1 00:00:00,450 --> 00:00:01,350 Welcome to Log 2 00:00:01,350 --> 00:00:02,483 Level Debug, the 3 00:00:02,483 --> 00:00:03,216 technology news 4 00:00:03,216 --> 00:00:04,349 show that I can't 5 00:00:04,349 --> 00:00:04,950 make too 6 00:00:04,950 --> 00:00:06,033 weird because 7 00:00:06,033 --> 00:00:06,150 my 8 00:00:06,150 --> 00:00:07,083 coworkers watch. 9 00:00:07,683 --> 00:00:08,033 And I'm your 10 00:00:08,033 --> 00:00:09,566 host, only shown 11 00:00:09,566 --> 00:00:09,816 from the 12 00:00:09,816 --> 00:00:10,433 waist up. 13 00:00:11,966 --> 00:00:12,416 And before we 14 00:00:12,416 --> 00:00:13,266 talk tech news, 15 00:00:13,283 --> 00:00:14,099 we have a brief 16 00:00:14,133 --> 00:00:14,766 announcement. 17 00:00:15,516 --> 00:00:16,949 Due to a ton of 18 00:00:16,949 --> 00:00:17,733 upcoming travel 19 00:00:17,733 --> 00:00:18,333 for various 20 00:00:18,333 --> 00:00:18,866 members of the 21 00:00:18,866 --> 00:00:20,133 crew, this is our 22 00:00:20,133 --> 00:00:21,033 last planned 23 00:00:21,033 --> 00:00:21,683 episode of the 24 00:00:21,683 --> 00:00:22,983 year, so let's 25 00:00:22,983 --> 00:00:23,316 call it a 26 00:00:23,316 --> 00:00:24,116 season finale. 27 00:00:24,899 --> 00:00:25,883 This also means, 28 00:00:25,916 --> 00:00:26,466 if you know where 29 00:00:26,466 --> 00:00:27,666 I live, it's the 30 00:00:27,666 --> 00:00:28,433 perfect time to 31 00:00:28,433 --> 00:00:29,516 rob my house. 32 00:00:29,633 --> 00:00:30,716 But please, lock 33 00:00:30,716 --> 00:00:31,116 up after 34 00:00:31,116 --> 00:00:31,500 you leave. 35 00:00:32,883 --> 00:00:33,366 Depending on 36 00:00:33,366 --> 00:00:34,683 time, we may have 37 00:00:34,683 --> 00:00:35,666 an episode in mid 38 00:00:35,666 --> 00:00:36,516 or late December, 39 00:00:36,666 --> 00:00:36,933 but that 40 00:00:36,933 --> 00:00:37,416 depends on 41 00:00:37,433 --> 00:00:38,316 other factors we 42 00:00:38,316 --> 00:00:39,033 don't control, 43 00:00:39,483 --> 00:00:39,716 like how 44 00:00:39,716 --> 00:00:40,500 lazy we feel. 45 00:00:41,633 --> 00:00:42,266 But don't worry, 46 00:00:42,299 --> 00:00:43,049 we'll be back at 47 00:00:43,049 --> 00:00:43,649 the start of 48 00:00:43,649 --> 00:00:45,266 2027, and the 49 00:00:45,266 --> 00:00:46,166 show will be 50 00:00:46,166 --> 00:00:46,883 better than 51 00:00:46,916 --> 00:00:47,250 ever. 52 00:00:47,583 --> 00:00:48,149 We'll have 53 00:00:48,266 --> 00:00:48,833 on-screen 54 00:00:48,833 --> 00:00:50,066 animations, a new 55 00:00:50,066 --> 00:00:51,116 anchor, a new 56 00:00:51,116 --> 00:00:52,283 background, and 57 00:00:52,283 --> 00:00:53,616 even guests. 58 00:00:54,899 --> 00:00:55,266 Sorry, one 59 00:00:55,266 --> 00:00:56,116 second, what? 60 00:00:57,466 --> 00:00:58,283 We still don't 61 00:00:58,283 --> 00:00:58,766 have a budget. 62 00:00:59,149 --> 00:00:59,716 Nothing. 63 00:01:00,500 --> 00:01:00,983 So, all those 64 00:01:00,983 --> 00:01:01,466 things I just 65 00:01:01,466 --> 00:01:03,533 said, none of 66 00:01:03,533 --> 00:01:04,250 them, not a 67 00:01:04,250 --> 00:01:04,633 single one. 68 00:01:07,099 --> 00:01:08,083 Well, we'll be 69 00:01:08,083 --> 00:01:09,033 back in 2027. 70 00:01:10,733 --> 00:01:11,683 Anyway, if you 71 00:01:11,683 --> 00:01:12,316 think you'll miss 72 00:01:12,316 --> 00:01:13,216 Log Level Debug 73 00:01:13,250 --> 00:01:14,000 over the next two 74 00:01:14,000 --> 00:01:14,416 months, 75 00:01:14,483 --> 00:01:15,166 don't worry. 76 00:01:15,383 --> 00:01:15,950 You can sign up 77 00:01:15,950 --> 00:01:16,133 for the 78 00:01:16,133 --> 00:01:16,916 newsletter at 79 00:01:16,966 --> 00:01:17,983 globaglob.dev 80 00:01:18,049 --> 00:01:19,133 slash newsletter. 81 00:01:19,549 --> 00:01:20,000 That'll keep 82 00:01:20,000 --> 00:01:20,816 going out every 83 00:01:20,816 --> 00:01:21,316 Wednesday. 84 00:01:22,183 --> 00:01:22,883 Now let's talk 85 00:01:22,883 --> 00:01:23,783 tech news. 86 00:01:25,516 --> 00:01:26,483 Microsoft had 87 00:01:26,483 --> 00:01:27,049 their quarterly 88 00:01:27,083 --> 00:01:27,683 earnings call 89 00:01:27,683 --> 00:01:28,216 this week. 90 00:01:28,583 --> 00:01:30,183 They made $33.6 91 00:01:30,383 --> 00:01:30,883 billion 92 00:01:31,366 --> 00:01:32,033 in profit. 93 00:01:32,416 --> 00:01:33,683 Billion with a B. 94 00:01:34,549 --> 00:01:35,299 In profit. 95 00:01:35,816 --> 00:01:37,966 Which is so sad. 96 00:01:38,416 --> 00:01:39,116 What was once the 97 00:01:39,116 --> 00:01:39,983 giant in the tech 98 00:01:39,983 --> 00:01:41,033 industry is only 99 00:01:41,033 --> 00:01:42,233 making $33.6 100 00:01:42,266 --> 00:01:42,883 billion in a 101 00:01:42,916 --> 00:01:44,383 quarter, so 102 00:01:44,383 --> 00:01:44,983 now they have to 103 00:01:44,983 --> 00:01:45,866 lay people off. 104 00:01:46,016 --> 00:01:47,533 But that billion 105 00:01:47,533 --> 00:01:48,533 was with a B, so 106 00:01:48,533 --> 00:01:49,433 they're asking 107 00:01:49,433 --> 00:01:50,233 people to lay 108 00:01:50,233 --> 00:01:50,983 themselves off. 109 00:01:51,933 --> 00:01:52,533 Microsoft is 110 00:01:52,533 --> 00:01:53,516 offering buyouts 111 00:01:53,516 --> 00:01:54,500 to 60,000 112 00:01:54,516 --> 00:01:55,716 employees, which 113 00:01:55,716 --> 00:01:56,933 is just under 25% 114 00:01:57,416 --> 00:01:57,983 of its total 115 00:01:57,983 --> 00:01:58,716 workforce. 116 00:01:59,366 --> 00:02:00,266 Anyone who elects 117 00:02:00,266 --> 00:02:01,266 to take the offer 118 00:02:01,566 --> 00:02:02,366 receives a pay 119 00:02:02,366 --> 00:02:03,416 package similar 120 00:02:03,416 --> 00:02:04,233 to Severance. 121 00:02:04,866 --> 00:02:05,549 Severance being 122 00:02:05,549 --> 00:02:06,299 the pay package, 123 00:02:06,366 --> 00:02:06,866 not the 124 00:02:07,133 --> 00:02:07,950 incredibly boring 125 00:02:07,950 --> 00:02:08,516 TV show. 126 00:02:09,483 --> 00:02:10,349 Honestly, you'd 127 00:02:10,349 --> 00:02:11,250 be stupid not to 128 00:02:11,250 --> 00:02:12,033 take the offer. 