1 00:00:00,040 --> 00:00:02,800 Just a question of who and when, not if anymore. 2 00:00:02,920 --> 00:00:06,480 Like if you talk to people like 5-10 years ago in quantum, you 3 00:00:06,480 --> 00:00:08,720 know, other vendors were talking about like having to build like 4 00:00:08,720 --> 00:00:12,800 this gigantic football field size quantum computer of you 5 00:00:12,800 --> 00:00:15,920 know, we don't need to do that anymore, then that's why we are 6 00:00:15,920 --> 00:00:19,160 so confident in the 2029 date. There's a lot of people talking 7 00:00:19,160 --> 00:00:21,280 about quantum computing that have never built a quantum 8 00:00:21,280 --> 00:00:24,600 computer but still hype it like they have. 9 00:00:25,520 --> 00:00:28,440 And then there's some of us who have built quantum computers and 10 00:00:28,440 --> 00:00:30,280 made them available for people to use. 11 00:00:30,760 --> 00:00:32,520 I actually believe this is not a good idea. 12 00:00:32,520 --> 00:00:36,160 I don't know. I mean, in fairness, I wouldn't 13 00:00:36,160 --> 00:00:40,800 need to think through his hypothesis for for why that is. 14 00:00:41,560 --> 00:00:45,280 Disruptors and curious minds, we are back in the realm of 15 00:00:45,280 --> 00:00:47,280 quantum. Listen, here's the question. 16 00:00:47,640 --> 00:00:52,320 Can scalable quantum centric supercomputing ever tackle the 17 00:00:52,320 --> 00:00:55,440 toughest scientific and humanitarian challenges? 18 00:00:55,640 --> 00:00:58,600 IBM says it can. We have a gentleman on the show 19 00:00:58,720 --> 00:01:00,120 that's going to help us unpack it? 20 00:01:00,120 --> 00:01:03,800 Scott Crowder is the vice president of IBM Quantum 21 00:01:03,800 --> 00:01:05,519 Adoption and Business Development. 22 00:01:05,920 --> 00:01:08,840 Before that, he spent seven years as the CTO and vice 23 00:01:08,840 --> 00:01:11,360 president of strategy of IBM Systems. 24 00:01:11,600 --> 00:01:14,080 So he knows infrastructure. He knows the history of 25 00:01:14,080 --> 00:01:16,720 computing. This is a show you will not want 26 00:01:16,720 --> 00:01:20,200 to miss. So I guess the message today is 27 00:01:20,200 --> 00:01:23,720 that computers are real. The concept started out 45 years 28 00:01:23,720 --> 00:01:25,800 ago. So it's it's not an old concept 29 00:01:25,920 --> 00:01:29,160 and it's been about 10 years. It's been 10 years in a couple 30 00:01:29,160 --> 00:01:33,280 weeks actually since IBM put a really baby five qubit chronic 31 00:01:33,280 --> 00:01:36,720 computer on the cloud and people who could actually run a program 32 00:01:36,720 --> 00:01:38,280 on a quantum computer and get a result. 33 00:01:38,280 --> 00:01:40,240 So it was kind of like sci-fi come to life. 34 00:01:40,400 --> 00:01:45,320 And then I think there's a lot of hype and confusion in the 35 00:01:45,320 --> 00:01:46,800 field. There's a lot of people talking 36 00:01:46,800 --> 00:01:48,920 about quantum computing that have never built a quantum 37 00:01:48,920 --> 00:01:51,720 computer. And then there's some of us who 38 00:01:51,720 --> 00:01:54,200 have built quantum computers and made them available for people 39 00:01:54,200 --> 00:01:57,640 to use. Where we are today is that we've 40 00:01:57,960 --> 00:02:02,520 now have quantum computers that can run a quantum algorithm that 41 00:02:02,520 --> 00:02:06,240 is too complex to run on any classical hardware. 42 00:02:06,560 --> 00:02:10,039 The confusing part, and This is why you get the bipolar response 43 00:02:10,720 --> 00:02:14,920 is that doesn't mean it's better then running a classical 44 00:02:14,920 --> 00:02:18,320 algorithm that can run on that classic computer to solve the 45 00:02:18,320 --> 00:02:21,320 problem in a different way. And that's like the next step in 46 00:02:21,320 --> 00:02:24,800 quantum computing is when the quantum way of doing it is 47 00:02:24,800 --> 00:02:26,320 better than the classical way of doing that. 48 00:02:26,920 --> 00:02:29,800 And because we're not there yet, you still have two camps in my 49 00:02:29,800 --> 00:02:32,320 personal opinion. And until we hit that threshold 50 00:02:32,320 --> 00:02:36,920 and prove it to people beyond a reasonable doubt, then those 51 00:02:36,920 --> 00:02:39,360 camps will converge into one. This is where we're headed. 52 00:02:39,360 --> 00:02:42,000 I mean, we have and I want to ask you a couple of questions. 53 00:02:42,000 --> 00:02:46,160 Historically, why certain reference architectures worked 54 00:02:46,400 --> 00:02:49,200 and why certain reference architectures got pushed to the 55 00:02:49,200 --> 00:02:52,920 side. So example TCIPTCPIP, everyone 56 00:02:52,920 --> 00:02:55,240 got on board, right? Unix everyone kind of got on 57 00:02:55,240 --> 00:02:57,600 board. OSI, everyone kind of got on 58 00:02:57,600 --> 00:03:01,000 board. Vetiver bushes, me Max, not so 59 00:03:01,000 --> 00:03:03,640 much. It was early and but it ended up 60 00:03:03,640 --> 00:03:06,600 being the kind of the same architecture that Berners Lee 61 00:03:06,600 --> 00:03:11,280 did with the World Wide Web. So what do you think makes a 62 00:03:11,280 --> 00:03:16,400 reference architecture powerful, compelling, and one that'll 63 00:03:16,400 --> 00:03:18,080 serve as scaffolding moving forward? 64 00:03:18,240 --> 00:03:19,640 Yeah, I think it's probably two things. 65 00:03:19,640 --> 00:03:23,160 One, it actually has to solve a problem like so that's kind of 66 00:03:23,160 --> 00:03:25,600 like fundamental and some of the examples that you gave didn't 67 00:03:25,600 --> 00:03:26,880 solve a problem but it didn't have. 68 00:03:26,920 --> 00:03:30,080 The second one is needs to get adopted for us, we're in the 69 00:03:30,080 --> 00:03:34,640 fairly early days of leveraging, you know, quantum computation to 70 00:03:34,640 --> 00:03:37,840 solve problems in a in a practical real way. 71 00:03:38,520 --> 00:03:42,440 And for us, the quantum centric supercomputing like reference 72 00:03:42,440 --> 00:03:47,520 architecture is kind of our perspective on what 73 00:03:47,520 --> 00:03:51,280 computational resources and how you put them together do you 74 00:03:51,280 --> 00:03:54,400 need to actually run a problem and solve a problem. 75 00:03:55,880 --> 00:03:59,560 And because it's early days in quantum computing, I think, you 76 00:03:59,560 --> 00:04:03,120 know, part of the the fuzz, like you mentioned before in the two 77 00:04:03,120 --> 00:04:06,560 camps, you got a similar fuzz in terms of what a quantum computer 78 00:04:06,560 --> 00:04:09,080 is. Because for a lot of people were 79 00:04:09,080 --> 00:04:13,360 still in this like if I come to life, you know, the excitement 80 00:04:13,360 --> 00:04:16,560 about flux capacitors and spooky action in a distance and 81 00:04:16,560 --> 00:04:19,760 teleportation like being used to do computation. 82 00:04:20,600 --> 00:04:23,080 People lose sight of the fact that the end of the day, all 83 00:04:23,080 --> 00:04:26,720 we're doing here is building a different kind of computer that 84 00:04:26,720 --> 00:04:30,280 runs a different kind of algorithm, which is for 85 00:04:30,360 --> 00:04:33,000 particular subroutines and a larger workflow. 86 00:04:33,320 --> 00:04:37,320 And quantum centric supercomputing for us is how do 87 00:04:37,320 --> 00:04:41,960 you take that acceleration for that subroutine that quantum 88 00:04:42,120 --> 00:04:45,160 computation is really good at and make it part of a larger 89 00:04:45,160 --> 00:04:46,880 workflow? You have these different 90 00:04:46,880 --> 00:04:49,200 modalities, you have trap tie and you have spin qubits. 91 00:04:49,200 --> 00:04:51,520 You have superconducting again those examples that Joe 92 00:04:51,520 --> 00:04:54,840 mentioned, one survived, best didn't. 93 00:04:54,840 --> 00:04:57,840 Is quantum computing the same in that one modality will survive 94 00:04:57,840 --> 00:05:00,160 or is it different in the all the modalities could come 95 00:05:00,160 --> 00:05:05,360 together and solve those problems as one global on some 96 00:05:05,360 --> 00:05:08,600 computer Or will they fall by the wayside too? 97 00:05:09,160 --> 00:05:11,880 I think that probably maybe it's my background, but to me the 98 00:05:11,880 --> 00:05:14,880 best analogy is probably silicon versus germanium versus gallium 99 00:05:14,880 --> 00:05:17,360 arsenide versus Indian phosphide, etcetera, etcetera. 