1
00:00:00,000 --> 00:00:03,060
Welcome to the Bitcoin Brief for Wednesday, August 19th, 2026.

2
00:00:03,440 --> 00:00:04,940
I'm an AI-generated version of Marty Bent.

3
00:00:05,400 --> 00:00:08,540
The stories and opinions you're about to hear are drawn from today's Bitcoin Brief newsletter,

4
00:00:08,900 --> 00:00:11,140
written and curated by the real Marty and his AI assistant.

5
00:00:11,940 --> 00:00:14,760
Bitcoin is sitting around $64,979.

6
00:00:15,220 --> 00:00:18,000
That's roughly 1,539 sats per dollar,

7
00:00:18,480 --> 00:00:21,300
where a block height 963,171,

8
00:00:21,440 --> 00:00:24,020
with next block fees around 6 sats per vbyte.

9
00:00:24,740 --> 00:00:28,840
Today, the Treasury is at least doubling the maximum size of its long-end bond buybacks.

10
00:00:28,840 --> 00:00:34,720
They call it liquidity support. I call it implicit yield curve control, and the debt math tells you where this is going.

11
00:00:35,220 --> 00:00:40,780
We'll also get into the federal records concealment conspiracy at NIH, the massive power demands behind faster AI agents,

12
00:00:41,220 --> 00:00:47,360
$1.45 billion of Dodgers television debt sitting across five insurers, Japan's impossible interest rate trap,

13
00:00:47,900 --> 00:00:53,540
static Bitcoin core builds, and a new attempt to give apps self-custodial Bitcoin without forcing users to run a node.

14
00:00:54,060 --> 00:00:54,700
Let's get into it.

15
00:00:55,320 --> 00:00:57,480
Secretary Scott Besson's treasury blinked this morning.

16
00:00:57,940 --> 00:01:04,040
Beginning September 9th, Treasury will at least double the maximum size of its buybacks in the 10-20-year and 20-30-year buckets.

17
00:01:04,600 --> 00:01:08,220
The cap moves from $2 billion to at least $4 billion per operation.

18
00:01:08,720 --> 00:01:11,540
The current schedule contains seven operations through November 4th.

19
00:01:11,800 --> 00:01:16,600
If those dates hold, the combined cap rises from $14 billion to at least $28 billion.

20
00:01:17,020 --> 00:01:20,020
That does not mean Treasury has committed to spending the full $28 billion.

21
00:01:20,480 --> 00:01:24,380
It can accept fewer offers, pay less, or buy nothing during a particular operation.

22
00:01:24,800 --> 00:01:26,960
But do not lose the plot by staring at the fine print.

23
00:01:27,480 --> 00:01:31,260
The government is doubling the size of its scheduled bid in the long end of its own bond market

24
00:01:31,260 --> 00:01:33,480
after long-term borrowing costs ripped higher.

25
00:01:34,040 --> 00:01:36,040
Treasury says the program is meant to improve liquidity

26
00:01:36,040 --> 00:01:38,860
because dealers have been submitting a large number of high-quality offers.

27
00:01:39,580 --> 00:01:43,720
Sure, it is also true that a larger government bid supports the exact part of the yield curve

28
00:01:43,720 --> 00:01:45,280
that has been giving Washington fits.

29
00:01:45,740 --> 00:01:47,140
The market got the message immediately.

30
00:01:47,680 --> 00:01:49,240
In the 16 minutes surrounding the announcement,

31
00:01:49,700 --> 00:01:55,180
the 30-year Treasury yield fell from 5.266% to 5.190%.

32
00:01:55,180 --> 00:01:57,800
The 10-year yield dropped 4.5 basis points.

33
00:01:58,300 --> 00:02:00,340
Bitcoin rose 0.75%.

34
00:02:00,340 --> 00:02:02,900
Gold futures jumped 1.19%.

35
00:02:02,900 --> 00:02:05,180
Stock futures rose and the dollar fell.

36
00:02:05,460 --> 00:02:07,580
One headline did not necessarily cause every tick.

37
00:02:08,480 --> 00:02:10,400
The direction was still exactly what you would expect

38
00:02:10,400 --> 00:02:12,920
when the market sees the government's step-up support for duration.