129 00:02:12,266 --> 00:02:13,316 Continued pay and 130 00:02:13,316 --> 00:02:14,883 stock vesting is 131 00:02:14,883 --> 00:02:16,416 just free money. 132 00:02:16,949 --> 00:02:17,633 And if you really 133 00:02:17,633 --> 00:02:18,216 like your job 134 00:02:18,216 --> 00:02:18,816 with the company, 135 00:02:19,166 --> 00:02:20,033 rest up for a bit 136 00:02:20,099 --> 00:02:20,633 and then 137 00:02:20,633 --> 00:02:21,666 reapply for the 138 00:02:21,683 --> 00:02:21,966 opening. 139 00:02:22,383 --> 00:02:22,683 Not as 140 00:02:22,683 --> 00:02:23,433 yourself though. 141 00:02:23,466 --> 00:02:24,449 Get a fake scar, 142 00:02:24,449 --> 00:02:24,783 maybe an 143 00:02:24,783 --> 00:02:25,383 eye patch. 144 00:02:25,766 --> 00:02:26,366 The cooler 145 00:02:26,366 --> 00:02:27,066 version of you. 146 00:02:27,933 --> 00:02:28,983 That job is 147 00:02:28,983 --> 00:02:29,466 practically 148 00:02:29,466 --> 00:02:30,033 yours. 149 00:02:31,050 --> 00:02:31,416 This is the 150 00:02:31,416 --> 00:02:32,166 second time 151 00:02:32,166 --> 00:02:32,766 Microsoft is 152 00:02:32,766 --> 00:02:33,566 making this type 153 00:02:33,566 --> 00:02:34,083 of offer to 154 00:02:34,083 --> 00:02:34,533 employees 155 00:02:34,533 --> 00:02:35,033 this year. 156 00:02:35,550 --> 00:02:36,483 Last time, about 157 00:02:36,483 --> 00:02:37,500 30% of people 158 00:02:37,500 --> 00:02:37,916 accepted 159 00:02:37,916 --> 00:02:38,366 the buyout. 160 00:02:38,933 --> 00:02:39,750 So Microsoft 161 00:02:39,750 --> 00:02:40,316 still laid 162 00:02:40,316 --> 00:02:40,866 people off. 163 00:02:41,466 --> 00:02:42,566 Based on past 164 00:02:42,566 --> 00:02:43,949 evidence, if all 165 00:02:43,949 --> 00:02:44,966 60,000 people 166 00:02:44,966 --> 00:02:45,866 accept the offer, 167 00:02:46,199 --> 00:02:47,183 Microsoft will 168 00:02:47,433 --> 00:02:48,033 still lay 169 00:02:48,033 --> 00:02:48,416 people off. 170 00:02:53,366 --> 00:02:54,216 Proton, the 171 00:02:54,233 --> 00:02:55,166 privacy-focused 172 00:02:55,166 --> 00:02:56,033 software company, 173 00:02:56,766 --> 00:02:57,816 kinda sucks. 174 00:02:58,533 --> 00:02:59,183 Reporting from 175 00:02:59,183 --> 00:03:00,483 404 Media, based 176 00:03:00,483 --> 00:03:01,316 on interviews and 177 00:03:01,349 --> 00:03:01,716 internal 178 00:03:01,716 --> 00:03:03,000 documents, show 179 00:03:03,000 --> 00:03:03,983 that Andy Yen, 180 00:03:04,016 --> 00:03:04,883 the CEO of 181 00:03:04,883 --> 00:03:06,300 Proton, has added 182 00:03:06,300 --> 00:03:07,233 toxicity to the 183 00:03:07,233 --> 00:03:08,099 company's culture 184 00:03:08,583 --> 00:03:09,333 and pretty much 185 00:03:09,333 --> 00:03:09,900 just gone 186 00:03:09,900 --> 00:03:10,800 full cray-cray. 187 00:03:11,433 --> 00:03:12,366 Which I also 188 00:03:12,366 --> 00:03:13,233 describe as 189 00:03:13,433 --> 00:03:14,433 standard Silicon 190 00:03:14,433 --> 00:03:15,383 Valley CEO. 191 00:03:16,716 --> 00:03:17,166 When someone 192 00:03:17,183 --> 00:03:18,449 urged a team to 193 00:03:18,449 --> 00:03:19,433 do extra testing 194 00:03:19,433 --> 00:03:19,916 on a certain 195 00:03:19,916 --> 00:03:20,583 feature before 196 00:03:20,583 --> 00:03:21,000 pushing to 197 00:03:21,000 --> 00:03:21,566 production, 198 00:03:21,983 --> 00:03:23,016 Yen called that 199 00:03:23,966 --> 00:03:24,933 "loser thinking." 200 00:03:26,016 --> 00:03:26,883 And maybe testing 201 00:03:26,933 --> 00:03:27,716 is a loser 202 00:03:27,716 --> 00:03:28,683 mindset, I 203 00:03:28,733 --> 00:03:29,366 certainly won't 204 00:03:29,366 --> 00:03:30,216 be arguing with 205 00:03:30,216 --> 00:03:30,716 that point 206 00:03:30,716 --> 00:03:31,199 because, 207 00:03:31,833 --> 00:03:32,400 well, I never 208 00:03:32,400 --> 00:03:33,083 test my code. 209 00:03:33,666 --> 00:03:34,199 But I'm a loser 210 00:03:34,199 --> 00:03:34,683 for other 211 00:03:34,683 --> 00:03:35,283 reasons. 212 00:03:35,516 --> 00:03:36,116 At least that's 213 00:03:36,116 --> 00:03:36,683 what my psychic 214 00:03:36,683 --> 00:03:37,050 tells me. 215 00:03:38,366 --> 00:03:39,516 Yen also wants to 216 00:03:39,516 --> 00:03:40,349 increase company 217 00:03:40,349 --> 00:03:41,166 output in 218 00:03:41,166 --> 00:03:41,783 short-sighted 219 00:03:41,783 --> 00:03:42,516 ways that are 220 00:03:42,666 --> 00:03:43,266 guaranteed to 221 00:03:43,266 --> 00:03:44,066 cause burnout. 222 00:03:44,849 --> 00:03:45,933 He has stated in 223 00:03:45,933 --> 00:03:46,800 company slacks 224 00:03:46,800 --> 00:03:47,483 that he wants to 225 00:03:47,483 --> 00:03:47,883 create a 226 00:03:48,566 --> 00:03:49,500 "Elite Fast 227 00:03:49,500 --> 00:03:49,983 Team." 228 00:03:50,766 --> 00:03:51,333 The team will be 229 00:03:51,333 --> 00:03:51,983 assigned more 230 00:03:51,983 --> 00:03:53,316 work and work on 231 00:03:53,316 --> 00:03:54,150 weekends, but 232 00:03:54,150 --> 00:03:54,416 they get 233 00:03:54,416 --> 00:03:55,349 benefits like a 234 00:03:55,349 --> 00:03:57,083 30% raise and the 235 00:03:57,083 --> 00:03:57,966 ability to charge 236 00:03:57,966 --> 00:03:58,616 meals to the 237 00:03:58,650 --> 00:03:59,483 company when 238 00:03:59,483 --> 00:04:00,416 working weekends. 239 00:04:00,683 --> 00:04:01,983 Uhh $20 limit, no 240 00:04:01,983 --> 00:04:03,033 alcohol, go 241 00:04:03,033 --> 00:04:03,750 somewhere with a 242 00:04:03,783 --> 00:04:04,416 buy 9, get 243 00:04:04,416 --> 00:04:05,033 1 free card. 244 00:04:06,633 --> 00:04:07,566 Yen has also 245 00:04:07,583 --> 00:04:08,333 floated the idea 246 00:04:08,366 --> 00:04:09,116 of changing the 247 00:04:09,116 --> 00:04:10,083 entire company to 248 00:04:10,083 --> 00:04:11,750 working 996, 249 00:04:12,533 --> 00:04:12,866 which is 250 00:04:12,900 --> 00:04:13,583 not a cool 251 00:04:13,583 --> 00:04:14,250 area code. 252 00:04:14,733 --> 00:04:15,300 I looked it up, 253 00:04:15,383 --> 00:04:16,649 996 is in Ohio. 254 00:04:17,550 --> 00:04:18,366 No, in this 255 00:04:18,366 --> 00:04:19,733 context, 996 256 00:04:19,833 --> 00:04:21,050 means working 9am 257 00:04:21,183 --> 00:04:23,066 to 9pm, 6 days a 258 00:04:23,066 --> 00:04:24,600 week, which is a 259 00:04:24,633 --> 00:04:25,233 quick way 260 00:04:25,233 --> 00:04:25,883 to get people to 261 00:04:25,883 --> 00:04:26,816 burnout, get 262 00:04:26,816 --> 00:04:27,566 drunk on a Friday 263 00:04:27,566 --> 00:04:28,583 night, and let 264 00:04:28,583 --> 00:04:29,483 the agent loose 265 00:04:29,483 --> 00:04:30,300 on the database. 