1
00:00:04,619 --> 00:00:08,220
OpenAI launches agentic commerce protocol and chat GPT.

2
00:00:08,659 --> 00:00:11,920
Every purchase decision now happens inside the model window itself.

3
00:00:12,449 --> 00:00:14,859
Merchants like Etsy gain instant reach to hundreds

4
00:00:14,859 --> 00:00:17,000
of millions without building their own agents.

5
00:00:17,219 --> 00:00:20,260
The open protocol looks generous until you notice settlement

6
00:00:20,260 --> 00:00:22,360
still clears through a single set of keys.

7
00:00:22,620 --> 00:00:25,979
This forces Stripe to either lower its per transaction fees or

8
00:00:25,979 --> 00:00:29,000
watch volume migrate to any fork that undercuts the rate.

9
00:00:29,299 --> 00:00:33,180
Watch mid-market sellers test the first alternative endpoint by next summer.

10
00:00:33,790 --> 00:00:36,860
OpenAI models and codecs now on Oracle Cloud.

11
00:00:37,110 --> 00:00:39,790
Universal Credit's already sitting in Oracle accounts

12
00:00:39,790 --> 00:00:42,610
now cover Frontier Inference without fresh paperwork.

13
00:00:42,909 --> 00:00:45,849
Enterprises skip the usual cloud procurement dance.

14
00:00:46,029 --> 00:00:48,110
The move quietly shifts inference spend

15
00:00:48,110 --> 00:00:50,250
toward Oracle's existing compliance stack.

16
00:00:50,549 --> 00:00:50,869
Azure.

17
00:00:50,955 --> 00:00:54,994
Must match the credit flexibility or concede ground in regulated sectors.

18
00:00:55,174 --> 00:00:58,595
Smaller providers without similar credit programs now face higher

19
00:00:58,595 --> 00:01:01,544
effective costs when bidding against bundled OCI deals.

20
00:01:01,834 --> 00:01:04,555
Watch what happens to per token margins when those

21
00:01:04,555 --> 00:01:08,214
contracts add strict latency SLAs that force dedicated capacity.

22
00:01:08,809 --> 00:01:11,860
OpenAI acquires AI agent startup OnA.

23
00:01:12,239 --> 00:01:15,599
Long-running agent sessions now inherit orchestration tools built

24
00:01:15,599 --> 00:01:18,440
for days-long execution instead of single turns.

25
00:01:18,730 --> 00:01:21,580
The acquisition bakes persistence into the core platform.

26
00:01:22,040 --> 00:01:24,599
Watch retention metrics on current agent platforms

27
00:01:24,599 --> 00:01:27,120
once customers test the Integrated alternative.

28
00:01:27,440 --> 00:01:30,739
Rivals must either license similar capabilities or lose ground

29
00:01:30,739 --> 00:01:34,510
on enterprise pilots that require unininterrupted agent workflows.

30
00:01:34,760 --> 00:01:38,500
Anthropic faces pressure to accelerate its own orchestration roadmap

31
00:01:38,650 --> 00:01:42,860
or risk seating the multi-day use cases entirely within 18 months.

32
00:01:43,379 --> 00:01:47,449
Anthropic files S-1 for near $1 trillion IPO.

33
00:01:47,779 --> 00:01:50,900
Confidential filing at this scale locks the valuation before

34
00:01:50,900 --> 00:01:53,620
the next training cycle reveals true cost curves.

35
00:01:53,940 --> 00:01:57,500
Public markets get a clear look at the burn rate tied to cluster expansion.

36
00:01:57,849 --> 00:02:00,169
The real pressure lands on governance terms that

37
00:02:00,169 --> 00:02:03,040
let early backers retain influence after the listing.

38
00:02:03,319 --> 00:02:04,970
Talent shifts follow quickly.

39
00:02:05,209 --> 00:02:09,529
OpenAI must now decide whether to accelerate its own IPO timeline or

40
00:02:09,529 --> 00:02:12,960
watch key talent chase liquid equity at the new public entity.

41
00:02:13,320 --> 00:02:16,250
Senate Banking Committee holds AI innovation hearing.