100 00:05:18,760 --> 00:05:22,520 And it's not like germanium, which was the original basis for 101 00:05:22,520 --> 00:05:27,120 transistor radios back in the 50s ever completely went away or 102 00:05:27,160 --> 00:05:30,280 Indium phosphide for specialized things ever really went away. 103 00:05:30,360 --> 00:05:33,160 But the silicon transistor CMOS transistors just became 104 00:05:33,160 --> 00:05:35,440 dominant. I'm old enough to just joined 105 00:05:35,440 --> 00:05:39,120 IBM at the beginning, at the end of the debate between bipolar 106 00:05:39,120 --> 00:05:43,080 and CMOS And it's I think it's more similar to something like 107 00:05:43,080 --> 00:05:46,080 that where, you know, in my opinion, yes, one of the 108 00:05:46,080 --> 00:05:50,600 modalities is probably going to become dominant and once it 109 00:05:50,600 --> 00:05:53,600 does, it's going to be hard for the other modalities to catch up 110 00:05:53,600 --> 00:05:56,600 just because of the like infrastructure etcetera, 111 00:05:56,600 --> 00:05:58,000 etcetera that's built up around it. 112 00:05:58,480 --> 00:06:02,760 For us, superconducting qubits is the right trade off of being, 113 00:06:02,960 --> 00:06:05,720 you know, fairly easily manufacturer because you use 114 00:06:05,720 --> 00:06:09,920 silicon fabrication techniques and also fast, which makes it 115 00:06:10,360 --> 00:06:12,560 feasible as a computational platform. 116 00:06:13,040 --> 00:06:14,840 You're not going to hear about athletic greens. 117 00:06:15,200 --> 00:06:18,400 We have no electrolyte drinks or AI agents to sell you. 118 00:06:19,040 --> 00:06:21,720 The show you're listening to, Thinking on Paper is funded 119 00:06:21,720 --> 00:06:25,480 entirely by me and Jeremy. So this is an advert for us 120 00:06:26,120 --> 00:06:28,560 because we sponsor ourselves and we need your help. 121 00:06:28,640 --> 00:06:32,240 So please subscribe wherever you're listening or watching 122 00:06:32,240 --> 00:06:34,880 Thinking on Paper. And if you're feeling really 123 00:06:34,880 --> 00:06:37,080 generous and kind, please leave a comment. 124 00:06:37,640 --> 00:06:42,720 And now back to Scott and IBM. How important you mentioned? 125 00:06:42,720 --> 00:06:45,360 You referenced this too. How important is accessibility 126 00:06:45,920 --> 00:06:48,760 to the technology? From our perspective is critical 127 00:06:48,760 --> 00:06:52,320 from like overly simply from our perspective, there's two things 128 00:06:52,320 --> 00:06:55,880 we need to do to like take quantum computing and make it to 129 00:06:55,880 --> 00:06:59,000 unlock its value. The 1st is we need to build more 130 00:06:59,000 --> 00:07:00,880 and more powerful quantum computers. 131 00:07:00,880 --> 00:07:02,040 That's both hardware and software. 132 00:07:02,320 --> 00:07:05,640 That's kind of obvious. That's kind of mostly on us. 133 00:07:05,880 --> 00:07:10,040 But the second one was equally important is algorithmic 134 00:07:10,040 --> 00:07:13,560 discovery of like stuff you run on quantum computers. 135 00:07:14,000 --> 00:07:19,040 I think we take it for granted that for binary math, we've been 136 00:07:19,040 --> 00:07:22,000 developing algorithms for centuries as humans. 137 00:07:22,800 --> 00:07:25,760 And when digital, quantum, digital classical computers came 138 00:07:25,760 --> 00:07:29,960 along in the 40s and the 50s, there was this huge increase in 139 00:07:29,960 --> 00:07:33,280 algorithmic development in the 50s and 60s that kind of like 140 00:07:33,280 --> 00:07:36,320 underpin a lot of what we run on classical computers today. 141 00:07:36,640 --> 00:07:41,720 We're really early stages in humans thinking about algorithms 142 00:07:41,720 --> 00:07:45,000 that run using quantum information science and how to 143 00:07:45,000 --> 00:07:49,200 apply them for for applications. Part of it is the algorithmic 144 00:07:49,200 --> 00:07:51,680 discovery. So part of it is getting the 145 00:07:52,280 --> 00:07:55,440 computers out there so people can explore and make sure that 146 00:07:55,440 --> 00:07:58,200 works. What is I think a little bit 147 00:07:58,200 --> 00:08:02,040 less appreciated, and bear with me is probably a long answer, 148 00:08:02,040 --> 00:08:06,920 but little bit less appreciated is how do I think about the 149 00:08:06,920 --> 00:08:12,320 problem and break up the problem so it I can leverage this tool 150 00:08:12,520 --> 00:08:16,520 effectively. And there's been a huge like sea 151 00:08:16,520 --> 00:08:20,480 change in that in the last couple years that I think people 152 00:08:20,480 --> 00:08:22,760 are not appreciating quite as much as they should. 153 00:08:23,400 --> 00:08:27,160 So that's kind of like what's underneath the need for quantum 154 00:08:27,160 --> 00:08:31,160 centric supercomputing is OK, I've got this new computational 155 00:08:31,160 --> 00:08:35,799 tool as quantum is good at certain things, but it's not big 156 00:08:35,799 --> 00:08:39,000 enough yet to run the entire problem on the quantum computer. 157 00:08:39,679 --> 00:08:42,679 Does that mean I wait for 20 years or 10 years or however 158 00:08:42,679 --> 00:08:45,720 long? From our perspective, no, 159 00:08:46,200 --> 00:08:49,400 because people didn't wait for classical for that either. 160 00:08:50,080 --> 00:08:55,280 And what you're seeing now is people are starting to run real 161 00:08:55,280 --> 00:08:58,960 world scale problems on quantum computers. 162 00:08:59,560 --> 00:09:03,360 So, for example, Cleveland Clinic published a paper where 163 00:09:03,360 --> 00:09:06,640 they basically show that they can simulate a protein of 164 00:09:06,720 --> 00:09:10,000 interest in life sciences with 303 atoms. 165 00:09:10,840 --> 00:09:14,680 They did that by breaking up the problem and taking the really 166 00:09:14,680 --> 00:09:17,760 hard parts and using a quantum computer to simulate the really 167 00:09:17,760 --> 00:09:21,880 hard parts and using classical to basically break up the 168 00:09:21,880 --> 00:09:23,480 problem and stitch it back together. 169 00:09:24,360 --> 00:09:28,720 So they can basically simulate that that protein, which is 170 00:09:28,720 --> 00:09:31,360 important for life sciences because the oxidation of that 171 00:09:31,360 --> 00:09:33,840 protein is actually one of the things that causes vaccines to 172 00:09:33,840 --> 00:09:36,920 lose their shelf life. So it's a real problem that 173 00:09:36,920 --> 00:09:39,840 they're really trying to simulate and they can simulate 174 00:09:39,840 --> 00:09:48,160 it on quantum computers today. So it's it's that work on how do 175 00:09:48,160 --> 00:09:51,520 I think about the problem? How do I apply quantum 176 00:09:51,520 --> 00:09:54,480 algorithms and how do I break up the problem to run it? 177 00:09:54,600 --> 00:09:58,840 And you can't do that without access to computers. 178 00:09:59,600 --> 00:10:01,160 It's kind of like the AI world, right? 179 00:10:01,160 --> 00:10:05,480 You can't do ChatGPT without access to GPUs today. 180 00:10:05,480 --> 00:10:08,520 I mean primarily like it's the same thing. 181 00:10:08,520 --> 00:10:11,640 You're not going to be able to develop the quantum applique, 182 00:10:11,640 --> 00:10:13,880 quantum enabled applications, quantum algorithms without 183 00:10:13,880 --> 00:10:16,160 access to the underlying technology. 184 00:10:16,280 --> 00:10:18,680 How complex of a protein is that? 185 00:10:18,680 --> 00:10:25,440 Cleveland study simulating so. So that one's 303 atoms is make 186 00:10:25,440 --> 00:10:27,680 sure I get the pronunciation right, but it's a tryptophan. 187 00:10:27,880 --> 00:10:30,480 303 atoms. I don't know how complex that 188 00:10:30,480 --> 00:10:32,760 is, so if you wanted to compare it to. 189 00:10:33,440 --> 00:10:39,160 It's, it's a, it's a, it's a relatively moderate size 190 00:10:39,320 --> 00:10:42,040 protein. But the exciting thing is that 191 00:10:42,040 --> 00:10:46,120 you can leverage the same approach to run 10,000 atoms. 