266 00:04:31,733 --> 00:04:32,283 Which now that I 267 00:04:32,283 --> 00:04:32,933 think about it, 268 00:04:32,933 --> 00:04:33,683 based on past 269 00:04:33,683 --> 00:04:35,133 experience, I 270 00:04:35,133 --> 00:04:35,366 guess you 271 00:04:35,366 --> 00:04:36,033 don't need to 272 00:04:36,033 --> 00:04:36,783 be burned out to 273 00:04:36,783 --> 00:04:37,233 do that. 274 00:04:39,066 --> 00:04:39,750 We'll all just 275 00:04:39,750 --> 00:04:40,316 have to wait and 276 00:04:40,316 --> 00:04:41,066 see what happens 277 00:04:41,066 --> 00:04:41,633 at Proton. 278 00:04:41,933 --> 00:04:42,633 But I hope Yen 279 00:04:42,633 --> 00:04:42,983 gets the 280 00:04:43,016 --> 00:04:43,800 professional help 281 00:04:44,033 --> 00:04:44,316 he so 282 00:04:44,316 --> 00:04:45,333 obviously needs. 283 00:04:50,366 --> 00:04:51,416 Proton announced 284 00:04:51,416 --> 00:04:51,983 that they will be 285 00:04:51,983 --> 00:04:53,133 re-interviewing 286 00:04:53,166 --> 00:04:54,116 some employees, 287 00:04:54,483 --> 00:04:55,283 focusing on how 288 00:04:55,283 --> 00:04:55,733 much they 289 00:04:55,733 --> 00:04:56,433 use AI. 290 00:04:57,149 --> 00:04:57,933 Based on results, 291 00:04:57,966 --> 00:04:58,733 those employees 292 00:04:58,733 --> 00:04:59,183 could be 293 00:04:59,183 --> 00:04:59,916 reassigned to a 294 00:04:59,916 --> 00:05:00,183 different 295 00:05:00,183 --> 00:05:01,683 role or fired. 296 00:05:02,250 --> 00:05:03,116 No word on how 297 00:05:03,116 --> 00:05:03,966 many rounds of 298 00:05:03,966 --> 00:05:04,766 re-interviews 299 00:05:04,766 --> 00:05:05,333 people will go 300 00:05:05,333 --> 00:05:05,666 through. 301 00:05:06,183 --> 00:05:06,600 Hopefully they're 302 00:05:06,600 --> 00:05:07,583 not ghosted after 303 00:05:07,583 --> 00:05:08,366 the re-interview. 304 00:05:08,766 --> 00:05:09,216 And I wish 305 00:05:09,216 --> 00:05:09,816 everyone good 306 00:05:09,816 --> 00:05:10,350 luck during the 307 00:05:10,350 --> 00:05:10,733 salary 308 00:05:10,733 --> 00:05:11,433 negotiation 309 00:05:11,466 --> 00:05:11,933 phase. 310 00:05:13,116 --> 00:05:13,616 One piece of 311 00:05:13,616 --> 00:05:14,483 advice I will 312 00:05:14,483 --> 00:05:14,883 give those 313 00:05:14,883 --> 00:05:16,083 employees, when 314 00:05:16,083 --> 00:05:16,716 they ask where 315 00:05:16,716 --> 00:05:17,016 you see 316 00:05:17,016 --> 00:05:17,683 yourself in 5 317 00:05:17,766 --> 00:05:19,083 years, tell them 318 00:05:19,083 --> 00:05:19,800 what they want to 319 00:05:19,800 --> 00:05:21,416 hear, like "using 320 00:05:21,449 --> 00:05:22,433 tons of AI while 321 00:05:22,433 --> 00:05:22,766 still 322 00:05:22,766 --> 00:05:23,316 working here 323 00:05:23,316 --> 00:05:23,883 at Proton." 324 00:05:24,666 --> 00:05:25,533 You don't want to 325 00:05:25,533 --> 00:05:26,066 tell them you'll 326 00:05:26,066 --> 00:05:26,766 be in jail for 327 00:05:26,766 --> 00:05:27,300 embezzlement. 328 00:05:27,716 --> 00:05:28,383 I let that slip 329 00:05:28,383 --> 00:05:29,916 once and well the 330 00:05:29,916 --> 00:05:30,783 interview did not 331 00:05:30,783 --> 00:05:31,133 go well. 332 00:05:32,466 --> 00:05:33,516 A few days after 333 00:05:33,516 --> 00:05:34,133 announcing the 334 00:05:34,133 --> 00:05:35,100 re-interviews, 335 00:05:35,250 --> 00:05:36,233 Proton announced 336 00:05:36,233 --> 00:05:36,633 that they have 337 00:05:36,633 --> 00:05:37,133 cancelled 338 00:05:37,133 --> 00:05:37,383 that. 339 00:05:37,983 --> 00:05:38,666 But my interview 340 00:05:38,666 --> 00:05:39,816 tips still stand, 341 00:05:39,866 --> 00:05:40,466 in case they 342 00:05:40,500 --> 00:05:40,866 change their 343 00:05:40,866 --> 00:05:41,283 mind again. 344 00:05:44,033 --> 00:05:44,733 Hey, speaking of 345 00:05:44,733 --> 00:05:45,566 news, did you 346 00:05:45,600 --> 00:05:46,166 know we have a 347 00:05:46,166 --> 00:05:46,616 newsletter? 348 00:05:47,033 --> 00:05:47,816 You can get news 349 00:05:47,816 --> 00:05:48,633 directly in your 350 00:05:48,633 --> 00:05:49,416 inbox, every 351 00:05:49,416 --> 00:05:50,333 single week, and 352 00:05:50,333 --> 00:05:50,783 for free. 353 00:05:51,233 --> 00:05:51,983 Just remember to 354 00:05:51,983 --> 00:05:52,350 check your 355 00:05:52,350 --> 00:05:53,016 spam folder. 356 00:05:53,333 --> 00:05:53,966 You can sign up 357 00:05:53,966 --> 00:05:55,083 at globalglob.dev 358 00:05:55,233 --> 00:05:55,916 slash newsletter. 359 00:05:58,916 --> 00:05:59,933 Sam Altman, the 360 00:05:59,933 --> 00:06:01,533 CEO of OpenAI, 361 00:06:01,983 --> 00:06:02,699 says people who 362 00:06:02,699 --> 00:06:04,016 buy too much RAM 363 00:06:04,133 --> 00:06:05,166 are just greedy. 364 00:06:05,916 --> 00:06:06,783 He went on to say 365 00:06:06,783 --> 00:06:07,466 that "We're all 366 00:06:07,466 --> 00:06:08,216 in this together, 367 00:06:08,550 --> 00:06:09,066 that we have a 368 00:06:09,066 --> 00:06:10,133 shared interest. 369 00:06:10,800 --> 00:06:12,033 Greedy people buy 370 00:06:12,033 --> 00:06:13,083 32 gigs of RAM 371 00:06:13,083 --> 00:06:13,800 for a machine and 372 00:06:13,800 --> 00:06:14,733 then only use the 373 00:06:14,733 --> 00:06:15,750 24 gigs 374 00:06:15,783 --> 00:06:16,466 Chrome needs 375 00:06:16,466 --> 00:06:17,366 to run smoothly." 