02 Why inference is still early 20:03 The backlash against data centers 28:21 Data centers in space, oceans and deserts 31:38 The rise of AI inference providers 33:53 AWS’s full-stack strategy 36:52 Why AWS is selling Trainium chips 38:11 How AWS allocates scarce compute 40:10 AWS’s frontier AI ambitions 41:37 Making AI agents safe 46:31 Open vs. closed AI models 49:14 Running AWS inside Amazon Links: Spotify: open.spotify.com/episode/4uE… YouTube: youtu.be/1C_MErgk2Z0 Substack: sources.news/p/aws-ceo-matt-… Apple Podcasts: sources.news/p/aws-ceo-matt-… Thanks to the show’s sponsors: @mercury, @meetgranola , and @Atlassian
9
17
10
135
65,764
My takeaway from moderating KPMG’s tech M&A panel last week: a company’s most valuable asset is a leadership team that can move quickly with conviction, especially now. I left expecting more big deals, including private equity teaming up with AI cos to remake existing businesses. Thanks to the KPMG team for inviting me to moderate!
3
2
7
2,784
“Physics is the law. Everything else is a recommendation.” @elonmusk told me that at DealBook in 2023. Today, SpaceX caught Starship after 14 orbits of Earth. Plenty doubted it was possible. Just extraordinary. @SpaceX.
431
1,931
189
16,650
1,013,165
Alex Heath retweeted
AWS CEO Matt Garman says AWS is at a $180B+ run rate but still has room to grow: most workloads are still on-premises, and AI is pushing companies to move them to the cloud “I’ve been with AWS for 20 years and we’ve been growing that whole time. And I think we over time have more and more gotten conviction that the business opportunity is just really, really, really large” “It’s surprising, but at the point that we’re at now, we’re at $182 billion run rate business. And so you’re over $180 billion and you’re growing at 39%. There’s not many companies that hit that spot” “As we’ve seen, we’ve actually increased our year-over-year growth rate each of the last six quarters” “We’ve always had this view that the majority of workloads still live in on-premise worlds. And as big as the cloud has gotten, there’s still a lot of data centers that people have” “Customers now see that there is a massive opportunity that AI affords them to gain efficiencies, to build new businesses. And they don’t get that benefit if their core workloads aren’t in a cloud environment. If they’re in a VMware cluster in their legacy data center, they just have a hard time having agents talk to those pieces”
AWS CEO on building a $1 trillion business Amazon thinks its cloud business could generate $1 trillion a year within the next five to ten years. I sit down with AWS CEO @mattsgarman in Seattle to understand what gives him that confidence and what could get in the way. Garman explains how AI is pushing companies to move their core operations into the cloud, accelerating the business AWS spent two decades building. We discuss why he thinks inference, the computing required to run AI models, is still below 1 percent of its eventual scale. He says AWS is adding one GPU every 14 seconds and expects capacity shortages for another couple of years. I ask how he decides who gets the GPUs, why he holds capacity back for startups that don’t exist yet, and where he sees risk in the industry’s spending. Recorded right after the midterm elections, our conversation also covers the backlash against data centers, what the industry needs to do to win over communities, and why he sees deserts as more promising than space. I also ask about the reset inside Amazon’s Nova team and whether it still intends to compete at the frontier of AI. We get into selling Trainium chips outside AWS, the rise of inference providers like Baseten and Fireworks, and Amazon’s big research bet on using mathematical proofs to help AI agents follow rules. Timestamps: 00:00 AWS’s trillion-dollar bet 02:24 How AI is accelerating cloud growth 09:33 The cost of scaling AI 14:10 The risks of the compute boom 17:02 Why inference is still early 20:03 The backlash against data centers 28:21 Data centers in space, oceans and deserts 31:38 The rise of AI inference providers 33:53 AWS’s full-stack strategy 36:52 Why AWS is selling Trainium chips 38:11 How AWS allocates scarce compute 40:10 AWS’s frontier AI ambitions 41:37 Making AI agents safe 46:31 Open vs. closed AI models 49:14 Running AWS inside Amazon Links: Spotify: open.spotify.com/episode/4uE… YouTube: youtu.be/1C_MErgk2Z0 Substack: sources.news/p/aws-ceo-matt-… Apple Podcasts: sources.news/p/aws-ceo-matt-… Thanks to the show’s sponsors: @mercury, @meetgranola , and @Atlassian
2
1
1
6
1,986
Alex Heath retweeted
Q: Could this be AI related? Mullin: We call it SI around here.