192 00:10:46,720 --> 00:10:49,120 So, so now that people have figured out how to break up the 193 00:10:49,120 --> 00:10:52,000 problem and leverage on the computer, they can actually 194 00:10:52,000 --> 00:10:56,560 scale it significantly larger. And I highly anticipate that 195 00:10:56,560 --> 00:11:00,320 you're going to start seeing publications around that those 196 00:11:00,320 --> 00:11:04,120 larger scale very soon, like I mean in the matter of months. 197 00:11:04,640 --> 00:11:07,480 This is kind of like learning how the tool works. 198 00:11:07,720 --> 00:11:11,040 Or is it the computational power of what you're able to do at the 199 00:11:11,040 --> 00:11:14,440 moment that is preventing or keeping it at 300 before it gets 200 00:11:14,440 --> 00:11:17,720 to 10,000? So it's a little bit of a little 201 00:11:17,720 --> 00:11:20,360 bit of both, but it's primarily the first. 202 00:11:20,880 --> 00:11:25,760 So like I said before, people like we're thinking of this as 203 00:11:26,160 --> 00:11:28,560 how do I fit the whole problem into a quantum computer? 204 00:11:28,640 --> 00:11:32,080 This is like five years ago and algorithms to do that. 205 00:11:32,720 --> 00:11:35,120 People are now thinking about now that the computational tool 206 00:11:35,120 --> 00:11:37,320 is real, that the quantities quantum computers are real. 207 00:11:37,640 --> 00:11:40,240 People are thinking about how do I best leverage them and 208 00:11:40,240 --> 00:11:42,840 realizing that if I use similar approaches that I use 209 00:11:42,840 --> 00:11:47,160 classically, I can use these quantum computational tool more 210 00:11:47,160 --> 00:11:48,800 efficiently and more effectively. 211 00:11:49,800 --> 00:11:52,280 Like the Cleveland Clinic example, they were doing like 8 212 00:11:52,280 --> 00:11:55,080 atoms, 14 atoms as recently as nine months ago. 213 00:11:55,840 --> 00:12:00,120 And then they had this breakthrough in algorithmic 214 00:12:00,560 --> 00:12:04,880 discovery on a better, a rhythmic approach which improve 215 00:12:04,880 --> 00:12:07,720 the accuracy. And then they had a breakthrough 216 00:12:07,720 --> 00:12:10,160 in figuring out how to break up the problem. 217 00:12:10,160 --> 00:12:13,800 So how do I construct the problem differently which allow 218 00:12:13,800 --> 00:12:15,680 them to increase the size of the atoms? 219 00:12:15,680 --> 00:12:18,320 And there's no reason why they can't, now that they figure this 220 00:12:18,320 --> 00:12:22,560 out, go from 303 to 11,012 thousand. 221 00:12:22,920 --> 00:12:25,320 And now not just simulate this protein, but simulate that 222 00:12:25,320 --> 00:12:28,800 protein insolvent in a much more complex thing, which is what is 223 00:12:28,800 --> 00:12:32,600 needed to do the real world digital twin kind of stuff that 224 00:12:32,600 --> 00:12:34,240 they want to do in healthcare, life sciences. 225 00:12:34,600 --> 00:12:37,400 Let's talk more about Cleveland Clinic and, and what they did, 226 00:12:37,400 --> 00:12:40,640 because I think whenever someone talks about computers and 227 00:12:40,640 --> 00:12:44,600 proteins, they tend to gravitate right towards oh, alpha fold. 228 00:12:44,680 --> 00:12:47,240 But I think what is interesting about what Cleveland Clinic is 229 00:12:47,240 --> 00:12:49,720 doing, they're taking it beyond structure and they're actually 230 00:12:49,720 --> 00:12:54,000 actually talking about the, the model works on the electronic 231 00:12:54,000 --> 00:12:56,680 structure between the molecules working together. 232 00:12:56,680 --> 00:12:59,360 It's more complex than just, hey, here's the studs and the 233 00:12:59,360 --> 00:13:01,040 framing. It's actually how things 234 00:13:01,040 --> 00:13:02,880 communicate. Can you unpack that a little for 235 00:13:02,880 --> 00:13:03,800 us? Yeah. 236 00:13:03,800 --> 00:13:05,680 I mean, I might might be oversimplifying this a little 237 00:13:05,680 --> 00:13:08,240 bit, but you can think of they're using quantum simulation 238 00:13:08,240 --> 00:13:10,600 to kind of like do more of a like a digital twin ish kind of 239 00:13:10,600 --> 00:13:14,040 thing of like simulating simulating it. 240 00:13:14,520 --> 00:13:18,480 Whereas alpha photon AI is basically taking the known 241 00:13:18,480 --> 00:13:22,480 information that we've got and trying to interpolate and finds, 242 00:13:22,920 --> 00:13:25,480 you know, better ways to like use that data to make a better 243 00:13:25,480 --> 00:13:29,160 prediction of what might work. So in one case you're basically 244 00:13:29,160 --> 00:13:33,520 doing the actual simulation of the molecule and the other case 245 00:13:33,520 --> 00:13:39,120 you're trying to use known human data that we've like wet benched 246 00:13:39,120 --> 00:13:43,480 or whatever and basically try to predict what might work better 247 00:13:43,480 --> 00:13:45,480 based on that. So we're here to talk about 248 00:13:45,480 --> 00:13:47,040 some. It's this exciting announcement 249 00:13:47,040 --> 00:13:49,480 you guys have let it. So it's a three phase 250 00:13:49,560 --> 00:13:52,720 architecture. Can you help us go through this 251 00:13:52,720 --> 00:13:56,720 phase by phase and help us understand it from a high level? 252 00:13:57,040 --> 00:14:00,320 Basically, you can think about it as we're going to have 253 00:14:01,240 --> 00:14:04,520 quantum processing units for quantum computers that are going 254 00:14:04,520 --> 00:14:06,840 to run those quantum subroutines. 255 00:14:07,880 --> 00:14:11,480 And we're going to have classical resources that we know 256 00:14:11,480 --> 00:14:15,040 and love today, CPU and GPU that are going to be continue to be 257 00:14:15,040 --> 00:14:19,480 good at what they're good at. So how do you build an 258 00:14:19,480 --> 00:14:25,040 architecture that allows you to run the subroutines on the right 259 00:14:25,680 --> 00:14:31,120 underlying processing units in a way that's sufficient and that's 260 00:14:31,200 --> 00:14:34,960 fundamentally at its heart, you know what the architecture is 261 00:14:34,960 --> 00:14:38,920 about? The choice between what part of 262 00:14:39,000 --> 00:14:44,000 the problem is done by which by either the QPU or the GPU is 263 00:14:44,000 --> 00:14:45,640 that? Is that what you mean? 264 00:14:45,840 --> 00:14:48,000 Exactly. And there, there's levels here 265 00:14:48,040 --> 00:14:50,880 and This is why it makes it a little bit complex, but there, 266 00:14:50,880 --> 00:14:52,400 there's levels of the onion here. 267 00:14:53,080 --> 00:14:56,120 There are things that need to be done to create a large scale 268 00:14:56,120 --> 00:15:02,600 fault tolerant quantum system that need to run on the the 269 00:15:02,600 --> 00:15:06,760 quantum processing unit. And there's pieces of it need to 270 00:15:06,760 --> 00:15:11,080 be run on classical accelerators, whether the ASICS 271 00:15:11,080 --> 00:15:14,880 or GP us or CP US. So that's kind of like the inner 272 00:15:14,880 --> 00:15:17,640 part of the onion. But then there's a second layer 273 00:15:18,000 --> 00:15:23,160 of classical resources that need to run the the classical parts 274 00:15:23,160 --> 00:15:27,560 of the workflow. So you're very similar to CP us 275 00:15:27,560 --> 00:15:33,320 and GP US working together today or GPU heavy nodes and CPU. 276 00:15:33,480 --> 00:15:36,360 So you need some way to basically coordinate the work 277 00:15:36,360 --> 00:15:40,960 across those and you need some kind of architecture that kind 278 00:15:40,960 --> 00:15:43,720 of lays out how are these connected, what bandwidth 279 00:15:43,840 --> 00:15:46,360 requirements are there, what latency requirements on there, 280 00:15:46,920 --> 00:15:50,160 those kinds of things. So the fault tolerant scalable 281 00:15:50,200 --> 00:15:52,920 thing is kind of the down the road like, hey, we're trying to 282 00:15:53,040 --> 00:15:57,200 point that direction what's happening in the near term, not 283 00:15:57,200 --> 00:16:01,000 today in the next year or so that that is exciting and and 284 00:16:01,000 --> 00:16:03,120 about this architecture and we can even go in the technical 285 00:16:03,120 --> 00:16:06,000 weeds a little bit here, like what are you excited about in in 286 00:16:06,000 --> 00:16:07,360 that realm? Yeah. 287 00:16:07,360 --> 00:16:09,360 So what I'm excited about is like some of the work that we 288 00:16:09,360 --> 00:16:14,280 did with Rican, which is, you know, the scientific research 289 00:16:14,280 --> 00:16:20,480 institution in Japan that has the Japan's largest HPC cluster. 