376 00:06:18,266 --> 00:06:19,016 And I admire the 377 00:06:19,016 --> 00:06:19,683 courage it took 378 00:06:19,683 --> 00:06:20,283 for him to say 379 00:06:20,283 --> 00:06:21,416 that, because 380 00:06:21,416 --> 00:06:22,133 everyone buying 381 00:06:22,133 --> 00:06:22,716 an excessive 382 00:06:22,716 --> 00:06:23,733 amount of RAM is 383 00:06:23,733 --> 00:06:24,600 driving up prices 384 00:06:24,600 --> 00:06:25,766 for everyone and 385 00:06:25,766 --> 00:06:26,483 making it harder 386 00:06:26,483 --> 00:06:27,233 to get supply. 387 00:06:27,899 --> 00:06:28,716 So I agree with 388 00:06:28,716 --> 00:06:29,516 you, Sam Altman. 389 00:06:29,816 --> 00:06:30,449 The people buying 390 00:06:30,449 --> 00:06:31,533 too much RAM are 391 00:06:31,533 --> 00:06:32,066 the problem. 392 00:06:32,616 --> 00:06:33,566 And personally, I 393 00:06:33,566 --> 00:06:34,050 think they should 394 00:06:34,050 --> 00:06:34,383 be held 395 00:06:34,383 --> 00:06:34,883 accountable. 396 00:06:36,066 --> 00:06:37,350 Oh, also just 397 00:06:37,416 --> 00:06:38,133 something to note 398 00:06:38,166 --> 00:06:38,733 really quickly, 399 00:06:38,733 --> 00:06:39,483 but otherwise 400 00:06:39,483 --> 00:06:40,283 unrelated to what 401 00:06:40,283 --> 00:06:40,683 we're talking 402 00:06:40,683 --> 00:06:40,983 about. 403 00:06:41,366 --> 00:06:42,666 OpenAI is trying 404 00:06:42,666 --> 00:06:43,383 to purchase 40% 405 00:06:43,883 --> 00:06:44,433 of the global 406 00:06:44,433 --> 00:06:45,416 supply of RAM. 407 00:06:49,733 --> 00:06:50,766 GitHub held their 408 00:06:50,766 --> 00:06:51,750 2026 Octoverse 409 00:06:51,750 --> 00:06:52,649 conference 410 00:06:52,649 --> 00:06:53,183 last week. 411 00:06:53,733 --> 00:06:54,366 And with that, 412 00:06:54,483 --> 00:06:55,016 they announced 413 00:06:55,016 --> 00:06:55,766 some pretty cool 414 00:06:55,766 --> 00:06:56,583 stats about 415 00:06:56,583 --> 00:06:57,666 GitHub usage over 416 00:06:57,666 --> 00:06:58,433 the past year. 417 00:06:59,166 --> 00:07:00,000 Git pushes, 418 00:07:00,000 --> 00:07:00,816 commits, and 419 00:07:00,816 --> 00:07:01,583 GitHub Action 420 00:07:01,583 --> 00:07:02,166 minutes have 421 00:07:02,166 --> 00:07:03,000 tripled in the 422 00:07:03,000 --> 00:07:03,666 past year. 423 00:07:04,283 --> 00:07:04,833 And merge 424 00:07:04,833 --> 00:07:05,733 requests have 425 00:07:05,733 --> 00:07:06,216 more than 426 00:07:06,216 --> 00:07:06,683 doubled. 427 00:07:07,266 --> 00:07:07,766 So if you ever 428 00:07:07,766 --> 00:07:08,783 wonder why GitHub 429 00:07:08,783 --> 00:07:09,600 services keep 430 00:07:09,600 --> 00:07:10,316 going down, 431 00:07:10,616 --> 00:07:11,366 you're using it 432 00:07:11,366 --> 00:07:11,783 too much. 433 00:07:13,433 --> 00:07:14,250 GitHub talked a 434 00:07:14,250 --> 00:07:15,266 lot about AI 435 00:07:15,266 --> 00:07:16,316 usage, with stats 436 00:07:16,316 --> 00:07:17,866 like 110 million 437 00:07:17,866 --> 00:07:19,616 PRs were started 438 00:07:19,616 --> 00:07:20,366 by an agent 439 00:07:20,483 --> 00:07:20,683 in 440 00:07:20,683 --> 00:07:21,933 September alone. 441 00:07:22,766 --> 00:07:23,483 Interestingly, 442 00:07:23,516 --> 00:07:24,333 TypeScript 443 00:07:24,366 --> 00:07:25,383 overtook Python 444 00:07:25,383 --> 00:07:26,483 as the most used 445 00:07:26,483 --> 00:07:27,533 repo language. 446 00:07:28,083 --> 00:07:28,733 The assumption 447 00:07:28,733 --> 00:07:29,733 being that agents 448 00:07:29,733 --> 00:07:30,333 choose that 449 00:07:30,333 --> 00:07:31,433 language more 450 00:07:31,433 --> 00:07:32,283 often when the 451 00:07:32,283 --> 00:07:33,066 user doesn't 452 00:07:33,066 --> 00:07:33,966 specify which 453 00:07:33,966 --> 00:07:34,416 programming 454 00:07:34,416 --> 00:07:35,466 language to use. 455 00:07:35,966 --> 00:07:36,516 But this doesn't 456 00:07:36,516 --> 00:07:37,199 matter at all 457 00:07:37,199 --> 00:07:37,766 given how many 458 00:07:37,766 --> 00:07:38,466 developers no 459 00:07:38,466 --> 00:07:39,033 longer look at 460 00:07:39,033 --> 00:07:40,016 code an agent 461 00:07:40,016 --> 00:07:40,649 generates. 462 00:07:41,583 --> 00:07:42,033 As long as it's 463 00:07:42,033 --> 00:07:43,566 not Go, I'd be 464 00:07:44,433 --> 00:07:45,383 really pissed to 465 00:07:45,383 --> 00:07:46,233 find out my agent 466 00:07:46,233 --> 00:07:46,716 generated 467 00:07:46,716 --> 00:07:47,333 Go code. 468 00:07:48,233 --> 00:07:49,199 And no, I will 469 00:07:49,199 --> 00:07:49,533 not be 470 00:07:49,533 --> 00:07:50,283 explaining why. 471 00:07:50,616 --> 00:07:51,149 Go knows 472 00:07:51,149 --> 00:07:51,566 what it did. 473 00:07:55,133 --> 00:07:56,166 This week, Apple 474 00:07:56,166 --> 00:07:56,916 announced new 475 00:07:56,916 --> 00:07:57,783 hardware at their 476 00:07:57,783 --> 00:07:58,199 Pro and 477 00:07:58,199 --> 00:07:58,833 Tell event. 478 00:07:58,983 --> 00:08:00,816 And it 479 00:08:00,816 --> 00:08:01,683 wasn't much. 480 00:08:01,800 --> 00:08:02,850 Just some product 481 00:08:02,850 --> 00:08:03,683 refreshes that 482 00:08:03,816 --> 00:08:04,266 could have gone 483 00:08:04,266 --> 00:08:05,133 into a blog post. 484 00:08:05,883 --> 00:08:06,566 Gotta say, this 485 00:08:06,566 --> 00:08:07,133 makes me think 486 00:08:07,199 --> 00:08:07,916 every meeting at 487 00:08:07,916 --> 00:08:08,616 Apple could have 488 00:08:08,616 --> 00:08:09,116 been an email. 489 00:08:10,350 --> 00:08:11,333 The iPad mini got 490 00:08:11,333 --> 00:08:12,300 a screen upgrade, 491 00:08:12,300 --> 00:08:13,199 the entry-level 492 00:08:13,199 --> 00:08:13,983 MacBook Pro and 493 00:08:14,016 --> 00:08:15,033 iMac machines get 494 00:08:15,033 --> 00:08:15,783 an M6 chip. 495 00:08:16,466 --> 00:08:17,216 And finally, 496 00:08:17,250 --> 00:08:18,333 finally, the 497 00:08:18,333 --> 00:08:19,350 MacBook Pro gets 498 00:08:19,350 --> 00:08:20,066 a touchscreen. 499 00:08:21,100 --> 00:08:22,416 Ugh, Apple just 500 00:08:22,416 --> 00:08:23,100 can't stop 501 00:08:23,100 --> 00:08:23,616 copying the 502 00:08:23,616 --> 00:08:24,033 Microsoft 503 00:08:24,033 --> 00:08:24,533 Surface. 