42
00:02:16,449 --> 00:02:19,210
Export rules on advanced chips are already moving faster.

43
00:02:19,445 --> 00:02:21,365
than any bill the committee could pass.

44
00:02:21,645 --> 00:02:24,125
The hearing instead surfaces how quickly domestic

45
00:02:24,125 --> 00:02:26,985
foundries must scale to replace restricted shipments.

46
00:02:27,205 --> 00:02:31,785
Nvidia now faces a hard choice between licensing older architectures to allies

47
00:02:31,904 --> 00:02:34,985
or absorbing the full cost of Arizona production ramps.

48
00:02:35,205 --> 00:02:38,205
Either path compresses gross margins on inference hardware

49
00:02:38,205 --> 00:02:40,725
by at least 8 points inside 18 months.

50
00:02:41,389 --> 00:02:43,820
China tests AI robot cleaners in homes.

51
00:02:44,190 --> 00:02:48,509
Beijing and Shenzhen homes are feeding petabyte scale video of daily routines

52
00:02:48,509 --> 00:02:50,470
into training loops that Western labs

53
00:02:50,470 --> 00:02:52,869
cannot replicate under current privacy rules.

54
00:02:53,229 --> 00:02:56,149
The resulting models will handle edge cases like laundry

55
00:02:56,149 --> 00:03:00,029
folding long before any US competitor clears regulatory review.

56
00:03:00,440 --> 00:03:02,520
That data advantage forces figure and

57
00:03:02,520 --> 00:03:05,380
agility to either acquire Chinese vision startups

58
00:03:05,520 --> 00:03:08,380
or accept a permanent lag in consumer deployments.

59
00:03:08,589 --> 00:03:11,399
Expect US firms to strike data partnerships or

60
00:03:11,399 --> 00:03:14,589
fall behind on household tasks by mid-2026.

61
00:03:15,039 --> 00:03:18,750
Supabase raises $500 million for AI developer tools.

62
00:03:19,029 --> 00:03:22,440
Valuation headlines missed the clause that lets GIC influence

63
00:03:22,440 --> 00:03:25,350
which post-gress extensions reach the open core first.

64
00:03:25,669 --> 00:03:26,660
With the fresh capital,

65
00:03:26,800 --> 00:03:28,199
Supabase can now ship production.

66
00:03:28,401 --> 00:03:32,231
Ready PG vector improvements 6 months ahead of any community fork.

67
00:03:32,561 --> 00:03:35,682
That timing locks developers building RAG apps into their

68
00:03:35,682 --> 00:03:39,391
hosted indexes before competitors finish their own managed offerings.

69
00:03:39,722 --> 00:03:41,921
The gap will show up first in query latency

70
00:03:41,921 --> 00:03:44,951
benchmarks that teams actually cite in production reviews.

71
00:03:45,281 --> 00:03:47,682
Firebase will have to match the extension velocity

72
00:03:47,682 --> 00:03:50,701
or watch mid-market migrations accelerate through next year.

73
00:03:51,151 --> 00:03:54,792
Suno secures $400 million Series D for AI Music.

74
00:03:55,121 --> 00:03:56,121
Copyright exposure.

75
00:03:56,203 --> 00:03:59,164
Still blocks the biggest sync deals with brands and studios.

76
00:03:59,324 --> 00:04:01,214
The new round buys time to settle with

77
00:04:01,214 --> 00:04:04,074
major labels before those contracts move to competitors.

78
00:04:04,333 --> 00:04:07,083
The settlement terms will determine whether Suno can embed

79
00:04:07,083 --> 00:04:10,233
full catalog tracks or stays limited to synthetic stems.

80
00:04:10,524 --> 00:04:12,623
Universal and Sony now face a clock.

81
00:04:12,884 --> 00:04:15,643
Either license their catalogs to Suno or watch

82
00:04:15,643 --> 00:04:19,204
independent artists route around traditional royalty flows entirely.

83
00:04:19,444 --> 00:04:23,824
Expect one major label to announce a licensing partnership inside 12 months.

84
00:04:24,246 --> 00:04:28,075
Generalist AI raises $400 million for robotics agents.