290 00:16:20,680 --> 00:16:23,840 And we actually have a quantum computer in the same building, 291 00:16:24,800 --> 00:16:28,280 which allows us to have, you know, connectivity, direct 292 00:16:28,280 --> 00:16:33,760 connectivity between the quantum system and the classical 293 00:16:33,760 --> 00:16:36,080 systems. You have to really explore this 294 00:16:36,080 --> 00:16:40,440 QCSC architecture. And what they're doing today is 295 00:16:40,440 --> 00:16:44,680 figuring out how do I orchestrate these two resources 296 00:16:44,680 --> 00:16:48,600 so they can work well together. And it's a little bit more 297 00:16:48,600 --> 00:16:51,800 challenging in the sense that, you know, a lot of the, like, 298 00:16:51,800 --> 00:16:54,320 orchestration today is you've got like a cheap resource and an 299 00:16:54,320 --> 00:16:57,120 expensive resource. And you're trying to like, you 300 00:16:57,120 --> 00:16:59,960 know, don't care too much about like managing the cheap stuff, 301 00:16:59,960 --> 00:17:03,160 but like, how do I optimize use of the expensive thing? 302 00:17:03,760 --> 00:17:06,240 In this case, in a quantum centric supercomputer, you've 303 00:17:06,240 --> 00:17:11,720 got your HPCAI resources and your quantum resources and 304 00:17:11,720 --> 00:17:14,440 they're both kind of expensive. So you basically want to come up 305 00:17:14,440 --> 00:17:18,040 with a way to orchestrate them so you're not wasting time on 306 00:17:18,040 --> 00:17:20,440 either side of it. And what we worked on with 307 00:17:20,440 --> 00:17:25,560 Weekend was the workflow in the orchestration for a chemistry 308 00:17:25,560 --> 00:17:29,760 experiment similar to what I described with Clinic of how do 309 00:17:29,760 --> 00:17:36,280 you run a chemistry workload that roughly runs about half the 310 00:17:36,280 --> 00:17:39,080 time on the classical resources, runs it half the time on the 311 00:17:39,080 --> 00:17:42,960 quantum resources in such a way that I'm not like letting either 312 00:17:42,960 --> 00:17:46,400 side of them just sit idle and I'm using fully using the 313 00:17:46,400 --> 00:17:48,560 computational on on both sides the. 314 00:17:48,960 --> 00:17:52,400 Phase one, phase two, phase 3, does that correlate in any way 315 00:17:52,400 --> 00:17:55,240 to the IBM chip? So you're talking about Starling 316 00:17:55,240 --> 00:17:58,240 in 2032, I think is on the road map. 317 00:17:58,240 --> 00:18:01,200 Do the phases relate to that? 2029. 318 00:18:01,360 --> 00:18:04,760 Thank you for correcting me. That's a Mystic. 319 00:18:05,240 --> 00:18:12,200 Yeah. So yes, because we're trying to 320 00:18:12,200 --> 00:18:19,520 intercept our road map for the quantum piece of the quantum 321 00:18:19,520 --> 00:18:23,680 centric supercomputer with the the classical piece of the 322 00:18:23,680 --> 00:18:25,600 quantum centric supercomputer, how they work together. 323 00:18:25,600 --> 00:18:30,840 So yes, the the Phase 3 is kind of aligned with with the the 324 00:18:30,840 --> 00:18:33,360 Starling 2029. Is that still on? 325 00:18:33,360 --> 00:18:37,080 Because 2029 is rolling up very quickly as I look at my. 326 00:18:37,080 --> 00:18:39,680 That's cool. That's why I basically, yeah. 327 00:18:40,080 --> 00:18:42,920 Yeah, yeah, yeah. You still on track for 20/29? 328 00:18:43,400 --> 00:18:47,520 We are, so we, we, we could for the reasons we mentioned before. 329 00:18:47,520 --> 00:18:49,920 Like you've got 2 camps, you've got a lot of hype. 330 00:18:50,200 --> 00:18:52,240 There's lots of different kind of like confusion hype, like I 331 00:18:52,240 --> 00:18:56,200 mentioned, the like, you know, apples to oranges confusion. 332 00:18:56,200 --> 00:18:58,480 There's the fault tolerant. What really fault tolerant is 333 00:18:58,480 --> 00:19:00,520 confusion. There's the, do you really have 334 00:19:00,520 --> 00:19:02,800 a quantum computer or you just have PowerPoint and like making 335 00:19:02,800 --> 00:19:05,760 it sound like you got a quantum computer hype. 336 00:19:06,000 --> 00:19:10,760 But we've put out basically a road map of what we're releasing 337 00:19:10,760 --> 00:19:12,560 to our clients year by year by year. 338 00:19:13,000 --> 00:19:17,360 And we put out a road map of key internal development milestones 339 00:19:17,360 --> 00:19:20,320 year by year by year. So you can like follow along at 340 00:19:20,320 --> 00:19:23,800 home of like, like, are we on track for 2029 right now? 341 00:19:23,840 --> 00:19:27,200 We're on track for 2029. Question I always get is like, 342 00:19:27,200 --> 00:19:30,800 what's your error bar? And you know what I would say 343 00:19:30,800 --> 00:19:33,080 the error bar is like, I don't think we're going to pull it in 344 00:19:33,080 --> 00:19:35,760 more than six months. Like I think there's like some 345 00:19:35,760 --> 00:19:39,040 really challenging engineering work necessary to put it all 346 00:19:39,040 --> 00:19:40,960 together. And I don't think we're going to 347 00:19:40,960 --> 00:19:43,280 miss by more than a year. And I don't think it's going to 348 00:19:43,280 --> 00:19:45,440 be fundamental. It's going to be like there's a 349 00:19:45,440 --> 00:19:46,680 lot of engineering work that we got to do. 350 00:19:46,920 --> 00:19:49,080 That's going to be all over the Internet, Scott. 351 00:19:49,080 --> 00:19:52,240 So my little thought, and then I'll leave it to you, Jamie. 352 00:19:52,280 --> 00:19:55,040 I was just thinking, we started the conversation speaking about 353 00:19:55,320 --> 00:19:59,480 solving a problem and then how that technology is adopted. 354 00:19:59,800 --> 00:20:03,960 And I'm thinking about chemistry and material science and 355 00:20:04,440 --> 00:20:07,280 molecular simulation and where that happens. 356 00:20:07,520 --> 00:20:10,040 So it happens in the universities, it happens in the 357 00:20:10,040 --> 00:20:12,920 big pharmaceutical companies, it happens in clinics. 358 00:20:13,280 --> 00:20:16,400 And then I try to. Connect that to the technology 359 00:20:16,400 --> 00:20:19,160 that you're building. And I, I, I think about learning 360 00:20:19,160 --> 00:20:21,960 curves and I think about removing what already in place 361 00:20:21,960 --> 00:20:24,880 and what's been working for these research institutes for so 362 00:20:24,880 --> 00:20:27,240 long. And I think about the scientists 363 00:20:27,240 --> 00:20:31,880 in the lab and how much say they have in what the, how the 364 00:20:31,880 --> 00:20:37,240 computations are run. And I and how do you change, 365 00:20:37,360 --> 00:20:44,000 update, evolve that system that seems very embedded culturally, 366 00:20:44,160 --> 00:20:46,240 or maybe I'm completely wrong on that because I'm not a 367 00:20:46,240 --> 00:20:49,680 scientist, but that's what I was thinking as you were speaking. 368 00:20:49,960 --> 00:20:51,520 I think there are two levels here, right? 369 00:20:51,520 --> 00:20:55,200 So there's the level of the people who are building the 370 00:20:55,200 --> 00:21:00,600 fundamental algorithm approaches and building the software assets 371 00:21:00,600 --> 00:21:06,520 or whatever you want to call them that instantiate that, that 372 00:21:06,520 --> 00:21:08,960 are repeatable. And then you got the people who 373 00:21:08,960 --> 00:21:11,840 are going to like leverage those models or leverage the software 374 00:21:11,840 --> 00:21:16,280 assets, right. So we're we're currently at the 375 00:21:16,280 --> 00:21:19,440 state where we're still in the algorithmic discovery and 376 00:21:19,440 --> 00:21:25,120 application research piece where you do need to basically still 377 00:21:25,120 --> 00:21:28,760 improve building those assets. So like in the Cleveland Clinic, 378 00:21:29,160 --> 00:21:32,120 you know, part of what they did was kind of use raw out, you 379 00:21:32,120 --> 00:21:35,840 know, raw concepts. We may have had some software 380 00:21:35,840 --> 00:21:37,680 assets that made it easier, blah, blah, blah, blah, blah. 381 00:21:37,920 --> 00:21:41,200 But it wasn't like a black box. They could just put their input, 382 00:21:41,200 --> 00:21:45,200 get the answer out and, and use, we're going to get there. 