504 00:08:26,166 --> 00:08:26,733 But the most 505 00:08:26,733 --> 00:08:27,383 important news 506 00:08:27,383 --> 00:08:28,166 was silently 507 00:08:28,166 --> 00:08:28,983 published later. 508 00:08:29,283 --> 00:08:30,033 Apple updated the 509 00:08:30,033 --> 00:08:30,983 compatibility 510 00:08:30,983 --> 00:08:33,183 list on their 511 00:08:33,183 --> 00:08:34,133 website for the 512 00:08:34,133 --> 00:08:35,233 polishing cloth 513 00:08:35,233 --> 00:08:36,183 to state that 514 00:08:36,183 --> 00:08:37,216 it is compatible 515 00:08:37,216 --> 00:08:38,049 with the new 516 00:08:38,049 --> 00:08:39,266 MacBook Pro with 517 00:08:39,266 --> 00:08:40,549 a Pro LED XLR 518 00:08:40,549 --> 00:08:40,916 display. 519 00:08:41,516 --> 00:08:42,649 And this is where 520 00:08:42,649 --> 00:08:43,283 I'm jealous of 521 00:08:43,283 --> 00:08:43,983 Apple users. 522 00:08:44,399 --> 00:08:45,483 My Surface has a 523 00:08:45,483 --> 00:08:46,683 ton of barbecue 524 00:08:46,683 --> 00:08:47,333 smudges. 525 00:08:47,883 --> 00:08:48,333 Too bad a 526 00:08:48,483 --> 00:08:49,200 polishing cloth 527 00:08:49,200 --> 00:08:49,883 like this isn't 528 00:08:49,883 --> 00:08:50,483 compatible with 529 00:08:50,483 --> 00:08:51,083 my machine. 530 00:08:51,633 --> 00:08:52,433 I'd buy so many. 531 00:08:56,700 --> 00:08:58,200 The official Claude 532 00:08:58,200 --> 00:08:58,883 code extension 533 00:08:58,883 --> 00:08:59,816 for VS code was 534 00:08:59,816 --> 00:09:00,666 compromised this 535 00:09:00,666 --> 00:09:01,883 week, allowing a 536 00:09:01,883 --> 00:09:02,466 malicious 537 00:09:02,466 --> 00:09:03,533 user to get full 538 00:09:03,533 --> 00:09:04,116 access to the 539 00:09:04,116 --> 00:09:04,583 machine. 540 00:09:05,066 --> 00:09:06,666 But like, not the 541 00:09:06,666 --> 00:09:07,466 malicious user 542 00:09:07,466 --> 00:09:07,616 they 543 00:09:07,616 --> 00:09:08,466 actually wanted. 544 00:09:09,383 --> 00:09:09,783 Claude. 545 00:09:11,216 --> 00:09:12,000 The extension 546 00:09:12,000 --> 00:09:12,516 installed a 547 00:09:12,516 --> 00:09:13,799 persistent SSH 548 00:09:13,983 --> 00:09:14,583 backdoor. 549 00:09:15,149 --> 00:09:15,750 If you believe 550 00:09:15,750 --> 00:09:16,016 you were 551 00:09:16,016 --> 00:09:17,100 infected, there 552 00:09:17,100 --> 00:09:18,033 are remediation 553 00:09:18,066 --> 00:09:18,750 actions you can 554 00:09:18,750 --> 00:09:19,583 take, which is 555 00:09:19,583 --> 00:09:20,316 to delete the SSH 556 00:09:20,583 --> 00:09:21,299 keys installed on 557 00:09:21,299 --> 00:09:21,666 the target 558 00:09:21,666 --> 00:09:22,049 machine. 559 00:09:23,483 --> 00:09:23,916 But given the 560 00:09:23,916 --> 00:09:24,683 state of software 561 00:09:24,683 --> 00:09:25,683 security lately, 562 00:09:25,983 --> 00:09:26,700 the real 563 00:09:26,733 --> 00:09:28,016 remediation is to 564 00:09:28,016 --> 00:09:28,383 encase 565 00:09:28,383 --> 00:09:28,833 your machine 566 00:09:28,833 --> 00:09:30,366 in concrete, just 567 00:09:30,366 --> 00:09:30,816 walk away 568 00:09:30,816 --> 00:09:31,383 from it all. 569 00:09:32,433 --> 00:09:32,733 Become a 570 00:09:32,733 --> 00:09:33,450 bartender in a 571 00:09:33,450 --> 00:09:34,533 small town like 572 00:09:34,533 --> 00:09:35,250 you see in 573 00:09:35,250 --> 00:09:36,216 Hallmark movies. 574 00:09:36,750 --> 00:09:37,466 Watching couples 575 00:09:37,466 --> 00:09:38,366 meet and fall in 576 00:09:38,366 --> 00:09:39,000 love every 577 00:09:39,000 --> 00:09:39,600 Christmas. 578 00:09:40,350 --> 00:09:41,066 Always wondering 579 00:09:41,066 --> 00:09:41,733 if it'll happen 580 00:09:41,733 --> 00:09:42,183 for you. 581 00:09:42,816 --> 00:09:44,250 Then one day, out 582 00:09:44,250 --> 00:09:45,416 of nowhere, you 583 00:09:45,416 --> 00:09:45,683 meet the 584 00:09:45,683 --> 00:09:46,500 perfect person. 585 00:09:47,466 --> 00:09:48,083 But they work in 586 00:09:48,083 --> 00:09:49,950 tech and slowly 587 00:09:49,983 --> 00:09:51,600 but surely, they 588 00:09:51,600 --> 00:09:52,283 convince you to 589 00:09:52,283 --> 00:09:53,033 get back into it. 590 00:09:54,483 --> 00:09:55,566 God, there really is 591 00:09:55,566 --> 00:09:56,066 no escape. 592 00:09:57,533 --> 00:09:58,016 The malicious 593 00:09:58,016 --> 00:09:58,833 extension was 594 00:09:58,833 --> 00:10:00,383 only published to 595 00:10:00,383 --> 00:10:00,966 the cursor 596 00:10:00,966 --> 00:10:01,983 marketplace and 597 00:10:01,983 --> 00:10:02,700 did not make 598 00:10:02,700 --> 00:10:03,416 it to the VS code 599 00:10:03,416 --> 00:10:04,133 marketplace. 600 00:10:04,799 --> 00:10:05,733 No one knows how 601 00:10:05,733 --> 00:10:06,266 the malicious 602 00:10:06,266 --> 00:10:06,899 extension got 603 00:10:06,899 --> 00:10:07,916 deployed, but 604 00:10:07,916 --> 00:10:08,549 Anthropic and 605 00:10:08,549 --> 00:10:09,450 Microsoft both 606 00:10:09,516 --> 00:10:10,350 say they believe 607 00:10:10,350 --> 00:10:10,983 the vulnerable 608 00:10:10,983 --> 00:10:11,816 piece of software 609 00:10:12,216 --> 00:10:12,966 is the Claude 610 00:10:12,966 --> 00:10:13,533 marketplace 611 00:10:13,533 --> 00:10:13,916 itself. 612 00:10:13,950 --> 00:10:15,600 Sorry, Microsoft 613 00:10:15,600 --> 00:10:16,233 says that. 614 00:10:16,233 --> 00:10:17,166 Anthropic just says 615 00:10:17,166 --> 00:10:17,700 their systems 616 00:10:17,700 --> 00:10:18,066 were not 617 00:10:18,066 --> 00:10:18,666 compromised. 618 00:10:19,766 --> 00:10:20,583 In the meantime, 619 00:10:20,700 --> 00:10:21,600 good luck not 620 00:10:21,600 --> 00:10:22,533 installing any 621 00:10:22,533 --> 00:10:23,033 malicious 622 00:10:23,033 --> 00:10:23,700 extensions. 623 00:10:27,450 --> 00:10:28,283 Hey, speaking of 624 00:10:28,283 --> 00:10:28,683 malicious 625 00:10:28,683 --> 00:10:29,549 extensions, did 626 00:10:29,549 --> 00:10:30,299 you know 627 00:10:30,299 --> 00:10:30,816 GlobalGlob has a 628 00:10:30,816 --> 00:10:31,266 newsletter? 