85
00:04:28,305 --> 00:04:30,686
Robots that learn from video rather than code

86
00:04:30,686 --> 00:04:33,276
will flip the economics of last mile logistics.

87
00:04:33,605 --> 00:04:36,165
One model handling both navigation and delicate

88
00:04:36,165 --> 00:04:38,765
manipulation cuts integration costs by 50%.

89
00:04:39,126 --> 00:04:43,585
This forces Fi to either match the data scale or retreat to research contracts.

90
00:04:43,786 --> 00:04:47,765
Expect pilot programs in two major retailers by Q2 next year,

91
00:04:47,925 --> 00:04:51,925
with full rollout decisions riding on whether error rates drop below 2%.

92
00:04:52,290 --> 00:04:55,329
The valuation premium assumes that flywheel spins

93
00:04:55,329 --> 00:04:57,989
without regulatory friction on job displacement.

94
00:04:58,309 --> 00:05:01,869
Flourish lands 500 a million for brain inspired AI.

95
00:05:02,170 --> 00:05:05,170
Models built on cortical columns promise lower energy

96
00:05:05,170 --> 00:05:07,640
per token than attention layers ever delivered.

97
00:05:07,929 --> 00:05:10,410
The catch surfaces when mapping those columns onto

98
00:05:10,410 --> 00:05:13,209
GPUs that were never designed for sparse spiking.

99
00:05:13,640 --> 00:05:17,109
This pushes OpenAI to either license the architecture or double

100
00:05:17,109 --> 00:05:20,100
down on custom silicon within the next product cycle.

101
00:05:20,429 --> 00:05:24,190
Early benchmarks suggest a 3x reduction in training flops,

102
00:05:24,309 --> 00:05:27,890
but only if the biological priors survive contact with noise.

103
00:05:28,109 --> 00:05:32,109
Retail investors chasing the Bezos name are missing the hardware dependency.

104
00:05:32,779 --> 00:05:35,739
What OU4's retail investors chasing the Bezos

105
00:05:35,739 --> 00:05:37,839
name are missing the hardware dependency.

106
00:05:38,279 --> 00:05:42,040
Super Micro announces $7 billion equity raise for AI.

107
00:05:42,339 --> 00:05:44,589
AI server margins have already peaked,

108
00:05:44,769 --> 00:05:47,059
and this volume only hastens the slide.

109
00:05:47,329 --> 00:05:48,500
Supply gluts from the cap.

110
00:05:48,845 --> 00:05:52,625
will compress margins for every GPU reseller starting Q3.

111
00:05:52,845 --> 00:05:54,994
Enterprise buyers gain leverage on custom

112
00:05:54,994 --> 00:05:57,915
configurations because inventory finally exists.

113
00:05:58,204 --> 00:06:01,394
This forces Dell to accelerate its own financing options

114
00:06:01,394 --> 00:06:04,565
or watch mid-market deals migrate to direct Asian sourcing.

115
00:06:04,869 --> 00:06:07,489
The 20% drop prices in the dilution risk,

116
00:06:07,549 --> 00:06:10,339
but ignores how faster server availability changes

117
00:06:10,339 --> 00:06:13,260
deployment timelines for the largest cloud operators.

118
00:06:13,589 --> 00:06:17,429
NSPM11 directs AI acceleration for national security.

119
00:06:17,790 --> 00:06:20,339
Security clearances create the actual moat here,

120
00:06:20,549 --> 00:06:21,850
not the technology itself.

121
00:06:22,309 --> 00:06:25,549
Clearance bottlenecks will determine winners long before any model.

122
00:06:25,695 --> 00:06:26,494
Performance does.

123
00:06:26,815 --> 00:06:29,494
This forces Google to spin up a classified division

124
00:06:29,494 --> 00:06:32,924
or lose DOD workloads to Palantir within 18 months.

125
00:06:33,214 --> 00:06:35,855
Smaller research groups without facility approvals get

126
00:06:35,855 --> 00:06:38,154
locked out of the largest budgets entirely.

127
00:06:38,415 --> 00:06:42,535
The memo's real effect surfaces in how quickly existing vendors must add air

128
00:06:42,535 --> 00:06:46,445
gapped inference options to stay eligible for the next round of task orders.