383 00:21:45,720 --> 00:21:48,960 Like we're going to get to the point probably in the next 3-4 384 00:21:48,960 --> 00:21:53,840 years where you're going to have chemistry solvers that are more 385 00:21:53,840 --> 00:21:57,840 like black boxes that like a wider set of computational 386 00:21:57,840 --> 00:22:01,040 chemists can basically just use as a tool as opposed to using 387 00:22:01,720 --> 00:22:04,960 your pick, pick a, pick a tool today, Julie, blah, blah, blah, 388 00:22:04,960 --> 00:22:07,200 blah, blah, you know, tool today that they use. 389 00:22:07,520 --> 00:22:11,440 So I think we'll get there. But in the short term, really 390 00:22:11,440 --> 00:22:16,480 the work is having, you know, the people who have domain 391 00:22:16,480 --> 00:22:21,040 expertise and math expertise, you know, building these 392 00:22:21,040 --> 00:22:24,120 algorithmic approaches for quantum, proving that they work, 393 00:22:24,640 --> 00:22:26,720 running them on real quantum computers, etcetera, etcetera. 394 00:22:28,680 --> 00:22:30,120 Scott, let's have some fun with this one. 395 00:22:30,120 --> 00:22:35,160 So let's imagine somehow Richard Feynman comes back to life and 396 00:22:35,160 --> 00:22:39,240 he comes across this paper and he reads the paper. 397 00:22:39,640 --> 00:22:41,800 Imagine what how he would react to. 398 00:22:41,800 --> 00:22:43,800 Let's just have some fun. There's no wrong answer here. 399 00:22:43,840 --> 00:22:46,880 Like how would he react? I think you, you probably so 400 00:22:46,960 --> 00:22:50,760 it's, it's not this one, maybe Les Feynman, but like, you know, 401 00:22:50,760 --> 00:22:54,080 they announced the Q4 bio winners today. 402 00:22:55,360 --> 00:22:57,160 They mentioned that Cleveland Clinic 1. 403 00:22:57,360 --> 00:23:01,040 So I think those I think he would be more excited about 404 00:23:01,040 --> 00:23:05,160 because like he had this postulation back in 1981 that 405 00:23:05,160 --> 00:23:09,360 like let's use quantum information science for, you 406 00:23:09,360 --> 00:23:13,120 know, computation. The fact that people are doing 407 00:23:13,120 --> 00:23:15,920 that for real world examples today. 408 00:23:15,920 --> 00:23:18,120 I think I think it would be a little mind blowing for him, 409 00:23:18,120 --> 00:23:21,080 honestly, even though he was the one who like postulated it, you 410 00:23:21,080 --> 00:23:23,280 know, you know, there's the half Mobius thing, there's the 411 00:23:23,280 --> 00:23:25,960 neutron sky. There's now a lot in the last 412 00:23:26,160 --> 00:23:29,080 this kind of my big point before in the last five months you've 413 00:23:29,080 --> 00:23:34,280 seen like a major uptake in like people simulating stuff on 414 00:23:34,280 --> 00:23:37,400 quantum computers that are real problems they want to solve. 415 00:23:38,200 --> 00:23:40,760 So, you know, we put out this paper on this half Mobius 416 00:23:40,760 --> 00:23:45,880 molecule thing that like the researchers at Zurich had this 417 00:23:46,080 --> 00:23:48,360 build, you know, build this molecule that doesn't exist in 418 00:23:48,840 --> 00:23:53,240 in, in, in nature with this, you know, crazy half Mobius 419 00:23:53,240 --> 00:23:56,080 property. We go around, it like goes 420 00:23:56,080 --> 00:23:59,560 halfway around like, you know, it crazy stuff. 421 00:23:59,560 --> 00:24:02,880 But they then simulated on a quantum computer to kind of 422 00:24:02,880 --> 00:24:06,600 prove that they had built it. You've got the neutron 423 00:24:06,600 --> 00:24:10,160 scattering experiments that like the DOE where they're use the 424 00:24:10,160 --> 00:24:13,880 quantum computer to simulate the experiment that they ran. 425 00:24:13,880 --> 00:24:16,360 You got the Cleveland Clinic thing simulating the protein. 426 00:24:16,360 --> 00:24:20,080 So I think this is the thing that's like Feynman's vision 427 00:24:20,080 --> 00:24:22,720 come to life. I mean, I didn't know him. 428 00:24:22,720 --> 00:24:25,040 Like, I don't know how excited he could be about like, you 429 00:24:25,040 --> 00:24:28,560 know, optimization for finance, but I'm just guessing he'd be 430 00:24:28,560 --> 00:24:33,680 more excited about the the like his concept that you can use 431 00:24:33,680 --> 00:24:37,800 this to really compute stuff that physicists and 432 00:24:37,800 --> 00:24:40,560 computational chemistry interested is like has come to 433 00:24:40,560 --> 00:24:43,480 fruition. Did Feynman write about this 434 00:24:43,560 --> 00:24:48,960 classical quantum unification, or was he very much of that? 435 00:24:48,960 --> 00:24:52,320 You'll have quantum computers and they'll be a separate idiom 436 00:24:52,320 --> 00:24:55,080 on their own. Or would he be disappointed by 437 00:24:55,080 --> 00:24:57,680 the reliance on the classical infrastructure? 438 00:25:01,040 --> 00:25:03,960 I doubt it. Like, I mean, you know, like I 439 00:25:03,960 --> 00:25:08,520 said, I didn't know him so complete speculation, but I 440 00:25:08,520 --> 00:25:11,320 think just the fact that you're you're leveraging quantum 441 00:25:11,320 --> 00:25:15,440 information science in order to simulate quantum mechanics on a 442 00:25:15,440 --> 00:25:19,240 real physics problem, I think would be pretty damn exciting to 443 00:25:19,240 --> 00:25:24,640 him because like that was like the big picture concept now. 444 00:25:25,440 --> 00:25:28,440 Let's can we transition to data Centers for a second Mark, Are 445 00:25:28,440 --> 00:25:32,120 you good with that? And could I before data centers? 446 00:25:32,280 --> 00:25:34,440 Just one more question then before we get on to data centers 447 00:25:34,440 --> 00:25:34,680 then. Yeah. 448 00:25:34,680 --> 00:25:37,680 Have you changed your mind about anything in the past year about 449 00:25:37,680 --> 00:25:40,200 quantum computers that you didn't think you would? 450 00:25:43,440 --> 00:25:45,200 I don't think I've changed my mind. 451 00:25:45,200 --> 00:25:52,040 I think. I think I've been surprised by 452 00:25:52,040 --> 00:25:56,520 how quickly some of these algorithmic approaches to do 453 00:25:56,520 --> 00:25:59,800 larger scale problems have really picked up. 454 00:26:00,560 --> 00:26:04,080 But I mean, I'm living at the daily, so I'm, yeah, I've been 455 00:26:04,080 --> 00:26:08,680 less surprised by other things. I am completely surprised by 456 00:26:08,680 --> 00:26:11,040 some of the hype and some of the how the hype is like 457 00:26:11,040 --> 00:26:12,680 interpreted. I shouldn't be. 458 00:26:12,920 --> 00:26:15,040 Just some. Sometimes you just shake your 459 00:26:15,040 --> 00:26:18,120 head at all of the hype and all of the nonsenses. 460 00:26:18,160 --> 00:26:20,480 Oh man. Probably can't comment on those, 461 00:26:20,480 --> 00:26:24,120 but. Yeah, Jeremy, data centers. 462 00:26:24,480 --> 00:26:26,040 What? What is this architecture going 463 00:26:26,040 --> 00:26:28,080 to do to the future of data centers is my question. 464 00:26:28,880 --> 00:26:30,720 Yeah. I mean the good news, bad news 465 00:26:30,720 --> 00:26:40,680 is that the rise of AI mega data centers have made it really easy 466 00:26:40,680 --> 00:26:43,840 for you to plop a quantum computer in any of those data 467 00:26:43,840 --> 00:26:47,640 centers. Our requirements are team by 468 00:26:47,640 --> 00:26:52,200 comparison actually I think you know people assume like quantum 469 00:26:52,200 --> 00:26:54,360 computing really complex etcetera, etcetera, etcetera 470 00:26:54,360 --> 00:26:57,680 that it must be like massively power hungry like all these 471 00:26:57,680 --> 00:27:01,120 requirements etcetera. So basically we require water 472 00:27:01,120 --> 00:27:05,400 cooling for our systems and not to get the heat out primarily 473 00:27:05,400 --> 00:27:09,600 but to keep the temperature steady across everything because 474 00:27:10,160 --> 00:27:13,760 variations in temperature leads to variation signal propagation 475 00:27:13,760 --> 00:27:16,400 which leads to timing kind of situations. 