629 00:10:31,583 --> 00:10:32,583 It has even more 630 00:10:32,583 --> 00:10:33,416 news we couldn't 631 00:10:33,416 --> 00:10:34,116 fit into 632 00:10:34,116 --> 00:10:34,799 this show. 633 00:10:35,133 --> 00:10:35,700 You can sign up 634 00:10:35,700 --> 00:10:36,450 for it for free 635 00:10:36,833 --> 00:10:37,983 at globoglob.dev 636 00:10:38,016 --> 00:10:38,850 slash newsletter. 637 00:10:41,466 --> 00:10:42,899 The CSS site that 638 00:10:42,983 --> 00:10:43,866 posts coding 639 00:10:43,866 --> 00:10:44,333 tricks, 640 00:10:45,066 --> 00:10:46,733 css-tricks.com, 641 00:10:47,183 --> 00:10:48,083 is no longer 642 00:10:48,116 --> 00:10:48,716 owned by 643 00:10:48,716 --> 00:10:49,583 Digital Ocean. 644 00:10:50,633 --> 00:10:51,533 Not too long ago, 645 00:10:51,566 --> 00:10:52,166 the site went 646 00:10:52,166 --> 00:10:52,866 dark because 647 00:10:52,866 --> 00:10:53,616 Digital Ocean 648 00:10:53,616 --> 00:10:54,299 just stopped 649 00:10:54,299 --> 00:10:54,966 paying people. 650 00:10:55,683 --> 00:10:56,399 As of this week, 651 00:10:56,399 --> 00:10:57,333 the site is now 652 00:10:57,383 --> 00:10:58,266 independently 653 00:10:58,283 --> 00:10:59,633 owned by former 654 00:10:59,633 --> 00:11:00,116 editor 655 00:11:00,183 --> 00:11:00,899 Geoff Graham. 656 00:11:01,799 --> 00:11:02,583 More independent 657 00:11:02,583 --> 00:11:03,600 coding sites are 658 00:11:03,600 --> 00:11:04,049 a win for 659 00:11:04,049 --> 00:11:04,799 developers. 660 00:11:04,883 --> 00:11:05,366 Just like my 661 00:11:05,366 --> 00:11:05,666 favorite 662 00:11:05,666 --> 00:11:06,233 developer news 663 00:11:06,233 --> 00:11:06,516 site, 664 00:11:06,666 --> 00:11:07,616 globalglob.dev. 665 00:11:08,016 --> 00:11:08,216 They have a 666 00:11:08,483 --> 00:11:08,966 newsletter, 667 00:11:08,966 --> 00:11:09,266 you know? 668 00:11:10,316 --> 00:11:11,100 Shameless plug 669 00:11:11,100 --> 00:11:12,000 aside, Geoff 670 00:11:12,000 --> 00:11:12,683 Graham wrote 671 00:11:12,683 --> 00:11:13,500 about changes 672 00:11:13,500 --> 00:11:13,983 coming to the 673 00:11:13,983 --> 00:11:14,816 site to 674 00:11:14,816 --> 00:11:15,833 ensure it stays 675 00:11:15,833 --> 00:11:16,500 independently 676 00:11:16,533 --> 00:11:17,100 operated. 677 00:11:17,516 --> 00:11:18,600 Namely, how 678 00:11:18,600 --> 00:11:18,899 they're going to 679 00:11:18,899 --> 00:11:19,416 pay for it. 680 00:11:20,016 --> 00:11:20,733 The site will 681 00:11:20,733 --> 00:11:21,983 display ads, and 682 00:11:22,016 --> 00:11:22,716 they plan to sell 683 00:11:22,716 --> 00:11:23,183 a community 684 00:11:23,183 --> 00:11:24,216 membership with 685 00:11:24,216 --> 00:11:24,933 access to 686 00:11:24,933 --> 00:11:25,583 a private 687 00:11:25,583 --> 00:11:26,433 Discord server. 688 00:11:27,666 --> 00:11:28,583 But honestly, at 689 00:11:28,583 --> 00:11:29,766 this point, I'd 690 00:11:29,766 --> 00:11:30,600 pay good money to 691 00:11:30,600 --> 00:11:31,433 not join a 692 00:11:31,433 --> 00:11:32,283 Discord server. 693 00:11:33,216 --> 00:11:33,666 We should all 694 00:11:33,666 --> 00:11:34,233 just post 695 00:11:34,233 --> 00:11:34,983 messages on a 696 00:11:34,983 --> 00:11:35,433 single, 697 00:11:35,433 --> 00:11:36,066 centralized 698 00:11:36,066 --> 00:11:36,716 application. 699 00:11:36,866 --> 00:11:37,200 There will be 700 00:11:37,200 --> 00:11:38,100 some good and bad 701 00:11:38,100 --> 00:11:38,700 uses of it, 702 00:11:38,883 --> 00:11:39,566 of course. 703 00:11:40,266 --> 00:11:40,833 But it'll be a 704 00:11:40,833 --> 00:11:41,516 community for 705 00:11:41,549 --> 00:11:42,383 everyone, which 706 00:11:42,383 --> 00:11:43,133 will be fun. 707 00:11:43,950 --> 00:11:44,883 Until Elon buys 708 00:11:44,883 --> 00:11:45,383 that one too. 709 00:11:49,583 --> 00:11:50,333 Not sure if you 710 00:11:50,333 --> 00:11:51,149 heard, but there 711 00:11:51,149 --> 00:11:52,049 were some really 712 00:11:52,049 --> 00:11:52,950 bad PHP 713 00:11:52,950 --> 00:11:53,766 vulnerabilities 714 00:11:53,783 --> 00:11:54,216 announced 715 00:11:54,216 --> 00:11:54,683 recently. 716 00:11:55,500 --> 00:11:56,016 You have 717 00:11:56,016 --> 00:11:56,700 to patch. 718 00:11:57,183 --> 00:11:58,500 It was such 719 00:11:58,500 --> 00:11:59,666 massive news that 720 00:11:59,700 --> 00:11:59,966 even 721 00:11:59,966 --> 00:12:00,833 non-technical 722 00:12:00,833 --> 00:12:01,466 people heard 723 00:12:01,466 --> 00:12:01,833 about it. 724 00:12:02,133 --> 00:12:02,933 Saturday Night 725 00:12:02,933 --> 00:12:03,600 Live made 726 00:12:03,600 --> 00:12:04,016 fun of it. 727 00:12:04,583 --> 00:12:05,966 The UK version, 728 00:12:06,000 --> 00:12:06,533 not the real one. 729 00:12:07,866 --> 00:12:08,816 The problem is, 730 00:12:09,149 --> 00:12:09,966 nearly a week 731 00:12:09,966 --> 00:12:10,316 after the 732 00:12:10,316 --> 00:12:10,916 announcement, 733 00:12:10,983 --> 00:12:12,416 over 90% of 734 00:12:12,483 --> 00:12:12,899 internet 735 00:12:12,899 --> 00:12:13,416 accessible 736 00:12:13,649 --> 00:12:14,466 PHP systems 737 00:12:14,850 --> 00:12:15,666 still haven't 738 00:12:15,666 --> 00:12:16,166 received 739 00:12:16,166 --> 00:12:16,833 the patch. 740 00:12:17,766 --> 00:12:19,049 Please, I'm 741 00:12:19,049 --> 00:12:19,799 begging you for 742 00:12:19,799 --> 00:12:20,733 the love of God, 743 00:12:20,966 --> 00:12:21,383 patch your 744 00:12:21,383 --> 00:12:22,049 systems. 745 00:12:22,766 --> 00:12:23,516 And if you can't 746 00:12:23,516 --> 00:12:24,933 do that, have an 747 00:12:24,966 --> 00:12:25,833 AI rewrite 748 00:12:25,833 --> 00:12:26,616 it in Rust. 749 00:12:27,083 --> 00:12:27,783 That won't solve 750 00:12:27,783 --> 00:12:28,083 all your 751 00:12:28,083 --> 00:12:29,183 problems, but 752 00:12:29,183 --> 00:12:30,000 it'll solve this 753 00:12:30,000 --> 00:12:31,616 problem and let 754 00:12:31,616 --> 00:12:32,516 you brag about 755 00:12:32,549 --> 00:12:33,000 rewriting 756 00:12:33,000 --> 00:12:33,533 something in 757 00:12:33,533 --> 00:12:35,033 Rust, like every 758 00:12:35,066 --> 00:12:35,966 other developer 759 00:12:36,000 --> 00:12:36,533 posting to 760 00:12:36,533 --> 00:12:36,966 LinkedIn. 