1,732
950
1,399
8,833
3,091,347
Alex Heath retweeted
@disruptivetech's Founder and CEO Alex Davis sat down with @alexeheath at Disruptive’s offices in Dallas for the Sources podcast. They covered a multitude of topics but in particular a discussion about the Flagship Fund, how the firm invests, and where Davis sees opportunity today and in the future. Watch or listen to the full conversation here: youtube.com/watch?v=dLPbh4Bm…
1
5
1,618
Woke up to hitting 300k subs for Sources pod on YouTube! All added in last two months. 🥳🚀
17
2
1
191
11,638
Confident tweet about a matter I've read a series of speculative tweets about, strongly asserting my insider wisdom, elevating my status above my peers on the feed
14
13
7
179
20,665
the source crossover of the century?
Sources Podcast's Alex Heath (@alexeheath), now a VC at Sound Ventures, says we're still "so early" on AI inference "I just had the CEO of AWS on the podcast telling me it's not even 1% of what he thinks it will be." "People are having a hard time seeing the exponential of inference in the same way they didn't see it necessarily for the foundational models a few years ago." "No one thought Anthropic would be IPOing at over $1T 3 years ago, in the same way I think we'll be talking about inference in a few years."
7
4
6
40
7,832
Alex Heath retweeted
united just banned ai agents from booking your flights
4
3
1
17
3,206
Alex Heath retweeted
Replying to @ajassy
@ajassy told investors @awscloud could be a trillion-dollar revenue business in the next five to 10 years. @alexeheath did the math in front of the man who has to deliver it, AWS CEO @mattsgarman: "based on the run rate you're already on, that requires roughly 40% compound growth over five years or 18-ish over 10 years." Then he asked what changed: "what happened between Q2 and Q3?" Garman: "I've been with AWS for 20 years and we've been growing that whole time. And I think we over time have more and more and more gotten conviction that the business opportunity is just really, really, really large." Then where it stands today: "we're at $182 billion run rate business." And the trend: "we've actually increased our year-over-year growth rate each of the last six quarters." The numbers behind it: → Jassy in @amazon's Q3 earnings release: "There has never been a trillion dollar revenue business, and we may have two of them in the next 5-10 years." The other one is its stores business. → AWS revenue that quarter: $46.0B, up 39%, its fastest growth in nearly five years. → Later in the same interview Garman said demand is "far outstripping the supply" and that he expects AWS to be short of capacity "for the next couple of years". My read: the target is less wild than it sounds and harder than the answer suggests. At 39%, AWS is already growing at almost exactly the pace Heath's five-year math needs, but a business that is short of capacity grows as fast as it can build, whatever customers want to buy. So the trillion depends less on demand than on chips, power and buildings arriving on time. From @alexeheath's Sources interview with @mattsgarman.
1
2
1
2,104
Alex Heath retweeted
For the past four years, Jensen Huang ran the Central Bank of AI. Today he resigned as chief executive of Nvidia. What are the implications for the AI economy?