476 00:27:16,520 --> 00:27:22,240 So that's why we water cool. It requires very minor power 477 00:27:22,400 --> 00:27:27,280 compared to AI like so rule of thumb is a state-of-the-art 478 00:27:27,280 --> 00:27:31,760 quantum computer is about the same amount of power as one rack 479 00:27:31,760 --> 00:27:35,360 of AI. And that's true today. 480 00:27:35,360 --> 00:27:37,040 It's a little under actually today. 481 00:27:37,040 --> 00:27:40,680 And that's true for the system in 2029 also. 482 00:27:40,840 --> 00:27:44,200 On rack of H1 hundreds like how much is that draining off the? 483 00:27:44,440 --> 00:27:47,880 When you get into like 2029 time frame you know they're 484 00:27:47,880 --> 00:27:51,640 estimating well over a MW per per rack. 485 00:27:52,120 --> 00:27:53,560 OK. Let's pause right there. 486 00:27:53,560 --> 00:27:58,120 Just just for reference, Mark, probably 10 years ago, 5 487 00:27:58,120 --> 00:28:01,720 kilowatts of rack was like chunky was like pretty big, 488 00:28:01,720 --> 00:28:03,120 right? And then we've gotten the 20 489 00:28:03,120 --> 00:28:06,000 kilowatts and 50 kilowatts, 100 kilowatts. 490 00:28:06,000 --> 00:28:10,360 But you're talking about a a MW in a single rack, which used to 491 00:28:10,360 --> 00:28:12,080 be like the whole data center, right? 492 00:28:12,360 --> 00:28:14,280 Yeah, it's a little bit insane. Yeah. 493 00:28:14,280 --> 00:28:15,480 So about the. Water. 494 00:28:15,480 --> 00:28:16,800 What about the water consumption? 495 00:28:16,960 --> 00:28:19,760 If could we get on the cooling, could we get a figure on that? 496 00:28:19,920 --> 00:28:22,840 I don't know what the AI data centers, water consumption are 497 00:28:22,840 --> 00:28:24,880 these or the. Quantum yours if you put it in a 498 00:28:24,880 --> 00:28:27,000 day. Oh, ours is fairly tame. 499 00:28:27,000 --> 00:28:29,560 Like, you know, if you've got water cooled in your data 500 00:28:29,560 --> 00:28:34,720 centre, we're like a blip on the, you know, less than the. 501 00:28:34,720 --> 00:28:36,240 Water for your coffee pot, right? 502 00:28:36,240 --> 00:28:38,440 Yeah. And then the, the, the, the 503 00:28:38,440 --> 00:28:41,920 weight of the system again is the weight of these like AI 504 00:28:42,440 --> 00:28:46,080 racks is like intense. So, you know, if you, if you can 505 00:28:46,080 --> 00:28:49,320 solve that problem, you know, the weight for our systems is 506 00:28:49,320 --> 00:28:51,840 not an issue. So the only constraint that 507 00:28:51,840 --> 00:28:56,920 we've run into a couple times is our systems are higher again. 508 00:28:56,920 --> 00:28:59,400 And so these AI mega data centers, no problem whatsoever. 509 00:28:59,400 --> 00:29:03,320 But in like a traditional data centre and even a traditional 510 00:29:03,560 --> 00:29:07,400 like big data centre, but like a departmental data centers kind 511 00:29:07,400 --> 00:29:10,080 of thing, you know, our systems tend to be a little bit higher. 512 00:29:10,520 --> 00:29:11,840 What do you mean? But what do you mean by higher 513 00:29:11,840 --> 00:29:13,800 like? Is it physically high? 514 00:29:14,000 --> 00:29:16,760 Physically high, that's it. They're taller. 515 00:29:17,400 --> 00:29:20,240 So in most of our deployments it hasn't been an issue at all. 516 00:29:21,080 --> 00:29:23,680 But it's it's usually the one that is the question mark. 517 00:29:24,200 --> 00:29:25,640 The other one is that. Design choice? 518 00:29:25,640 --> 00:29:29,400 Or is that just necessity? It's a little bit of both. 519 00:29:29,400 --> 00:29:33,400 It's a lot easier to get the signals up on the top, up over 520 00:29:33,400 --> 00:29:39,920 the top because there's some advantages you know of Basically 521 00:29:41,280 --> 00:29:44,920 your cryogenics like hanging down as opposed to coming up. 522 00:29:47,040 --> 00:29:50,960 There's somewhere behind me. But if you have some show and 523 00:29:50,960 --> 00:29:53,080 tell Scott, we'd love to if you I don't think it's. 524 00:29:53,280 --> 00:29:54,640 Oh, yes, yes. So what are we? 525 00:29:54,680 --> 00:29:56,800 What are we looking? At that's kind of like an 526 00:29:56,800 --> 00:29:59,720 example of the thing that's hangs down inside what we call 527 00:30:00,120 --> 00:30:03,120 the fridge. So that's like where the cubits 528 00:30:03,120 --> 00:30:06,200 live. So that's where the cubits live, 529 00:30:06,200 --> 00:30:09,960 and they live there because like overly simplistically, they need 530 00:30:09,960 --> 00:30:11,760 to be isolated from the rest of the universe. 531 00:30:12,240 --> 00:30:15,160 So you know you do this quantum information. 532 00:30:15,160 --> 00:30:17,280 I love that sentence so much. I love this. 533 00:30:17,280 --> 00:30:19,600 One of my favorite sentence in quantum I think. 534 00:30:19,920 --> 00:30:23,000 Yeah, so like this is a sci-fi come to life part that's like, 535 00:30:23,040 --> 00:30:26,880 you know, for us it's like 1.5 millikel, 15 millikelvin. 536 00:30:28,080 --> 00:30:29,720 Yeah, so it's really, really freaking cold. 537 00:30:29,920 --> 00:30:33,240 But it's also like light isolation by racial isolation, 538 00:30:33,240 --> 00:30:36,240 etcetera, etcetera. Yeah, it's like a Bond movie. 539 00:30:36,240 --> 00:30:39,480 Like either you you're you're cryogenic at some point, like 540 00:30:39,480 --> 00:30:41,960 even the trapped ions are cryogenic at some temperature 541 00:30:42,640 --> 00:30:45,720 and then it's either just cooling it or you shoot friggin 542 00:30:45,720 --> 00:30:48,320 laser beams at it. It's like one of the two in 543 00:30:48,360 --> 00:30:50,840 order to get the entropy out. So that's kind of like the 544 00:30:50,840 --> 00:30:52,720 sci-fi come to life part of part of it. 545 00:30:53,320 --> 00:30:59,120 But anyway, so, so in order to basically get the signals most 546 00:30:59,120 --> 00:31:02,320 efficiently in and out, it's easier to do it over the top. 547 00:31:02,600 --> 00:31:04,920 If you're doing it over the top, it means that it's a little bit 548 00:31:04,920 --> 00:31:07,400 higher. So, so qubits, qubits are very 549 00:31:07,400 --> 00:31:09,760 distractible. They require they require some 550 00:31:09,760 --> 00:31:12,080 focusing right, which you were, which you're talking about the. 551 00:31:12,440 --> 00:31:15,640 Universe and the trick the. Trick is you want them to be 552 00:31:15,640 --> 00:31:18,280 distractible because if they're not distractible at all, you 553 00:31:18,280 --> 00:31:20,360 can't program them or they're really, really slow. 554 00:31:20,760 --> 00:31:23,720 So it's trying to find the right balance of how distractible do 555 00:31:23,720 --> 00:31:26,920 you want your qubits and like giving them the right medication 556 00:31:26,920 --> 00:31:29,120 so they don't, they don't lose focus, so. 557 00:31:29,800 --> 00:31:31,120 What a reference. Wow. 558 00:31:31,840 --> 00:31:33,920 All right. Well, so speaking, Speaking of 559 00:31:33,920 --> 00:31:39,120 being higher and having the having the computation happening 560 00:31:39,120 --> 00:31:42,600 higher, let's talk about orbital, orbital data centers 561 00:31:42,600 --> 00:31:45,520 and let's talk about quantum in space. 562 00:31:45,640 --> 00:31:48,960 And you know, I, I know you probably don't think a lot about 563 00:31:48,960 --> 00:31:51,160 that, but my job is to weave and connect the dots. 564 00:31:51,160 --> 00:31:53,920 On the show, we talked a lot about space tech, talk a lot 565 00:31:53,920 --> 00:31:56,320 about orbital data center, star cloud and others. 566 00:31:56,720 --> 00:31:59,440 Is there a place for quantum computing in space? 567 00:32:01,000 --> 00:32:06,360 I think it would be similar. And so like overly 568 00:32:06,360 --> 00:32:09,040 simplistically, I don't think it's for cooling or anything 569 00:32:09,040 --> 00:32:12,160 like that. So the real question is, you 570 00:32:12,160 --> 00:32:16,200 know, space, energy, etcetera, etcetera, etcetera. 571 00:32:16,880 --> 00:32:19,880 Honestly, it's not something we've looked at very carefully, 572 00:32:20,440 --> 00:32:22,520 Elon. Musk is is Elon Musk is on 573 00:32:22,520 --> 00:32:25,040 Twitter saying oh, they should put quantum computers on the 574 00:32:25,040 --> 00:32:28,480 South Pole of the moon in the in the dark shadows of the craters. 