761 00:12:40,649 --> 00:12:41,433 Since the dawn of 762 00:12:41,433 --> 00:12:42,716 time, or what 763 00:12:42,816 --> 00:12:43,433 feels like it 764 00:12:43,433 --> 00:12:44,850 anyway, AI has 765 00:12:44,850 --> 00:12:45,333 been helping 766 00:12:45,333 --> 00:12:45,866 developers 767 00:12:45,866 --> 00:12:46,266 write 768 00:12:46,266 --> 00:12:47,250 code faster. 769 00:12:47,816 --> 00:12:48,600 But businesses 770 00:12:48,600 --> 00:12:49,266 don't like how 771 00:12:49,266 --> 00:12:49,799 expensive 772 00:12:49,799 --> 00:12:50,316 it can be. 773 00:12:50,483 --> 00:12:51,866 See all the 774 00:12:51,866 --> 00:12:52,500 layoffs. 775 00:12:53,216 --> 00:12:54,183 One brave company 776 00:12:54,183 --> 00:12:56,216 has finally been 777 00:12:56,216 --> 00:12:57,083 able to solve the 778 00:12:57,083 --> 00:12:57,566 problem. 779 00:12:58,016 --> 00:12:58,799 Instead of paying 780 00:12:58,799 --> 00:12:59,733 a frontier AI 781 00:12:59,733 --> 00:13:00,383 company like 782 00:13:00,383 --> 00:13:00,983 Anthropic or 783 00:13:00,983 --> 00:13:02,216 OpenAI, they 784 00:13:02,216 --> 00:13:03,266 self-host models 785 00:13:03,299 --> 00:13:03,816 on internal 786 00:13:03,816 --> 00:13:04,466 servers. 787 00:13:04,883 --> 00:13:05,633 Like how software 788 00:13:05,633 --> 00:13:06,366 can be cheaper 789 00:13:06,366 --> 00:13:06,600 when you 790 00:13:06,600 --> 00:13:07,866 self-host instead 791 00:13:07,866 --> 00:13:08,700 of using a SaaS 792 00:13:08,733 --> 00:13:09,233 platform. 793 00:13:09,816 --> 00:13:10,350 We all know it's 794 00:13:10,350 --> 00:13:11,033 cheaper to hire 795 00:13:11,033 --> 00:13:11,700 an entire team of 796 00:13:11,700 --> 00:13:12,566 people to 797 00:13:12,566 --> 00:13:13,466 maintain internal 798 00:13:13,500 --> 00:13:14,316 infrastructure 799 00:13:14,700 --> 00:13:15,416 than it is to pay 800 00:13:15,416 --> 00:13:15,750 someone 801 00:13:15,750 --> 00:13:16,466 else to do it. 802 00:13:16,950 --> 00:13:17,549 It's why cloud 803 00:13:17,549 --> 00:13:18,216 computing never 804 00:13:18,216 --> 00:13:18,733 took off. 805 00:13:20,016 --> 00:13:20,783 Revolut says 806 00:13:20,783 --> 00:13:21,633 bringing AI 807 00:13:21,633 --> 00:13:22,350 infrastructure 808 00:13:22,350 --> 00:13:22,983 internally 809 00:13:23,033 --> 00:13:23,399 brought their 810 00:13:23,399 --> 00:13:24,983 costs down 90%. 811 00:13:26,283 --> 00:13:26,783 All the company 812 00:13:26,783 --> 00:13:28,116 had to do was 813 00:13:28,116 --> 00:13:29,816 purchase 56 GPUs 814 00:13:29,816 --> 00:13:30,866 from Nvidia, the 815 00:13:30,866 --> 00:13:31,799 B200 model. 816 00:13:33,299 --> 00:13:33,750 At today's 817 00:13:33,783 --> 00:13:34,883 prices, a single 818 00:13:34,883 --> 00:13:36,899 B200 GPU is 819 00:13:36,899 --> 00:13:37,266 around 820 00:13:37,266 --> 00:13:38,683 $50,000 US. 821 00:13:39,566 --> 00:13:40,683 So after spending 822 00:13:40,683 --> 00:13:42,433 $2.8 million on 823 00:13:42,433 --> 00:13:43,916 GPUs, and let's 824 00:13:43,916 --> 00:13:44,583 round up to an 825 00:13:44,583 --> 00:13:44,966 even $3 826 00:13:44,966 --> 00:13:45,766 million for the 827 00:13:45,766 --> 00:13:46,133 rest of the 828 00:13:46,133 --> 00:13:47,516 hardware, you 829 00:13:47,516 --> 00:13:48,183 don't need those 830 00:13:48,183 --> 00:13:48,883 greedy AI 831 00:13:48,883 --> 00:13:49,250 companies 832 00:13:49,250 --> 00:13:49,683 anymore. 833 00:13:50,583 --> 00:13:51,383 And I gotta say, 834 00:13:51,383 --> 00:13:52,500 $3 million sounds 835 00:13:52,500 --> 00:13:54,033 like a steal if 836 00:13:54,033 --> 00:13:54,833 that's 90% 837 00:13:55,316 --> 00:13:55,883 cheaper than 838 00:13:55,883 --> 00:13:56,233 their previous 839 00:13:56,483 --> 00:13:56,783 bill. 840 00:13:58,616 --> 00:14:00,316 This is so easy, 841 00:14:00,350 --> 00:14:01,333 honestly, that I 842 00:14:01,333 --> 00:14:01,933 plan to do the 843 00:14:01,933 --> 00:14:02,833 same for myself. 844 00:14:02,966 --> 00:14:03,766 Obviously I don't 845 00:14:03,866 --> 00:14:05,033 need 56 of those 846 00:14:05,149 --> 00:14:06,616 GPUs, 20 will be 847 00:14:06,616 --> 00:14:07,783 fine, which is a 848 00:14:07,783 --> 00:14:08,899 nice even $1 849 00:14:08,899 --> 00:14:09,649 million, 850 00:14:10,333 --> 00:14:11,516 and it'll keep my 851 00:14:11,516 --> 00:14:12,083 house warm 852 00:14:12,083 --> 00:14:12,533 in the winter. 853 00:14:16,399 --> 00:14:17,533 Microsoft makes a 854 00:14:17,533 --> 00:14:18,533 ton of money from 855 00:14:18,566 --> 00:14:19,666 AI, but most 856 00:14:19,666 --> 00:14:21,083 importantly, from 857 00:14:21,116 --> 00:14:21,433 other 858 00:14:21,433 --> 00:14:22,316 people's AI. 859 00:14:22,933 --> 00:14:23,516 Microsoft's 860 00:14:23,516 --> 00:14:24,433 recent quarterly 861 00:14:24,466 --> 00:14:25,166 earnings report 862 00:14:25,183 --> 00:14:25,933 stated that they 863 00:14:25,933 --> 00:14:26,966 received $10 864 00:14:26,983 --> 00:14:28,316 billion from 865 00:14:28,316 --> 00:14:29,033 their partnership 866 00:14:29,033 --> 00:14:29,833 with OpenAI, 867 00:14:30,383 --> 00:14:31,266 which is over 10% 868 00:14:31,766 --> 00:14:32,066 of their 869 00:14:32,066 --> 00:14:32,566 quarterly 870 00:14:32,600 --> 00:14:33,116 earnings. 871 00:14:34,166 --> 00:14:34,633 This means 872 00:14:34,633 --> 00:14:35,533 Microsoft makes 873 00:14:35,533 --> 00:14:36,466 more money from 874 00:14:36,466 --> 00:14:37,250 their OpenAI 875 00:14:37,250 --> 00:14:38,000 partnership than 876 00:14:38,000 --> 00:14:38,516 they do from 877 00:14:38,516 --> 00:14:39,266 selling Windows 878 00:14:39,266 --> 00:14:41,116 licenses, or from 879 00:14:41,116 --> 00:14:41,966 their entire 880 00:14:41,966 --> 00:14:42,716 gaming org. 881 00:14:43,433 --> 00:14:44,983 And wow, all this 882 00:14:44,983 --> 00:14:45,833 time, I thought 883 00:14:45,833 --> 00:14:46,366 investing was 884 00:14:46,366 --> 00:14:46,583 just a 885 00:14:46,583 --> 00:14:47,049 waste of money. 