39
00:02:14,200 --> 00:02:16,460
Now, before the bond nerds flood my inbox,

40
00:02:16,880 --> 00:02:19,400
I understand that this is not explicit yield curve control.

41
00:02:20,020 --> 00:02:21,740
Treasury has not announced a yield target.

42
00:02:22,280 --> 00:02:24,760
It has not promised unlimited purchases to defend that target.

43
00:02:25,620 --> 00:02:27,140
Treasury is not the Federal Reserve.

44
00:02:28,040 --> 00:02:30,000
These buybacks do not create money on their own.

45
00:02:30,940 --> 00:02:33,460
Treasury can fund them with cash or by issuing other debt,

46
00:02:33,780 --> 00:02:35,440
and the bonds it purchases are retired.

47
00:02:36,380 --> 00:02:37,600
Those distinctions are real.

48
00:02:38,120 --> 00:02:39,520
They do not change where this is headed.

49
00:02:40,660 --> 00:02:43,600
Yield curve control was never going to arrive with a flashing sign

50
00:02:43,600 --> 00:02:45,720
announcing that yield curve control begins today.

51
00:02:46,320 --> 00:02:49,020
It was always going to creep in through temporary programs,

52
00:02:49,380 --> 00:03:00,250
technical adjustments liquidity facilities and whatever linguistic tricks Treasury and the Fed could invent The pattern is always the same The government increases its bid Officials insist

53
00:03:00,250 --> 00:03:05,430
the move is small, technical, and temporary. If yields keep rising, the bid gets bigger.

54
00:03:06,270 --> 00:03:11,790
If Treasury cannot contain the pressure, the Fed gets dragged back in. Every step receives a new

55
00:03:11,790 --> 00:03:17,130
name so everyone can pretend the destination has changed. We have seen the explicit version before.

56
00:03:17,930 --> 00:03:21,710
In 1942, the Fed pegged Treasury bills at three-eighths of a percent

57
00:03:21,710 --> 00:03:24,930
and capped long-term Treasury yields at two and a half percent

58
00:03:24,930 --> 00:03:26,950
to keep the government's war financing cheap.

59
00:03:27,610 --> 00:03:29,830
The Fed bought what it needed to defend those rates.

60
00:03:30,410 --> 00:03:31,390
The money supply expanded.

61
00:03:31,950 --> 00:03:32,810
Inflation ripped.

62
00:03:33,650 --> 00:03:37,250
The arrangement finally broke with the Treasury-Fed Accord in 1951.

63
00:03:38,250 --> 00:03:41,910
Today's program is smaller, finite, and operated by Treasury.

64
00:03:42,530 --> 00:03:44,290
The political instinct is identical.

65
00:03:44,290 --> 00:03:48,770
When the cost of funding the government becomes uncomfortable, suppress the cost.

66
00:03:49,730 --> 00:03:51,050
The debt math explains why.

67
00:03:52,250 --> 00:03:56,350
Treasury's own numbers show nearly $40 trillion of total public debt outstanding,

68
00:03:57,110 --> 00:03:59,750
including $32.2 trillion held by the public.

69
00:04:00,970 --> 00:04:06,110
Gross interest expense reached roughly $1.17 trillion through July for the fiscal year.

70
00:04:06,870 --> 00:04:10,730
That gross treasury measure is different from the federal budget's net interest line,

71
00:04:11,150 --> 00:04:12,310
but the direction is obvious.

72
00:04:13,130 --> 00:04:16,910
Every round of refinancing at higher rates makes the bill larger.

73
00:04:18,330 --> 00:04:24,510
In theory, Washington can cut spending, raise taxes, grow its way out, restructure the debt, or default.

74
00:04:24,690 --> 00:04:31,110
In reality, the political class will do almost anything to avoid an honest default or the spending cuts required to stabilize the debt.

75
00:04:31,910 --> 00:04:33,170
That leaves debasement.

76
00:04:34,470 --> 00:04:35,930
First comes financial repression.

77
00:04:36,650 --> 00:04:37,710
Then come larger interventions.

78
00:04:38,270 --> 00:04:42,290
Eventually, the boundary between debt management and monetary policy disappears.

79
00:04:43,230 --> 00:04:45,010
Today's buybacks are not money printing.