23
25
9
153
18,852
Alex Heath retweeted
Here are the emails sent to Nvidia employees: Company meeting Nov 18
112
121
32
2,450
1,275,748
Alex Heath retweeted
Google's @sundarpichai tells @alexeheath of Sources that Gemini 4’s early token usage is triple Gemini 3’s
1
3
32
3,982
AWS CEO tells me the AI inference market is "super early" "I don't even think it's at 1%"
AWS CEO on building a $1 trillion business Amazon thinks its cloud business could generate $1 trillion a year within the next five to ten years. I sit down with AWS CEO @mattsgarman in Seattle to understand what gives him that confidence and what could get in the way. Garman explains how AI is pushing companies to move their core operations into the cloud, accelerating the business AWS spent two decades building. We discuss why he thinks inference, the computing required to run AI models, is still below 1 percent of its eventual scale. He says AWS is adding one GPU every 14 seconds and expects capacity shortages for another couple of years. I ask how he decides who gets the GPUs, why he holds capacity back for startups that don’t exist yet, and where he sees risk in the industry’s spending. Recorded right after the midterm elections, our conversation also covers the backlash against data centers, what the industry needs to do to win over communities, and why he sees deserts as more promising than space. I also ask about the reset inside Amazon’s Nova team and whether it still intends to compete at the frontier of AI. We get into selling Trainium chips outside AWS, the rise of inference providers like Baseten and Fireworks, and Amazon’s big research bet on using mathematical proofs to help AI agents follow rules. Timestamps: 00:00 AWS’s trillion-dollar bet 02:24 How AI is accelerating cloud growth 09:33 The cost of scaling AI 14:10 The risks of the compute boom 17:02 Why inference is still early 20:03 The backlash against data centers 28:21 Data centers in space, oceans and deserts 31:38 The rise of AI inference providers 33:53 AWS’s full-stack strategy 36:52 Why AWS is selling Trainium chips 38:11 How AWS allocates scarce compute 40:10 AWS’s frontier AI ambitions 41:37 Making AI agents safe 46:31 Open vs. closed AI models 49:14 Running AWS inside Amazon Links: Spotify: open.spotify.com/episode/4uE… YouTube: youtu.be/1C_MErgk2Z0 Substack: sources.news/p/aws-ceo-matt-… Apple Podcasts: sources.news/p/aws-ceo-matt-… Thanks to the show’s sponsors: @mercury, @meetgranola , and @Atlassian
7
2
4
42
12,252
Alex Heath retweeted
AWS's CEO described a pattern i've also been observing. the Fable 5.2 launch is a live example of it in his interview with @alexeheath , matt garman says every 50% drop in inference prices (the cost of running a model) has brought ~10x more usage, with revenue still rising. asked how far along inference is, he said "I don't think it's even 1%" now look at fable 5.2. anthropic cut the token price 20%, from $10/$50 to $8/$40 per million tokens in/out. but on semianalysis's GDPx benchmark (professional tasks), the average cost per finished task went up 31%, from $67.16 to $88.15 the reason is how hard the new model checks its own work. the last version wrote ~20 test cases per job. 5.2 sometimes writes hundreds or thousands, and spins up helper agents (subagents) to recheck the answer. all that checking also helped it take #1 on the benchmark so the price cut didn't shrink the bill. it bought more thinking per job i think this is what gets lost when ppl model AI compute demand off falling token prices. cheaper tokens get spent, and each smarter model finds (more) new ways to spend them h/t @SemiAnalysis_, @latentspacepod teapawt.net/alexeheath/status/2120…
AWS CEO on building a $1 trillion business Amazon thinks its cloud business could generate $1 trillion a year within the next five to ten years. I sit down with AWS CEO @mattsgarman in Seattle to understand what gives him that confidence and what could get in the way. Garman explains how AI is pushing companies to move their core operations into the cloud, accelerating the business AWS spent two decades building. We discuss why he thinks inference, the computing required to run AI models, is still below 1 percent of its eventual scale. He says AWS is adding one GPU every 14 seconds and expects capacity shortages for another couple of years. I ask how he decides who gets the GPUs, why he holds capacity back for startups that don’t exist yet, and where he sees risk in the industry’s spending. Recorded right after the midterm elections, our conversation also covers the backlash against data centers, what the industry needs to do to win over communities, and why he sees deserts as more promising than space. I also ask about the reset inside Amazon’s Nova team and whether it still intends to compete at the frontier of AI. We get into selling Trainium chips outside AWS, the rise of inference providers like Baseten and Fireworks, and Amazon’s big research bet on using mathematical proofs to help