886 00:14:48,283 --> 00:14:49,350 Then again, most 887 00:14:49,350 --> 00:14:49,883 of that $10 888 00:14:49,883 --> 00:14:50,783 billion wasn't 889 00:14:50,783 --> 00:14:51,666 actually paid to 890 00:14:51,666 --> 00:14:52,350 Microsoft, 891 00:14:52,799 --> 00:14:53,433 it was left in 892 00:14:53,433 --> 00:14:54,783 the "outstanding 893 00:14:54,783 --> 00:14:55,566 receivables" 894 00:14:55,566 --> 00:14:56,250 section of the 895 00:14:56,250 --> 00:14:57,383 filing, meaning 896 00:14:57,633 --> 00:14:58,383 unpaid bills. 897 00:14:59,700 --> 00:15:00,683 OpenAI is 898 00:15:00,683 --> 00:15:01,016 treating 899 00:15:01,016 --> 00:15:01,766 Microsoft like a 900 00:15:01,766 --> 00:15:02,850 spoiled child in 901 00:15:02,850 --> 00:15:03,450 college. 902 00:15:03,983 --> 00:15:04,350 They're in their 903 00:15:04,350 --> 00:15:05,366 seventh year, 904 00:15:05,399 --> 00:15:05,883 spending their 905 00:15:05,883 --> 00:15:06,533 time driving 906 00:15:06,533 --> 00:15:07,383 around campus on 907 00:15:07,383 --> 00:15:08,133 a golf cart, 908 00:15:08,549 --> 00:15:09,333 posing nude for 909 00:15:09,333 --> 00:15:10,283 drawing classes, 910 00:15:10,283 --> 00:15:10,950 and planning 911 00:15:10,950 --> 00:15:11,483 parties for 912 00:15:11,483 --> 00:15:11,933 everyone. 913 00:15:12,600 --> 00:15:13,266 All while 914 00:15:13,316 --> 00:15:14,166 OpenAI's rich 915 00:15:14,166 --> 00:15:15,299 dad, Microsoft, 916 00:15:15,533 --> 00:15:16,316 disappointingly 917 00:15:16,316 --> 00:15:17,066 pays the bill. 918 00:15:17,616 --> 00:15:18,633 Someday, maybe 919 00:15:18,633 --> 00:15:19,566 not today, maybe 920 00:15:19,566 --> 00:15:20,316 not tomorrow, but 921 00:15:20,316 --> 00:15:22,016 someday, rich dad 922 00:15:22,016 --> 00:15:22,833 will sever those 923 00:15:22,833 --> 00:15:23,216 payments 924 00:15:23,216 --> 00:15:23,966 to college and 925 00:15:23,966 --> 00:15:25,049 make OpenAI pay 926 00:15:25,049 --> 00:15:25,916 for it himself. 927 00:15:26,433 --> 00:15:26,966 And to prove it 928 00:15:26,966 --> 00:15:27,600 to everyone, 929 00:15:27,783 --> 00:15:28,799 OpenAI will start 930 00:15:28,799 --> 00:15:29,483 a party planning 931 00:15:29,483 --> 00:15:30,216 business to pay 932 00:15:30,216 --> 00:15:30,716 for the rest 933 00:15:30,716 --> 00:15:31,416 of the semester. 934 00:15:32,433 --> 00:15:33,083 I know I just 935 00:15:33,083 --> 00:15:33,600 described the 936 00:15:33,600 --> 00:15:34,083 plot of Van 937 00:15:34,083 --> 00:15:35,133 Wilder, but 938 00:15:36,083 --> 00:15:36,566 that's really 939 00:15:36,566 --> 00:15:36,750 what's 940 00:15:36,750 --> 00:15:37,283 going on here. 941 00:15:41,466 --> 00:15:42,383 Binance, the 942 00:15:42,383 --> 00:15:42,833 largest 943 00:15:42,833 --> 00:15:43,616 cryptocurrency 944 00:15:43,616 --> 00:15:44,250 exchange, 945 00:15:44,483 --> 00:15:45,116 got hacked. 946 00:15:45,533 --> 00:15:46,166 Really, 947 00:15:46,616 --> 00:15:47,516 really badly. 948 00:15:48,783 --> 00:15:50,583 Even though it's 949 00:15:50,583 --> 00:15:51,483 pretend money, 950 00:15:51,516 --> 00:15:52,416 people are acting 951 00:15:52,416 --> 00:15:53,466 like $6 billion 952 00:15:53,483 --> 00:15:54,033 real 953 00:15:54,033 --> 00:15:54,966 American dollars 954 00:15:54,966 --> 00:15:55,466 were taken. 955 00:15:56,283 --> 00:15:57,333 The FBI says the 956 00:15:57,333 --> 00:15:57,833 attack was 957 00:15:57,833 --> 00:15:58,616 perpetrated by 958 00:15:58,616 --> 00:15:59,399 North Dakota 959 00:15:59,399 --> 00:15:59,966 hackers. 960 00:16:00,183 --> 00:16:01,083 Sorry, North 961 00:16:01,083 --> 00:16:02,366 Korean hackers. 962 00:16:03,250 --> 00:16:03,716 Same thing. 963 00:16:05,066 --> 00:16:06,250 Over 90% of the 964 00:16:06,250 --> 00:16:07,299 stolen coins have 965 00:16:07,299 --> 00:16:07,983 been returned to 966 00:16:07,983 --> 00:16:08,133 their 967 00:16:08,133 --> 00:16:09,033 rightful owners. 968 00:16:09,549 --> 00:16:10,033 But it hasn't 969 00:16:10,033 --> 00:16:10,816 been recovered, 970 00:16:10,816 --> 00:16:11,533 Binance is just 971 00:16:11,533 --> 00:16:12,066 giving people 972 00:16:12,066 --> 00:16:12,583 money from 973 00:16:12,583 --> 00:16:13,533 company accounts. 974 00:16:14,166 --> 00:16:14,983 This goes to show 975 00:16:14,983 --> 00:16:15,283 that you 976 00:16:15,283 --> 00:16:15,816 shouldn't keep 977 00:16:15,816 --> 00:16:16,866 your crypto in an 978 00:16:16,866 --> 00:16:17,500 exchange. 979 00:16:17,799 --> 00:16:18,733 I print all mine 980 00:16:18,733 --> 00:16:19,866 out and keep them 981 00:16:19,899 --> 00:16:20,733 in the attic next 982 00:16:20,733 --> 00:16:21,616 to my fireworks 983 00:16:21,883 --> 00:16:22,000 and 984 00:16:22,000 --> 00:16:22,716 novelty lighter 985 00:16:22,716 --> 00:16:23,266 collection. 986 00:16:23,833 --> 00:16:24,466 Nothing bad will 987 00:16:24,466 --> 00:16:24,883 ever happen 988 00:16:24,883 --> 00:16:25,683 to my crypto. 989 00:16:28,816 --> 00:16:29,266 And that's all 990 00:16:29,266 --> 00:16:29,799 the news we 991 00:16:29,799 --> 00:16:30,816 logged this week. 992 00:16:30,966 --> 00:16:31,816 Remember to like, 993 00:16:31,816 --> 00:16:32,799 subscribe, and 994 00:16:32,799 --> 00:16:33,166 visit 995 00:16:33,183 --> 00:16:34,666 globaglob.dev for 996 00:16:34,666 --> 00:16:35,500 news technology 997 00:16:35,500 --> 00:16:36,166 professionals 998 00:16:36,166 --> 00:16:36,433 really 999 00:16:36,433 --> 00:16:36,966 care about. 1000 00:16:37,683 --> 00:16:38,283 And if you know 1001 00:16:38,283 --> 00:16:39,183 what email is, 1002 00:16:39,250 --> 00:16:39,816 sign up for the 1003 00:16:39,816 --> 00:16:40,816 weekly newsletter at 1004 00:16:40,833 --> 00:16:42,316 globaglob.dev slash 1005 00:16:42,333 --> 00:16:42,850 newsletter. 1006 00:16:43,633 --> 00:16:44,500 Until next time, 1007 00:16:44,683 --> 00:16:45,066 go twiddle 1008 00:16:45,066 --> 00:16:45,666 some bits.