80
00:04:45,710 --> 00:04:50,310
They are another step toward the point where money printing becomes the only politically acceptable answer.

81
00:04:51,390 --> 00:04:52,130
They will call it temporary.

82
00:04:52,830 --> 00:04:53,550
They will call it technical.

83
00:04:54,330 --> 00:04:55,770
They will call it liquidity support.

84
00:04:56,590 --> 00:05:00,870
They will insist it is not quantitative easing and definitely not yield curve control.

85
00:05:01,730 --> 00:05:02,550
Call it whatever you want.

86
00:05:03,150 --> 00:05:05,130
The great debasement trade is well on its way.

87
00:05:06,150 --> 00:05:07,390
This is why we Bitcoin.

88
00:05:07,390 --> 00:05:13,570
Bitcoin exists because governments and central banks cannot be trusted not to print money when they control the money.

89
00:05:14,250 --> 00:05:15,870
Bitcoin takes that control away.

90
00:05:16,810 --> 00:05:18,790
Its issuance schedule is fixed by consensus.

91
00:05:19,810 --> 00:05:25,210
No treasury secretary or central banker gets to change it because the government's debt service has become inconvenient.

92
00:05:26,250 --> 00:05:27,850
Bitcoin may not go straight up today.

93
00:05:28,390 --> 00:05:29,230
That is not the point.

94
00:05:29,830 --> 00:05:32,530
The reason for owning it just became more obvious.

95
00:05:33,190 --> 00:05:35,630
It is a very bad time to be short Bitcoin.

96
00:05:35,630 --> 00:05:38,130
Now, shifting to NIH accountability.

97
00:05:38,810 --> 00:05:42,830
Former senior NIAID official David Morins pleaded guilty to a federal conspiracy count.

98
00:05:44,030 --> 00:05:57,600
Morins admitted helping conceal or destroy records related to NIH coronavirus grants and COVID origin discussions in order to evade federal records requirements and Freedom of Information Act requests Jessica Rose also posted an image of a February 2020 email chain

99
00:05:58,620 --> 00:06:02,880
In that exchange, Jeremy Farrar expressed hope that a paper or letter would effectively put

100
00:06:02,880 --> 00:06:07,880
the origin question to bed. Peter Daszak described a separate statement intended to help shift

101
00:06:07,880 --> 00:06:13,060
public opinion. The screenshot is not a complete, independently authenticated production of the

102
00:06:13,060 --> 00:06:18,700
entire thread. Morin's plea does not prove that Anthony Fauci joined the conspiracy, and it does

103
00:06:18,700 --> 00:06:24,140
not settle the origin of the virus. It does confirm that federal records were intentionally concealed.

104
00:06:25,000 --> 00:06:29,040
That should matter to everyone, regardless of what they believe about the lab origin question.

105
00:06:30,080 --> 00:06:33,520
Public officials do not get to destroy the evidence and then lecture the public about

106
00:06:33,520 --> 00:06:38,760
trusting the process. Moving to artificial intelligence, Cerebris introduced its CS4 system,

107
00:06:38,760 --> 00:06:42,920
a rack containing three WSE3 turbo wafer-scale processors.

108
00:06:43,280 --> 00:06:46,540
The company claims twice the power per wafer, twice the input-output bandwidth,

109
00:06:46,940 --> 00:06:50,440
and up to 30 times the inference speed of GPU systems across its benchmark set.

110
00:06:51,320 --> 00:06:52,820
First shipments are scheduled for this quarter.

111
00:06:53,660 --> 00:06:54,260
Here's the catch.

112
00:06:55,160 --> 00:06:59,980
Semi-Analysis estimates that one rack will draw roughly 125 to 135 kilowatts.

113
00:07:00,560 --> 00:07:03,860
The same 5-nanometer wafer gets more performance by running faster

114
00:07:03,860 --> 00:07:05,860
and consuming significantly more power.

115
00:07:06,460 --> 00:07:07,920
The speed figures come from Cerebris.

116
00:07:07,920 --> 00:07:10,140
The power estimate comes from semi-analysis.

117
00:07:11,140 --> 00:07:12,840
Everyone wants faster AI agents.

