AI is moving from experiment… to essential.

Every major industry is integrating it.
Every major company is investing in it.

By late 2025, AI was already an $800B market — growing at a pace that could push it well beyond $1 trillion in the years ahead.

Cloud infrastructure is scaling fast.
AI-enabled devices are multiplying.
Automation is becoming standard.

But here’s the real question…

When trillions flow into this transformation — which stocks stand to benefit most?

Our new report reveals 10 AI stocks positioned across the backbone of this shift — from the companies powering the infrastructure… to those embedding intelligence into everyday systems.

If you want exposure to one of the defining growth trends of this decade, start here.

Download the Report Now

For years, crypto holders have been told that the safest private key is one that never touches the internet. The latest hardware-wallet crisis exposes the dangerous gap in that promise: an air gap cannot protect a secret that was predictable from the moment it was created.

Beginning on July 30, attackers started draining bitcoin from wallets associated with affected COLDCARD devices—without needing to connect to or physically possess them. The failure was not a conventional remote break-in. It was a flaw in the process that generated certain wallet seeds, giving attackers a path to reconstruct candidate keys on their own computers and test them against public blockchain information.1

Every crypto wallet begins with entropy: unpredictable data used to create the seed from which its private keys are derived. Block Engineering’s technical analysis found that an integration error in affected COLDCARD firmware directed a seed-generation path to MicroPython’s deterministic Yasmarang software generator instead of the intended STM32 hardware random-number generator.1

On Mk2 and Mk3 devices using affected version 4 firmware, Block reported that the faulty path added no cryptographic entropy. On later Mk4, Q, and Mk5 models, secure-element entropy was included, but Block found that the reseeding process retained only four bytes before combining it with other device state. Coinkite separately estimated that affected seeds on those later models had about 72 bits of entropy rather than the expected 128 bits.1

The practical danger was straightforward. If attackers could sufficiently narrow the device identifier, timing state, and sequence of random-number calls, they could reproduce candidate seeds offline. A wallet address, extended public key, or other public value could then act as a check: derive a candidate, compare the result, and stop when it matches.1

That sentence captures the defining risk. Updating the device fixes future seed generation; it does not add randomness to a seed that already exists.

Because funds continued to move and different research teams used different clustering methods, the total loss figures changed rapidly in the days following the initial sweep.

Both firms suggested that multiple attackers might be involved. Galaxy Research shared attacker and victim address data with U.S. law enforcement, exchanges, and crypto-investigation groups, urging the industry to track opportunistic copycats.4

The variation in the address counts—7,300 by Galaxy and 5,200 by TRM—illustrates the difficulty of defining a “victim” solely by on-chain behavior. The safest summary is that at least roughly $116 million was confirmed stolen across multiple waves, with contemporaneous industry estimates warning the total could reach about $130 million.4

Exposure depends on the firmware running when the seed or other secret was generated, not merely the wallet model a user owns today.1

Coinkite says seeds created with at least 50 fair, independent, and private dice rolls are not considered at risk from this specific RNG issue alone. A strong, unique BIP-39 passphrase also adds an independent barrier, but it does not repair the underlying seed; the company still advises migration as soon as practical. A device PIN is not the same as a BIP-39 passphrase.2

Coinkite stated on August 8 that fixed firmware was available for every supported COLDCARD model. It also cited independent checks of specific remediation mechanisms, including real-device testing, source review, and a reproducible-build workflow. The company cautioned that these checks were not a complete audit and said a formal technical postmortem remained in progress.3

For an affected seed, the response is a controlled migration. The user should verify backups and wallet fingerprints, install and verify the correct fixed firmware for the device’s release track, generate a new seed, confirm a receive address on the hardware-wallet screen, and send a small test transaction before transferring the remaining balance. The old backup should be retained until the migration is fully confirmed.2

This is a moment for urgency without panic. A rushed seed rotation can introduce a different failure—an incorrect address, an incomplete backup, or loss of access—so the migration process should be deliberate and independently verified.

The incident does not prove that every hardware wallet is insecure. It demonstrates that self-custody changes the location of trust rather than eliminating it. Instead of trusting an exchange, the holder trusts a chain of hardware, firmware, build settings, entropy sources, backup procedures, and personal operational discipline.

The failure also shows why statistically random-looking output is not enough. A deterministic generator can produce values that appear random while still coming from a search space small enough for an attacker to explore. In hardware-wallet security, the critical question is not simply, “Does this device contain a secure component?” It is, “Can we verify that this exact firmware version actually used that component correctly?”1

Wallet makers are likely to combine independent sources of randomness so that no single component or line of code can determine the final seed. Current industry approaches already include blending randomness from separate hardware components or combining device-generated entropy with host-supplied entropy.6

Broad claims that a brand is “secure” will increasingly give way to assurance tied to a specific device, firmware release, and build. Reproducible builds, signed binaries, public source archives, and real-device instrumentation will become more important because they help reviewers prove that production firmware matches published code and invokes the intended hardware RNG.3

Expect stronger startup and seed-generation checks that refuse to proceed when the intended entropy source is unavailable. The industry’s goal will be to prevent a silent downgrade from trusted hardware randomness to an untrusted software fallback.1

Bug bounties, coordinated disclosure programs, third-party audits, and constructive engagement with outside researchers will become visible purchasing criteria. Transparency will no longer be judged only by whether source code is public, but by whether the vendor makes independent verification practical.3

For large treasuries and long-term holders, multisignature arrangements using independently designed devices and independently generated seeds will gain momentum. The important detail is diversity: a quorum made entirely from devices sharing the same vulnerable implementation can preserve the same single point of failure.1

Cold storage will become less “set and forget.” Watch-only alerts, periodic firmware reviews, documented wallet inventories, and rehearsed seed-rotation plans can shorten the interval between disclosure and safe migration. This trend is an operational inference from the speed of the theft: when an exploit becomes public and multiple attackers are competing for vulnerable funds, response time becomes part of the security model.4