575 00:32:28,960 --> 00:32:33,000 Is he saying that just to to get likes or is he saying that 576 00:32:33,000 --> 00:32:35,440 because he believes that that's actually a good idea? 577 00:32:36,960 --> 00:32:38,680 I actually believe it's not a good idea. 578 00:32:38,680 --> 00:32:43,440 I don't know. I mean, in fairness, I wouldn't 579 00:32:43,440 --> 00:32:49,720 need to think through why his hypothesis for for why that is. 580 00:32:53,120 --> 00:32:56,680 You know, there's a bunch floating around of like helium 581 00:32:56,680 --> 00:32:58,360 sources, etcetera, etcetera, etcetera. 582 00:32:58,360 --> 00:33:01,840 But. Well, Helium 3, I'm sorry, I'd 583 00:33:01,840 --> 00:33:04,840 like to interrupt you a train, a train of thought, but is he, is 584 00:33:04,840 --> 00:33:08,000 Helium 3 needed in the quantum computing industry? 585 00:33:08,680 --> 00:33:14,520 It is for the if you're using cryogenics to isolate it down to 586 00:33:14,520 --> 00:33:19,560 the temperatures that we are, it is, you know, it's a real thing. 587 00:33:21,400 --> 00:33:23,280 It is something that we need to address. 588 00:33:24,000 --> 00:33:27,000 Do you have a shortage of it? It's not a short term problem. 589 00:33:27,000 --> 00:33:30,080 It's more of a if you believe this technology is going to 590 00:33:30,120 --> 00:33:33,760 scale and you've got lots and lots and lots of quantum 591 00:33:33,760 --> 00:33:37,120 computers by the middle of the next decade, then it is 592 00:33:37,120 --> 00:33:38,600 something that we need to consider. 593 00:33:40,200 --> 00:33:42,280 Excellent. The question is, yes, it's not a 594 00:33:42,280 --> 00:33:45,080 problem today, but it is something that we need to think 595 00:33:45,160 --> 00:33:46,800 about. Let's try to, let's try to land 596 00:33:46,800 --> 00:33:50,200 a plane a little bit. We, we talk about not just 597 00:33:50,200 --> 00:33:53,440 technology, but what what technology means for humans, 598 00:33:53,440 --> 00:33:56,680 human solving problems, humans doing work, that sort of thing. 599 00:33:56,680 --> 00:34:01,760 We do have one question that Kevin Kelly left us to ask all 600 00:34:01,760 --> 00:34:04,360 of our guests. And I want to I want to ask you 601 00:34:04,360 --> 00:34:07,440 this question. Little bit, little bit social 602 00:34:07,440 --> 00:34:08,679 question, little bit tech question. 603 00:34:08,679 --> 00:34:11,480 But what do we want humans to be? 604 00:34:11,800 --> 00:34:16,040 And how does technology like quantum computing potentially 605 00:34:16,040 --> 00:34:23,199 help us get there? I would say we want humans to be 606 00:34:23,199 --> 00:34:29,800 happy and productive. And I think it's, again, for me, 607 00:34:29,800 --> 00:34:32,360 quantum computing is just a new computational tool when you 608 00:34:32,360 --> 00:34:34,239 really get it at the heart of it. 609 00:34:34,679 --> 00:34:38,199 So the question is like, what kinds of applications are we 610 00:34:38,199 --> 00:34:40,800 going to be able to do better with quantum computers? 611 00:34:40,800 --> 00:34:44,400 So you know, from my side, I think there's a lot of societal 612 00:34:44,400 --> 00:34:47,760 benefit in doing a better job of simulating materials. 613 00:34:48,280 --> 00:34:50,800 Well, that's chemistry for life, you know, chemistry for 614 00:34:50,800 --> 00:34:53,280 fertilizer, et cetera, et cetera, et cetera, et cetera. 615 00:34:55,199 --> 00:34:59,400 And then there's the like doesn't sound so great for 616 00:34:59,400 --> 00:35:02,400 societal good. It's not, you know, I'm not sure 617 00:35:02,400 --> 00:35:05,200 it would be on the UN like list of things that they want to use 618 00:35:05,200 --> 00:35:07,720 a quantum computer for. But there are a lot of things in 619 00:35:07,720 --> 00:35:13,200 optimization that can save money or make more money for, you 620 00:35:13,200 --> 00:35:19,320 know, a lot of industry where if they could optimize better, they 621 00:35:19,400 --> 00:35:25,320 have happy customers, higher return portfolios, lower risk, 622 00:35:25,840 --> 00:35:31,160 all those kinds of things. So I think it's the combination 623 00:35:31,160 --> 00:35:36,280 of those kind of things like, but yeah, I mean, I, I, I think 624 00:35:36,280 --> 00:35:41,760 it's more in the line of just the, the more meta question of 625 00:35:42,040 --> 00:35:47,040 how can we continue to leverage computation to make humans 626 00:35:47,320 --> 00:35:55,200 happier and productive with all the like ethical thinking about 627 00:35:56,000 --> 00:35:59,400 how to not use computers for the opposite. 628 00:36:00,200 --> 00:36:02,200 Do you have any other show and tell quantum? 629 00:36:02,200 --> 00:36:04,440 Show and tell some anything that you could hold physics in your 630 00:36:04,440 --> 00:36:07,400 hand to show us? That would be stunning because 631 00:36:07,400 --> 00:36:10,360 that always is well appreciated. It's going to come in here, but 632 00:36:11,960 --> 00:36:18,320 so state-of-the-art quantum computer circle 282018 So this 633 00:36:18,320 --> 00:36:21,840 is system one. This is like the first one that 634 00:36:21,840 --> 00:36:25,800 we did from a design point of view to kind of show it like all 635 00:36:25,800 --> 00:36:30,400 in one effective, really, really pretty box to show that 636 00:36:30,400 --> 00:36:34,240 basically this is a computer. It's not a lab experiment and 637 00:36:34,280 --> 00:36:37,800 the early days like we've got like archival pictures of like, 638 00:36:38,600 --> 00:36:43,920 you know, we form generator equipment and like wires flying 639 00:36:43,920 --> 00:36:46,760 all over the place and looks like a lab experiment. 640 00:36:48,960 --> 00:36:52,600 This block, which was like at the bottom of that, that quantum 641 00:36:52,600 --> 00:36:57,920 computer back in the day, you know, there has turned into, you 642 00:36:57,920 --> 00:37:01,880 know, the thing that I'm showing back back here, which has like 643 00:37:02,440 --> 00:37:06,680 flex cables, you know, running, you know, hundreds of lines in 644 00:37:06,680 --> 00:37:09,000 and out of this thing in order to program it. 645 00:37:10,280 --> 00:37:14,040 Yeah, I've got the flex cable. And the and the Heron we. 646 00:37:14,280 --> 00:37:16,160 We, we multi purpose here at IBM. 647 00:37:16,200 --> 00:37:19,200 So this is actually the flex cable that we're talking about 648 00:37:19,560 --> 00:37:22,160 like a prototype of it. You're one of the things that we 649 00:37:22,160 --> 00:37:25,320 need to do is get more microwave signals in and out of the out of 650 00:37:25,320 --> 00:37:28,760 the cryostat. So and also for manufacturing 651 00:37:28,760 --> 00:37:33,040 reasons, you can't have people like 1 by 1 plugging in wires 652 00:37:33,040 --> 00:37:36,800 like this anymore when you got like thousands or 10 thousands 653 00:37:36,800 --> 00:37:38,360 of lines coming out of your computer. 654 00:37:38,840 --> 00:37:43,480 So we need to come up with ways that we can like significantly 655 00:37:43,480 --> 00:37:48,920 increase the the the amount of connections coming in at and 656 00:37:48,920 --> 00:37:51,240 being able to just like click it into the computer from a 657 00:37:51,240 --> 00:37:54,280 manufacturing point of view as opposed to like hand connect it. 658 00:37:54,440 --> 00:37:56,600 That's like a ribbon cable, right, That we were looking at 659 00:37:57,080 --> 00:37:58,960 multiple connections. So what if, what if? 660 00:37:58,960 --> 00:38:02,240 So you mentioned Mobius before. What happens if you if you had 661 00:38:02,320 --> 00:38:06,680 half turn full turn the ribbon? If it doesn't break, I think 662 00:38:06,680 --> 00:38:09,040 you're OK. Yeah, I mean, the trick on these 663 00:38:09,040 --> 00:38:12,360 things is that it looks pretty standard. 664 00:38:12,360 --> 00:38:16,360 But yes, but it needs to be super conducting and obviously 665 00:38:16,360 --> 00:38:19,920 it needs to like not break as you're cooling it down to really 666 00:38:19,920 --> 00:38:21,800 low temperatures and heating it up occasionally. 