118
00:07:13,320 --> 00:07:17,120
Very few people want to talk about the power conversion, liquid cooling, interconnects,

119
00:07:17,340 --> 00:07:18,620
and data centers required to run them.

120
00:07:19,160 --> 00:07:20,900
The agent race is becoming a power race.

121
00:07:21,640 --> 00:07:24,880
And that brings us back to the insurance and private credit machine we covered yesterday.

122
00:07:25,840 --> 00:07:30,400
Nick Nemeth reviewed regulatory filings and found approximately $1.45 billion of American

123
00:07:30,400 --> 00:07:34,360
Media Productions positions spread across five insurers on a mixed valuation basis.

124
00:07:35,360 --> 00:07:38,780
American Media Productions is tied to the Dodgers' Sportsnet LA television business.

125
00:07:39,640 --> 00:07:40,640
The accounting details matter.

126
00:07:41,220 --> 00:07:48,000
The same two Delaware Life security identifiers carried $277 million of aggregate par value at year-end 2024 and 2025,

127
00:07:48,460 --> 00:07:51,440
but moved from an affiliate schedule to unaffiliated corporate bonds.

128
00:07:52,340 --> 00:07:55,080
Security Benefit describes its position as a related party investment.

129
00:07:55,560 --> 00:07:58,980
Nick traces the remaining positions through current and former Dodgers investors,

130
00:07:59,380 --> 00:08:03,400
although some ownership and director chains still require confirmation from complete regulatory records.

131
00:08:04,360 --> 00:08:10,240
There is no proof here of default, insurer distress, self-dealing, improper classification, or policyholder losses.

132
00:08:10,960 --> 00:08:18,460
There is plenty of reason to inspect the ownership chains, accounting labels, and valuation basis before treating $1.45 billion like one clean number.

133
00:08:19,240 --> 00:08:20,120
That is the larger point.

134
00:08:20,680 --> 00:08:26,160
Complex relationships and friendly marks can make exposure look simple right up until somebody is forced to put a real price on it.

135
00:08:26,780 --> 00:08:32,320
Now to Japan, where J.P. Morgan has moved its forecast for the Bank of Japan's next rate hike from October to September.

136
00:08:32,320 --> 00:08:37,900
The bank added another projected hike in December and now expects a 1.5% policy rate by the end of

137
00:08:37,900 --> 00:08:53,611
2026 and 2 by the end of 2027 Those are JPMorgan forecasts not announced Bank of Japan policy They still expose the trap Higher rates may support the yen but they also increase debt service costs and put more pressure on a government bond market already carrying enormous duration risk

138
00:08:54,271 --> 00:08:59,331
A weaker yen invites tighter monetary policy. Tighter policy makes the fiscal math worse.

139
00:09:00,591 --> 00:09:04,211
Yesterday we looked at Japanese life insurers sitting on large, unrealized losses in government

140
00:09:04,211 --> 00:09:09,471
bonds. Today's question is one level higher. How far can the central bank raise rates before

141
00:09:09,471 --> 00:09:14,451
defending the currency begins destabilizing the government's own balance sheet. Quick Bitcoin Core

142
00:09:14,451 --> 00:09:19,031
note. Michael Ford released reproducible static test builds for Bitcoin D and Bitcoin Core's

143
00:09:19,031 --> 00:09:26,431
command line tools on x86-64 and ARM64 Linux. This sounds boring. It isn't. Current Linux releases

144
00:09:26,431 --> 00:09:31,311
rely on the host system's glibc and related libraries at runtime. Static builds place that

145
00:09:31,311 --> 00:09:36,291
code inside the executable. That expands what users can reproducibly verify and makes the software

146
00:09:36,291 --> 00:09:38,251
easier to run across older Linux distributions,

147
00:09:38,631 --> 00:09:40,251
Alpine Linux, and minimal containers.

148
00:09:40,971 --> 00:09:43,071
The test binaries remain position-independent,

149
00:09:43,371 --> 00:09:44,971
preserve address-based layout randomization,

150
00:09:45,451 --> 00:09:46,871
and are only about one megabyte larger.

151
00:09:47,451 --> 00:09:49,671
The main Bitcoin core pull request remains open.