AI agents follow rules. Timestamps: 00:00 AWS’s trillion-dollar bet 02:24 How AI is accelerating cloud growth 09:33 The cost of scaling AI 14:10 The risks of the compute boom 17:02 Why inference is still early 20:03 The backlash against data centers 28:21 Data centers in space, oceans and deserts 31:38 The rise of AI inference providers 33:53 AWS’s full-stack strategy 36:52 Why AWS is selling Trainium chips 38:11 How AWS allocates scarce compute 40:10 AWS’s frontier AI ambitions 41:37 Making AI agents safe 46:31 Open vs. closed AI models 49:14 Running AWS inside Amazon Links: Spotify: open.spotify.com/episode/4uE… YouTube: youtu.be/1C_MErgk2Z0 Substack: sources.news/p/aws-ceo-matt-… Apple Podcasts: sources.news/p/aws-ceo-matt-… Thanks to the show’s sponsors: @mercury, @meetgranola , and @Atlassian
11
2
12
5,045
Amazon AWS CEO expects to be short on compute "for the next couple of years" "Right now the demand is far outstripping the supply"
AWS CEO on building a $1 trillion business Amazon thinks its cloud business could generate $1 trillion a year within the next five to ten years. I sit down with AWS CEO @mattsgarman in Seattle to understand what gives him that confidence and what could get in the way. Garman explains how AI is pushing companies to move their core operations into the cloud, accelerating the business AWS spent two decades building. We discuss why he thinks inference, the computing required to run AI models, is still below 1 percent of its eventual scale. He says AWS is adding one GPU every 14 seconds and expects capacity shortages for another couple of years. I ask how he decides who gets the GPUs, why he holds capacity back for startups that don’t exist yet, and where he sees risk in the industry’s spending. Recorded right after the midterm elections, our conversation also covers the backlash against data centers, what the industry needs to do to win over communities, and why he sees deserts as more promising than space. I also ask about the reset inside Amazon’s Nova team and whether it still intends to compete at the frontier of AI. We get into selling Trainium chips outside AWS, the rise of inference providers like Baseten and Fireworks, and Amazon’s big research bet on using mathematical proofs to help AI agents follow rules. Timestamps: 00:00 AWS’s trillion-dollar bet 02:24 How AI is accelerating cloud growth 09:33 The cost of scaling AI 14:10 The risks of the compute boom 17:02 Why inference is still early 20:03 The backlash against data centers 28:21 Data centers in space, oceans and deserts 31:38 The rise of AI inference providers 33:53 AWS’s full-stack strategy 36:52 Why AWS is selling Trainium chips 38:11 How AWS allocates scarce compute 40:10 AWS’s frontier AI ambitions 41:37 Making AI agents safe 46:31 Open vs. closed AI models 49:14 Running AWS inside Amazon Links: Spotify: open.spotify.com/episode/4uE… YouTube: youtu.be/1C_MErgk2Z0 Substack: sources.news/p/aws-ceo-matt-… Apple Podcasts: sources.news/p/aws-ceo-matt-… Thanks to the show’s sponsors: @mercury, @meetgranola , and @Atlassian
2
7
7
77
19,932
AWS CEO Matt Garman says AWS lands a GPU every 14 seconds and the world doesn't do one launch a day, so space data centers are a decade or two out [ Alex: What do you think about data centers in space? ] "I think that it is highly unlikely to be economically feasible or practical in the next decade or two." [ Alex: Elon's trying to fix that. ] "Like it's a great, I appreciate it, but one person is not going to fix it... we as a world don't do one launch a day today." "We're landing in a data center today, one GPU every 14 seconds." [ Alex: Wow. ] "So, we don't do a launch a day for the world... a rack of GPUs, you know, might weigh a ton."
Intel CEO Lip-Bu Tan recalls meeting Trump after Trump called for his resignation, and Trump ending it by asking if the government could own 10% of Intel "That phone number is mainly to get me the appointment to see President Trump." "First time I went to Oval Office. It's a beautiful office, but I focus on him." "I think they have some misinformation and I just telling the truth. And I born in Malaysia, I grown up in Singapore, and I did my nuclear engineering at MIT. And then I've been US citizens after that, never live in other country. This is my home." "At the end he said, 'Well, today is 5:00 Monday. Friday I want to announce. Do you mind US government own 10% of the company?'"
1
1
1
9
3,353
Alex Heath retweeted
STATE OF THE TECH JOB MARKET IN 2026 Twice a year, I dig into trends in hiring and job openings. This is the fifth report I've done, and the most surprising yet: 1. PMs are winning in the AI era 2. Engineering demand has stalled 3. Design openings are at a record low 4. Every role now requires AI skills 5. Tech jobs are concentrating in the Bay Area 6. Remote roles are at a record low 7. The tech job market is growing fast, but mostly outside of software (gasp!) Full report here: lennysnewsletter.com/p/state…
345
594
334
5,329
1,651,146