667 00:38:23,120 --> 00:38:27,680 So it it has more demands than than like, you know, the typical 668 00:38:28,440 --> 00:38:32,920 like Fox cables. So it's like example of kind of 669 00:38:32,920 --> 00:38:35,280 like an example I was giving of like, you know, where do I put 670 00:38:35,280 --> 00:38:39,600 the risk on 20/29? Yeah, it's basically lots and 671 00:38:39,600 --> 00:38:42,120 lots of piece parts very similar to that. 672 00:38:42,480 --> 00:38:44,720 It's not just one thing. It's like this entire 673 00:38:44,720 --> 00:38:49,320 engineering stack that we need to not only get the right 674 00:38:49,320 --> 00:38:52,480 quality on, but we need to get the right quality on at scale. 675 00:38:53,000 --> 00:38:55,640 And we need to do all of that like together. 676 00:38:56,040 --> 00:38:58,200 And that's why it's taking us several years, even though we've 677 00:38:58,200 --> 00:39:01,760 demonstrated like all the piece parts on our road map, we need 678 00:39:01,760 --> 00:39:06,600 to put it put it all together. Yeah. 679 00:39:06,680 --> 00:39:11,560 So this is kind of like a picture of what the 2029 system, 680 00:39:12,120 --> 00:39:13,640 you know, roughly is going to look like. 681 00:39:13,960 --> 00:39:16,800 So it looks, you know, kind of similar to what what our system 682 00:39:16,800 --> 00:39:19,040 is today. But you know, this is not a 683 00:39:19,040 --> 00:39:21,800 football field anymore. Like if you talk to people like 684 00:39:21,920 --> 00:39:25,760 5-10 years ago in quantum because of the overheads and 685 00:39:25,760 --> 00:39:28,680 error correction at the time, they, you know, people were 686 00:39:28,680 --> 00:39:31,160 talking like, you know, other vendors were talking about like 687 00:39:31,160 --> 00:39:34,240 having to build like this gigantic football field size, 688 00:39:34,920 --> 00:39:38,960 like quant computer of like, you know, we don't need to do that 689 00:39:38,960 --> 00:39:41,440 anymore. Then that's why we are so 690 00:39:41,440 --> 00:39:46,440 confident in the 2029 date because we believe we've come up 691 00:39:46,440 --> 00:39:50,720 with the architecture and the all the peace parts to, to build 692 00:39:50,720 --> 00:39:53,440 something that is of reasonable scale for us to deliver. 693 00:39:53,800 --> 00:39:59,840 It's the idea of this architecture to remove NVIDIA 694 00:40:00,240 --> 00:40:03,800 the reliance on NVIDIA or is it to to build the relationship 695 00:40:03,800 --> 00:40:07,680 between IBM and NVIDIA? What role do or will the NVIDIA 696 00:40:08,720 --> 00:40:12,760 QPU while they play? NVIDIA GPUs are a really 697 00:40:12,760 --> 00:40:17,360 important part of the larger computing, you know thing, and 698 00:40:17,480 --> 00:40:21,680 GPUs will be a critical part of quantum centric supercomputing. 699 00:40:22,280 --> 00:40:26,680 You know it. It is the right platform to run 700 00:40:26,680 --> 00:40:29,560 certain kinds of work. Those certain kinds of work are 701 00:40:29,560 --> 00:40:32,640 definitely part of building a fault on quant computer, is part 702 00:40:32,640 --> 00:40:35,400 of building a quantocentric supercomputer and it's part of 703 00:40:35,400 --> 00:40:38,640 like AI mega farms, right? And all those are going to exist 704 00:40:38,640 --> 00:40:42,520 in 2029. This is not about like like 705 00:40:43,560 --> 00:40:45,640 competition, etcetera, etcetera, etcetera. 706 00:40:46,480 --> 00:40:51,320 The one thing I will say though is that we firmly believe that 707 00:40:51,320 --> 00:40:56,640 in order to drive adoption, it needs to be based on truly open 708 00:40:56,640 --> 00:41:02,920 source software that isn't tied to anyone like back end, whether 709 00:41:02,920 --> 00:41:06,360 it be the GPU back end, the CPU back end or the quantum back 710 00:41:06,360 --> 00:41:12,520 end. So that's why we would say that 711 00:41:12,720 --> 00:41:17,480 the layer for, you know, QCSC from an orchestration point of 712 00:41:17,480 --> 00:41:20,680 view or from a like how you program the quantum elements of 713 00:41:20,680 --> 00:41:25,640 it needs to be open and needs to be in our perspective, cloud 714 00:41:25,640 --> 00:41:29,440 native. Because a lot of the, the usage 715 00:41:29,440 --> 00:41:32,880 accessibility, to your point, making it accessible means right 716 00:41:32,880 --> 00:41:36,440 now making it cloud accessible. It's just a lot easier to get 717 00:41:36,440 --> 00:41:39,480 board adoption. If it's cloud accessible, from 718 00:41:39,480 --> 00:41:43,800 our perspective, it needs to be open, you know, kiss, KISS. 719 00:41:43,800 --> 00:41:45,560 I still think this is a true statement. 720 00:41:45,560 --> 00:41:49,040 The only software development kit you can run on Google, 721 00:41:49,400 --> 00:41:53,440 Microsoft and AWS, obviously IBM also. 722 00:41:53,640 --> 00:41:57,160 We fundamentally believe that that's the case both in how you 723 00:41:57,160 --> 00:41:59,360 connect down and how you connect up. 724 00:42:00,200 --> 00:42:03,760 Connecting into a video is absolutely part of what's going 725 00:42:03,760 --> 00:42:07,640 to have to be the case. We just want to make sure that 726 00:42:08,040 --> 00:42:12,320 quantum computing and how quantum computing links into all 727 00:42:12,320 --> 00:42:15,920 the other forms of computing is done in a way that people have 728 00:42:15,920 --> 00:42:18,960 flexibility in what hardware they choose for the different 729 00:42:18,960 --> 00:42:20,840 elements. Did we miss anything really 730 00:42:20,840 --> 00:42:22,800 fundamentally important? Yeah. 731 00:42:22,800 --> 00:42:26,880 I mean, I think the question you hit on on, you know, the 732 00:42:28,000 --> 00:42:33,160 accessibility and the algorithm development like threads are 733 00:42:33,160 --> 00:42:35,440 really, really, really, really important. 734 00:42:35,720 --> 00:42:41,840 Like I think that's there's, we obviously believe we're in the 735 00:42:41,840 --> 00:42:45,440 league here. Like we think that we're ahead 736 00:42:45,440 --> 00:42:47,480 of everybody else. We think we've right there 737 00:42:47,560 --> 00:42:49,560 technology choices, obviously we're biased, right? 738 00:42:50,640 --> 00:42:53,200 And we have confidence we're going to hit 2029. 739 00:42:54,880 --> 00:42:57,720 But there are enough other players with enough other money 740 00:42:57,720 --> 00:43:01,760 behind it that, you know, even if we're off, I don't think 741 00:43:01,760 --> 00:43:04,360 they're going to be 10 years behind us. 742 00:43:04,560 --> 00:43:08,080 So we do believe that there's enough money in this field that 743 00:43:08,080 --> 00:43:11,560 quantum computing and fault on chronic be is going to be a 744 00:43:11,560 --> 00:43:13,160 reality. It's just a question of who and 745 00:43:13,160 --> 00:43:20,160 when, not if anymore. And I think the big thing is 746 00:43:20,160 --> 00:43:25,000 going to be how much algorithmic discovery, how much application 747 00:43:25,000 --> 00:43:29,880 research, how much focus on how do you use this tool is really 748 00:43:29,880 --> 00:43:34,920 going to be the bigger unknown in my personal opinion, which is 749 00:43:34,920 --> 00:43:40,120 why we're so like passionate about, you know, getting the 750 00:43:40,120 --> 00:43:41,920 technology out there, getting people to use it, etcetera, 751 00:43:41,920 --> 00:43:45,120 etcetera. So there's been like a big tick 752 00:43:45,120 --> 00:43:50,000 up, like I mentioned in the last three years, but the curve is 753 00:43:50,000 --> 00:43:52,840 actually even increasing. But there's still a long way to 754 00:43:52,840 --> 00:43:54,840 go. Like the number of people 755 00:43:54,840 --> 00:43:58,400 thinking about quantum algorithm is still small compared to the 756 00:43:58,400 --> 00:44:00,280 number of people thinking about classical algorithms. 757 00:44:00,440 --> 00:44:04,280 Scott Crowder, Vice President of IBM Quantum Adoption. 758 00:44:04,280 --> 00:44:06,960 Thank you for thinking on paper with us today. 759 00:44:07,960 --> 00:44:10,960 If you enjoyed the show, please subscribe while you're listening 760 00:44:10,960 --> 00:44:14,720 to it and share with 1 quantum curious friend, 1 quantum 761 00:44:14,720 --> 00:44:17,000 curious person in your life who needs to hear this. 762 00:44:17,680 --> 00:44:19,920 And until next week, stay disruptive. 763 00:44:20,160 --> 00:44:22,520 Be curious. Keep thinking on paper.