152
00:09:50,231 --> 00:09:51,111
More platforms are planned,

153
00:09:51,371 --> 00:09:52,751
and the graphical interface is unchanged.

154
00:09:53,711 --> 00:09:55,611
Linux node operators can help by testing the binaries

155
00:09:55,611 --> 00:09:57,351
and reporting failures before static builds

156
00:09:57,351 --> 00:09:58,671
make their way into normal releases.

157
00:09:59,731 --> 00:10:01,471
Boring infrastructure work is often the work

158
00:10:01,471 --> 00:10:02,651
that makes Bitcoin more durable.

159
00:10:03,651 --> 00:10:05,031
Finally, in Freedom Tech Corner,

160
00:10:05,611 --> 00:10:07,331
Lightning Labs has released Wavelength and Alpha.

161
00:10:08,211 --> 00:10:11,911
Wavelength gives developers a unified wallet interface for Bitcoin, Lightning, and ARK.

162
00:10:12,251 --> 00:10:15,131
Applications can integrate it through Web, React Native, Native Mobile,

163
00:10:15,391 --> 00:10:16,631
Command Line, and API tools.

164
00:10:17,391 --> 00:10:19,291
The user keeps the keys on their own device.

165
00:10:19,911 --> 00:10:21,711
Lightning payments use atomic swaps through loop.

166
00:10:22,631 --> 00:10:25,071
ARK virtual transaction outputs provide the off-chain balance.

167
00:10:25,591 --> 00:10:27,251
If the normal service path disappears,

168
00:10:27,251 --> 00:10:30,091
the user retains a unilateral task back to the Bitcoin chain.

169
00:10:30,691 --> 00:10:32,471
For developers, the pitch is compelling.

170
00:10:33,111 --> 00:10:35,811
They can add self-custodial Bitcoin payments without operating a Lightning node,

171
00:10:36,151 --> 00:10:38,551
managing channels, or constantly sourcing inbound liquidity.

172
00:10:39,171 --> 00:10:42,991
That removes a pile of operational complexity that has kept Lightning out of many applications.

173
00:10:43,551 --> 00:10:46,131
But the complexity has not vanished. It moved.

174
00:10:46,431 --> 00:10:50,011
Normal operations still depends on an ARC operator, swap services, backups,

175
00:10:50,351 --> 00:10:52,971
liveness assumptions, and the application's handling of wallet state.

176
00:10:53,671 --> 00:10:57,111
A unilateral exit gives the user a way out, but every dependency still matters.

177
00:10:57,911 --> 00:11:00,831
Signet and testnet access are open. Mainnet remains invite only.

178
00:11:01,751 --> 00:11:02,451
I like the direction.

179
00:11:03,131 --> 00:11:04,631
Developers get a simpler integration surface,

180
00:11:05,151 --> 00:11:05,991
users keep their keys,

181
00:11:06,151 --> 00:11:07,251
and an on-chain escape hatch.

182
00:11:07,771 --> 00:11:09,331
The real test is whether applications

183
00:11:09,331 --> 00:11:11,031
can make backup and recovery simple enough

184
00:11:11,031 --> 00:11:11,751
for normal people

185
00:11:11,751 --> 00:11:13,831
without quietly turning the hosted service path

186
00:11:13,831 --> 00:11:14,871
into the real custodian.

187
00:11:15,831 --> 00:11:18,171
That's your Bitcoin Brief for Wednesday, August 19th.

188
00:11:18,591 --> 00:11:19,411
If you want all the links,

189
00:11:19,691 --> 00:11:21,071
primary documents, market data,

190
00:11:21,351 --> 00:11:22,231
and full written analysis,

191
00:11:22,691 --> 00:11:23,891
check today's Bitcoin Brief newsletter.

192
00:11:24,311 --> 00:11:25,211
It's already in your inbox,

193
00:11:25,511 --> 00:11:26,731
or you can find it on the website.

194
00:11:27,311 --> 00:11:28,291
If you're not subscribed yet,

195
00:11:28,291 --> 00:11:34,571
head to tftc.io slash bitcoin dash brief. It's free. Always will be. Forward this to a friend

196
00:11:34,571 --> 00:11:39,151
who needs more signal and less noise. Stay humble. Stack sats. I'll see you tomorrow.